Leveraging Disruptive Technologies for Faster and More Efficient Disaster Response Management

Abstract

Natural disasters cause extensive infrastructure and significant economic losses, hindering sustainable development and impeding social and economic progress. More importantly, they jeopardize community well-being by causing injuries, damaging human health, and resulting in loss of life. Furthermore, communities often experience delayed disaster response. Aggravating the situation, the frequency and impact of disasters have been continuously increasing. Therefore, fast and effective disaster response management is paramount. To achieve this, disaster managers must proactively safeguard communities by developing quick and effective disaster management strategies. Disruptive technologies such as artificial intelligence (AI), machine learning (ML), and robotics and their applications in geospatial analysis, social media, and smartphone applications can significantly contribute to expediting disaster response, improving efficiency, and enhancing safety. However, despite their significant potential, limited research has examined how these technologies can be utilized for disaster response in low-income communities. The goal of this research is to explore which technologies can be effectively leveraged to improve disaster response, with a focus on low-income communities. To this end, this research conducted a comprehensive review of existing literature on disruptive technologies, using Covidence to simplify the systematic review process and NVivo 14 to synthesize findings.

1. Introduction

Natural disasters arise from the interaction of natural hazards with the exposure and vulnerabilities of communities that are unable to withstand and cope with such threats [1,2]. Such disasters include (1) extreme geological events, including earthquakes, and (2) climate- and weather-related events, including hurricanes or cyclones, tornadoes, and floods [2,3]. These disasters often occur and have devastating power [2,4].

Between 1960 and 2019, 11,360 disaster events where either more than ten people lost their lives or more than 100 individuals were affected occurred globally [3,5]. Furthermore, over the past two decades, natural disasters have resulted in economic losses exceeding USD 2.96 trillion, claimed 1.23 million lives, and impacted over 4.2 billion individuals [4,6,7]. Disasters not only cause substantial damage to property and critical infrastructure but also pose significant risks to human lives and well-being, leading to injuries, adverse health effects, income loss, displacement, and restricted access to essential resources such as food, electricity, and water [4,6,8,9,10]. That said, these events not only stand as the main source of destruction to property and infrastructure systems, particularly in underdeveloped communities, but also act as a significant obstacle to sustainable development, hindering social and economic progress [5,6].

The frequency and intensity of natural disasters, coupled with the resulting damages and losses, have shown a persistent increase [11,12,13,14,15,16,17]. Furthermore, communities often receive delayed disaster response and recovery, especially low-income communities that, due to lacking resources for prevention, preparation, and adequate response, are more exposed and vulnerable to such threats [2,3,18,19,20,21]. This lack of a prompt and adequate response further exacerbates the already elevated risks associated with these destructive events [14,22]. Consequently, fast and effective disaster management is of the utmost importance.

Effective disaster management is crucial for protecting vulnerable communities and critical infrastructure while minimizing the overall adverse impacts of disasters [23,24]. Disaster management can be divided into four stages: (1) mitigation, which involves managerial actions aimed at preventing or reducing the impact of future disasters, yielding long-term benefits; (2) preparedness, occurring before the disaster, involves preparatory measures to safeguard lives, enhance response and rescue operations, as well as improve early warning systems and monitoring capabilities; (3) response, taking place during and after the disaster, entails search and rescue activities, initial damage assessments, first-aid provision, humanitarian assistance, and shelter provision; and (4) recovery, occurring after the disaster, involves debris removal, damage assessment, reconstruction, financial assistance, and community development [16,23,24]. Disaster managers must take on a growing responsibility to actively protect communities by developing swift and effective disaster management strategies, ensuring resilience, and minimizing the overall adverse impacts of disasters [24]. This proactive approach is crucial in safeguarding lives and critical infrastructure.

The impacts of these events present significant challenges for disaster response managers, who grapple with increasingly limited resources and a fatigued workforce [24]. Disruptive technologies, including artificial intelligence (AI), machine learning (ML), the Internet of Things (IoT), robotics, and information and communication technology (ICT), and their applications in geospatial analysis, smartphone applications, and social media, can significantly enhance response efforts by expediting operations, improving efficiency, and ensuring safety. In addition to improving and expediting disaster response, these technologies support sustainable development by fostering community recovery and strengthening infrastructure resilience. Thus, they play a pivotal role in disaster response. However, despite their significant potential, there is limited research analyzing how disruptive technologies can be effectively utilized in disaster response management, particularly in low-income communities that are highly exposed and vulnerable to such threats, often receiving delayed disaster response and recovery. This study addresses this gap by synthesizing existing knowledge to comprehensively explore how disruptive technologies can be utilized in disaster response, including their strengths and limitations. It contributes to both theoretical understanding and practical applications by offering insights that guide future research and inform the development of strategies tailored to the unique challenges of low-income communities. To this end, the goals of this research are to investigate how disruptive technologies can be effectively utilized to improve the efficiency and speed of disaster response management and to identify which technologies are most effective and feasible for enhancing resilience and accelerating response in low-income communities, considering the adoption barriers and limited resources of these communities. To achieve these goals, a comprehensive review of the existing literature on disruptive technologies and their applications in disaster response was conducted using (1) Covidence to streamline the systematic review process and (2) NVivo 14 to synthesize the findings.

2. Background

Disruptive technologies are innovations that alter or disrupt established practices, initially attracting a small number of users but gradually expanding and displacing previously dominant technologies [25,26]. These technologies offer opportunities for continued innovations, enhanced productivity, cost reduction, analysis and synthesis of vast amounts of data, improved decision-making, and increased efficiency [25,27,28]. Examples of these technologies encompass AI, ML, and robotics, as well as their applications in geospatial analysis, smartphone applications, and social media.

AI refers to the simulation of human intelligence processes by computer systems, including learning, reasoning, problem-solving, perception, and decision-making [14,16,29,30]. This disruptive technology plays a crucial role in disaster response management, aiding in hazard prediction, decision-making, damage assessment, and resource allocation [16,23]. AI applications such as remote sensing (RS), real-time data analysis, and optimization algorithms enhance situational awareness, expedite response efforts, improve response prioritizations, and support equitable resource distribution [31,32]. Key components of AI include machine learning (ML), data mining, deep learning (DL), large language models, natural language processing (NLP), neural networks, machine–human interaction, machine vision, the Internet of Things (IoT), and robotics [29,30,33].

ML enables systems to learn and improve through experience without explicit programming. This is achieved by training computer systems and developing algorithms that enable them to recognize patterns in data and make decisions or predictions based on that data [30,34]. ML can be categorized into three types: (1) supervised learning, where the algorithm learns from training datasets labeled by the user as correct or incorrect to make predictions or decisions; (2) unsupervised learning, where the algorithm employs statistical methods to identify patterns and relationships in data without the need for labeled output; and (3) DL, which utilizes artificial neural networks (ANNs) with multiple layers, inspired by the structure and function of the human brain, to learn complex representations of data [24,29,30,35]. Furthermore, DL models utilize convolutional neural networks (CNNs) for multi and recurrent neural networks (RNNs) to learn complex representations of data, such as image captioning, language modeling, and speech recognition [36]. Within ML, NLP enables machines to comprehend and interact with human language, while computer vision grants machines the ability to interpret visual data captured by cameras [30]. Large language models have made notable advancements in the field of NLP. These models undergo training on vast volumes of text data, enabling them to generate text that closely resembles human writing, provide accurate answers to questions, and perform other language-related tasks with high accuracy [33].

Data mining refers to the process of collecting extensive datasets from various systems and using the insights gained from this data to make predictions [37]. Machine–human interaction refers to how people and automated systems interact with each other [35].

Robotics and unmanned aerial vehicles (UAVs) refer to the interdisciplinary field of engineering and science that involves the design, construction, operation, and use of robots. These robots are autonomous or semi-autonomous machines that can perform intended tasks through programmed instructions or by remote control [38].

The Internet of Things (IoT) is a network of physical objects embedded with electronics, software, sensors, and connectivity, enabling them to exchange data with other connected devices [39]. This connectivity allows for RS and control, facilitating direct integration between the physical world and computer-based systems [39]. IoT enhances efficiency, accuracy, and economic benefits by enabling the acquisition and measurement of a wide variety of signals during disasters, which can be used for meaningful interpretation of events [39].

Geospatial data refer to information that identifies the geographic location and characteristics of natural or constructed features on Earth’s surface, typically represented in the form of maps, images, or datasets [17,30]. Geographic information systems (GISs) can capture, store, manipulate, analyze, manage, and present spatial or geographic data, allowing visualization, interpretation, and understanding of patterns and relationships in data through maps and spatial analysis [17,23,34]. Furthermore, GIS can utilize AI techniques and algorithms to enhance spatial analysis and decision-making processes.

The rapid collection, management, and processing of large datasets are crucial for enabling effective and efficient disaster response management [24,39,40,41,42,43]. Therefore, leveraging disruptive technologies can greatly contribute to effective, faster, and safer disaster response management due to their ease of use, high-speed operation, and acceptable accuracy [16].

3. Materials and Methods

This literature review is intended to provide a comprehensive understanding of how disruptive technologies can be utilized in disaster response. The research is guided by two main research questions: (1) How can disruptive technologies (e.g., AI, ML, robotics, and their applications in geospatial analysis, smartphone applications, and social media) be effectively used in disaster response to improve the efficiency, effectiveness, and speed of disaster management? And (2) which disruptive technologies are the most effective and feasible for enhancing resilience and expediting recovery in low-income communities, considering the adoption barriers and limited resources of these communities?

3.1. Literature Retrieval and Selection

This study addressed these two questions by conducting a scoping literature review to investigate and synthesize the current literature on the use of disruptive technologies in disaster response management. The scoping review methodology was selected to ensure a broad exploration of available literature while identifying key themes and gaps relevant to the research questions. This review included conference proceedings, journal articles, and review articles sourced from the Scopus database.

To identify relevant publications, a preliminary search was conducted to identify key terms within titles, abstracts, keywords, and indexed keywords. This initial search included the terms “artificial intelligence” AND “natural disaster.” Subsequently, a more extensive search, limited to English studies, was conducted using all identified keywords and index terms to collect all relevant publications. Table 1 presents the specific keyword combinations used in the search. Each line of the table contains multiple keyword combinations aimed at gathering all relevant publications. Quotation marks delineated phrases; “AND” required multiple terms to appear together, while “OR” allowed for either term to be included. The asterisk symbol (*) acted as a wildcard, enabling variations of a word, such as robot or robotics.

Table 1. Keyword Combinations for Scoping Literature Review.

	Keyword Combinations
1	“disaster management” OR “emergency management” AND “artificial intelligence” OR robot* OR “machine learning” AND “natural disaster” OR hurricane OR earthquake OR cyclone OR tornado OR flood*
2	rescue OR evacuation OR recovery OR response AND “artificial intelligence*” OR robot* OR “machine learning” AND management AND “natural disaster” OR hurricane OR earthquake OR cyclone OR tornado OR flood*
3	“disruptive technolog*” AND management AND “natural disaster” OR hurricane OR earthquake OR cyclone OR tornado OR flood*
4	“post-disaster management” AND “disruptive technolog*” OR “artificial intelligence” OR “machine learning” OR robot*

* The asterisk symbol (*) acts as a wildcard, allowing variations of a word, such as robot or robotics and technology or technologies.

A total of 1314 papers were retrieved and imported into Covidence, an online platform tailored for simplifying the systematic review process [44]. Covidence automatically identified and removed duplicates. The authors screened the titles and abstracts to assess the relevance of the studies and to identify potentially eligible studies for inclusion. Studies were excluded at this stage if they were deemed irrelevant based on title and abstract screening, such as focusing on unrelated disasters or lacking relevance to disaster response management. Subsequently, full-text studies of potentially eligible publications were evaluated for eligibility, with the focus being on the application of disruptive technologies in disaster response management for earthquakes, tornadoes, hurricanes, and floods. Studies were excluded during the full-text review phase based on the following criteria: duplicates, non-English publications, no full-text availability, and wrong focus or out-of-scope studies (e.g., focusing on other types of disasters or other phases of disaster management).

3.2. Data Extraction and Analysis

After completing the screening phase in Covidence, all relevant studies were extracted and imported into NVivo 14 for organization and thematic analysis. NVivo is a specialized qualitative analysis software designed for literature reviews, providing unique functionalities aimed at enhancing transparency and confidence in synthesizing findings [45]. Customized attributes were created to categorize essential information for synthesis, such as country of origin, year, and article type (i.e., conference proceedings, article, or review).

Codes were created to categorize themes, subthemes, and all crucial information for synthesis purposes. The themes created by the authors included Artificial Intelligence, Big Data, Disaster Management, Geospatial Analysis, Robotics, Smartphone Applications, and Social Media. Subthemes included a more detailed breakdown of the aforementioned themes, such as Cloud Computing, Machine Learning, and the Internet of Things under the Artificial Intelligence theme. Thematic coding allowed the authors to systematically extract and categorize key information from each study, including the study’s focus as well as the strengths and limitations of disruptive technologies.

The combined approach of utilizing Covidence for the systematic screening and data extraction, alongside NVivo 14 for in-depth thematic analysis, allows for a comprehensive and transparent synthesis of the literature.

4. Results

A total of 1314 publications were identified from Scopus. Covidence automatically identified and removed two duplicates. After screening titles and abstracts, 744 were considered irrelevant. Then, 569 publications underwent full-text review for eligibility. After a final review in Covidence, 343 studies were excluded, leaving 225 remaining. The main reasons for exclusion were incorrect focus or being out of scope (e.g., focusing on other types of disasters or other phases of disaster management), lack of full text, or not being in English. The 225 included studies were imported into NVivo 14 for the scoping review. Figure 1 outlines the flow diagram of the literature search and study selection process.

Figure 1. Flow diagram of the literature search and study selection.

Figure 2 and Figure 3 present the general characteristics of the 225 studies included in this review. The literature: (1) spans from 1996 to 2024, as shown in Figure 2; (2) includes 110 journal articles, 104 conference proceedings, and 11 review articles, as shown in Figure 2; and (3) encompasses studies from 58 countries, as illustrated in Figure 3, with 65 from the United States, 50 from India, 21 from China, 14 from Japan, 11 from Italy, 10 from the United Kingdom, among others.

Figure 2. Distribution of included studies by year of publication and article type.

Figure 3. Distribution of included studies by country of origin.

Improving disaster resilience and effective disaster response management stand out as critical global imperatives [23,46,47,48,49,50,51]. Efficient strategies for disaster management can mitigate the impact of disasters on people, infrastructure, and the environment, as well as reduce damage and, most importantly, casualties [47,50,52,53,54,55]. Effective disaster management systems rely on accurate data, reliable and streamlined communication networks, and collaboration among various stakeholders [23,49,53,56,57,58]. They also emphasize community engagement to integrate local insights and requirements into planning and response efforts [53]. By implementing robust disaster response systems, societies can enhance preparedness, reduce disaster impact, save lives, and mitigate social, economic, and environmental consequences [31,53].

Disruptive technologies offer significant potential to expedite processes, improve efficiency, and ensure safe disaster response management [23,31,50,51,52,53,54,55,59,60,61,62]. These technologies not only facilitate and expedite disaster response but are also essential for managing and distributing available resources efficiently and equitably, removing human biases [31,32,52,63]. This study provides a comprehensive examination of disruptive technologies, exploring their potential for effective disaster response, particularly in low-income communities that are highly exposed and vulnerable to disasters and often experience delayed response.

4.1. Artificial Intelligence (AI) in Disaster Management

Artificial intelligence (AI) plays a crucial role in disaster management, aiding in prediction, timely decision-making, and effective response across all disaster phases [16,23,31,49,51,52,55,57,59,61,64]. It effectively manages vast and diverse data types, enhancing the understanding of disasters [16,55]. Key AI applications include hazard assessment, data collection, prediction, and infrastructure damage assessment [49,51,55,57]. Computational intelligence supports disaster control, while computer vision utilizes remote sensing (RS) data for effective mitigation, resource allocation, traffic management, and response prioritization [59]. Optimization algorithms, such as Particle Swarm Optimization (PSO), offer advantages over other optimization techniques, including ease of implementation, robustness, scalability, and simplicity in mathematical calculations [65]. Equitable resource distribution models, like game theory, further enhance efficient and socially equitable emergency responses [66]. Table 2 summarizes findings on AI-based technologies in disaster management. The table highlights each study’s focus, strengths, and limitations.

Table 2. Summary of AI studies for disaster management.

Study	Technology	Study Focus	Strengths	Limitations
[67]	AI, fuzzy programming	Develops a fuzzy multi-objective disaster relief model for nonprofits, addressing prioritizations, effectiveness, efficiency, and equity in logistics.	Multi-objective relief model improves delivery efficiency.	Complex optimization with fuzzy parameters.
[66]	Game-based DSS, game theory	Utilizes game theory for fair and efficient resource allocation during multiple emergency events, optimizing limited resources across multiple locations.	Fair and equitable resource allocation, proven effective in case studies.	High demands reduce satisfaction and complexity increases with more crisis locations.

Major components of AI encompass machine learning (ML) and its applications (e.g., DL, NLP, neural networks, large language models), data mining, machine–human interaction, machine vision, the Internet of Things (IoT), robotics, and UAVs, as well as their applications in geospatial analysis, smartphone applications, and social media [29,30,33]. All these components will be presented in the following sections.

4.2. Machine Learning (ML) in Disaster Management

Machine learning (ML), including specialized techniques such as deep learning (DL), natural language processing (NLP), convolutional neural networks (CNNs), artificial neural networks (ANNs), and recurrent neural networks (RNNs), plays a key role in analyzing extensive datasets to forecast disasters, assess impacts, and identify survivors [23,47,52,53,55,57,61,62,68,69,70,71,72]. These techniques facilitate time series analysis, accurately predicting disaster events, improving forecasts, mitigating disaster threats, and reducing false alarms and noise [23,55,61]. Therefore, ML is essential in early warning systems for various natural disasters, including earthquakes, flooding, and severe weather events. It supports disaster monitoring, mapping, damage assessment, rescue operations, crowd evacuation, and informed decision-making [23,55,61,62,71,73].

ML applications in image processing, DL, and NLP contribute significantly to disaster management efficiency [55,72]. Image recognition and classification assess damages by analyzing images, while predictive analytics examines historical events to detect patterns and vulnerable populations. Sentiment analysis on social media data provides early warnings and real-time reports [59]. These technologies enable the analysis of unstructured data from diverse sources, such as social media and news articles, facilitating the identification of disaster-affected areas, monitoring misinformation spread, and enabling communication among response and recovery groups. Thus, facilitating intelligent and efficient decision-making [55,59,61,69,71]. Table 3 presents ML technologies in disaster management, underscoring each study’s focus, strengths, and limitations.

Table 3. Summary of ML studies for disaster management.

Study	Technology	Study Focus	Strengths	Limitations
[74]	CNN, DL, RS	Uses DL to assess building damage and affected population after an earthquake to provide quick information for rescue operations.	High accuracy in identifying building damage and affected population.	High computational requirements.
[63]	bilevel optimization, ML	Proposes a hybrid heuristic model combining ML and bilevel optimization method for improving to improve traffic control for efficient evacuation.	Efficient optimization of large traffic networks through road direction changes.	Complex implementation for large-scale networks, requiring high computational resources.
[75]	AI, CNN, DL, SNN	Proposes a Siamese neural network (SNN) with light detection and ranging (LIDAR) data to assess post-earthquake building damage.	High classification accuracy and effective feature extraction.	Limited to LIDAR data.
[76]	ML, RL	Proposes a hierarchical multi-agent RL framework for real-time resource allocation in disasters.	Improves efficiency and flexibility in resource allocation.	High computational resources for multi-agent systems
[77]	ML	Uses supervised ML technique using Gaussian process regression for post-seismic damage assessment, with minimal observation data.	Accurate damage prediction with limited data.	Accuracy depends on quality of available data.
[78]	Aerial imaging, AI, CNN	Assess the generalizability of CNN models for post-hurricane damage assessment across different disaster events.	Good classification accuracy in detecting high-damage areas.	Limited generalizability due to unique disaster characteristics and reduced accuracy in intermediate damage levels.
[79]	CNN, DL, smart infrastructure	Explores the use of CNNs to classify images from disaster sites, providing support for emergency decision-making.	High accuracy in image classification with lower computational requirements	Requires large amounts of high-quality image data for accuracy.
[60]	AI, CNN, computer vision, transfer learning (TL)	Develops an ML model using DenseNet201 to classify disaster damage in structures to prioritize rescue efforts.	High classification accuracy (90%) in distinguishing levels of structural damage.	Requires high-quality images and extensive labeled data.
[80]	DNN, ML	Employs a deep neural network (DNN) and factorization machine for assessing post-earthquake building damage, tested on Nepal’s earthquake.	High classification accuracy for damage grades.	Limited by model complexity and the need for large-scale labeled datasets.
[81]	DL, ML, TL	Proposes a DL framework for post-earthquake damage recognition.	Provides a quick structural assessment.	Lacks comprehensive validation, limiting generalizability.
[82]	AI, CNN, DL	Introduces a multi-view (i.e., aerial and ground views) CNN to improve damage classification.	High accuracy due to multi-view imagery.	Requires extensive image datasets from multiple angles.
[83]	CNN, DL	Evaluates a multi-hazard damage detection dataset using CNNs and DL for building damage classification.	Improves damage classification across multiple disasters and damage classes.	Limited performance for some disaster types due to class imbalance and lack of diverse training data.
[84]	TL	Proposes a physics-informed model for transferring knowledge between buildings to diagnose seismic damage, using adversarial domain adaptation to overcome data scarcity.	Accurate seismic damage diagnosis using limited labeled data.	Performance may vary based on structural differences.
[85]	AI-based image recognition, DL, GAN	Develops a novel AI-based model for urban flood detection using generative adversarial network (GAN) with high precision using DL techniques.	High accuracy in flood detection with 84% precision and 91% recall rate.	Limited to urban flooding scenarios and varying CCTV angles and heights complicate accurate calculations.
[86]	Deep RL	Proposes a deep reinforcement learning (RL) approach to rescue path planning in complex urban flood scenarios, considering obstacles and high-risk areas.	Navigates complex urban flood conditions effectively.	Requires comprehensive training for RL models and real-time updates.

4.3. Internet of Things (IoT) in Disaster Management

The Internet of Things (IoT) plays a paramount role in improving disaster response by utilizing sensor technology for real-time data collection, enabling informed decisions, and addressing community needs [23,31,39,61]. Sensors are key in search and rescue operations, with acoustic sensors and microphone arrays detecting and locating survivors based on sound or voice [23,50]. Vision systems use various types of cameras, such as thermal, color, and infrared (IR), to detect victims, while computer vision algorithms facilitate pattern recognition, tracking, and warnings in disaster scenarios [50].

IoT frameworks facilitate various functions, including data collection, analytics, early warning systems, hazard identification, remote event monitoring, and victim location [50]. By integrating IoT devices with complementary data sources, including AI, ML, big data analytics, satellite images, and drone videos, IoT enhances decision-making and response initiatives [31]. The integration of robotic systems with IoT technology, known as the Internet of Robotic Things (IoRT), is particularly effective in surveillance and disaster management scenarios [50,87]. Table 4 provides a summary of IoT technologies in disaster management, presenting each study’s focus, strengths, and limitations.

Table 4. Summary of IoT studies for disaster management.

Study	Technology	Study Focus	Strengths	Limitations
[88]	Fog computing, IoT, ML, WSNs	Presents a smart IoT and ML-based framework for collecting real-time data, predicting flood threats, issuing warnings, and enhancing response efficiency.	99% accuracy in flood detection using decision tree random forest.	Tested only in the city of Jeddah, affecting generalizability and scalability.
[89]	Cloud computing, IoT, MANETs,	Introduces a cloud ecosystem for disaster response that uses IoT-based mesh networks for improved communications when traditional networks fail.	Reliable communication during network failures, supporting SOS data exchange in crisis.	Requires adoption of IoT devices and dependent on connectivity within the mesh.
[56]	5G, AI, cloud computing, CIoV, IoT	Proposes a cognitive internet of vehicles (CIoV) framework for real-time disaster data sharing, leveraging 5G, cloud computing, AI, and ML.	Supports efficient communication for disaster response.	Depends on infrastructure stability, which can be compromised in disasters.
[90]	IoT, IoV, ML, sensors	Proposes an IoT and Internet of Vehicles (IoV) safety system using vibration and smoke sensors to monitor and respond to seismic and fire hazards.	Real-time monitoring enables rapid response and evacuation.	Depends on stable IoT infrastructure, which may be compromised in severe disasters.
[87]	IoRT, RL	Uses deep RL for resource allocation in IoRT networks, optimizing energy and connectivity.	Ensures communication for post-disaster relief through efficient resource allocation.	Complex implementation, high computational and power demand, which can affect scalability.
[91]	IoT, NFV, SDN	Proposes a scalable IoT gateway architecture using network functions virtualization (NFV) and software defined networking (SDN) to enhance network coverage during disasters.	Facilitates flexible and scalable network support in disaster settings.	Requires high computational resources and network resilience, which may be limited during disasters.
[92]	AI, IoT	Develops an AI-powered post-flood management system using IoT to monitor and control flood response.	Real-time post-flood management with automated decision making.	Requires robust IoT device infrastructure.
[93]	Edge computing, IoT	Presents a system that uses wearable sensors and edge computing to monitor the vitals of search and rescue team members, enhancing health and safety.	Real-time health monitoring for search and rescue personnel.	Edge computing relies on stable connectivity, limiting applicability in remote areas.

4.4. Robotics and Unmanned Aerial Vehicles (UAVs) in Disaster Management

Robotics, empowered by microprocessors, sensors, and wireless technology, are invaluable in disaster scenarios where human intervention may be risky [32,38,50,57,87,94,95,96,97,98]. Equipped with wireless communication, cameras, and sensors, robots perform surveillance, navigate through obstacles, access hazardous spaces, assess damages, and conduct search and rescue operations [32,50,57,72,87,94,96,97,98,99,100,101]. Robotics enables remote operation, reducing risks for rescuers and enhancing response effectiveness [32,38,94,99]. Cloud integration further enhances robotics flexibility and accessibility [99].

Unmanned Aerial Vehicles (UAVs), commonly known as drones, provide high-resolution imagery and reconnaissance capabilities, offering unique perspectives and minimizing risks for rescue teams [32,38,50,52,55,95,98,102,103]. These versatile robots can be remotely controlled or operate autonomously using pre-programmed software to survey disaster-stricken areas and analyze real-time data, aiding in terrain mapping and locating victims [52,95]. Compact, cost-effective, and maneuverable, UAVs are ideal for navigating challenging environments, inspecting infrastructure, evaluating damage, delivering essential supplies, and identifying safe rescue routes [50,52,65,95,98,102]. UAVs equipped with sensors detect disasters and assist with 3D mapping of unfamiliar areas to prevent accidents [95]. Unmanned underwater vehicles (UUVs), unmanned ground vehicles (UGVs), and unmanned surface vehicles (USVs) complement UAVs by providing extensive transportation capabilities for search operations and assisting survivors [38,50]. Microrobots are especially promising for navigating small spaces within collapsed structures, increasing the chances of detecting survivors, while evacuation robots offer essential support in chaotic disaster scenarios [50].

Advanced technologies like the Cognitive Internet of Vehicles (CIoVs) further enhance UAV capabilities for real-time disaster management by facilitating information exchange among intelligent vehicles, data visualization, analysis, and dissemination [56]. Tethered UAVs provide reliable communication infrastructures in disrupted areas [104]. UAVs equipped with data communication capabilities, including flying witness units (FWUs), can form collaborative architectures such as flying ad hoc networks (FANETs), enhancing connectivity and data exchange among rescue teams [102]. Integrating the collected data with GIS further strengthens disaster relief efforts [102]. Moreover, coordinating multiple UAVs in a swarm configuration enables the achievement of collective tasks, such as identifying and locating humans, while efficiently sharing crucial information [65]. This ensures that even if some UAVs fail, others can seamlessly continue the mission, maintaining overall rescue efficacy without interruption [65]. Table 5 summarizes robotics in disaster management, highlighting each study’s focus, strengths, and limitations.

Table 5. Summary of robotics and UAVs studies for disaster management.

Study	Technology	Study Focus	Strengths	Limitations
[105]	AI, image recognition, ML	Develops an autonomous robot with rough terrain navigation, obstacle avoidance, and open-source ML for identifying disaster survivors trapped in disaster debris.	Cost-effective and adaptable for deployment in small towns with limited technology.	Limited resources hinder the robot’s performance.
[106,107]	LoRa, LPWAN, robotics	Proposes a LoRa-based low-power wide area network (LPWAN) communication system for search and rescue robots in disaster scenarios, featuring robot coordination and autonomous navigation.	Low-power, wide-area network for robust communication.	Limited communication range in heavily obstructed areas.
[108]	Cloud-computing, robotics	Presents a cloud-based multi-agent framework for managing aerial robots (drones) in disaster response.	Enhances disaster surveillance and decision support using autonomous drones.	System’s timeliness may suffer under high data volumes due to computational demands.
[109]	HPC, photogrammetry UAVs	Develops a high-performance UAV photogrammetry workflow with high-performance computing (HPC), reducing processing time for large datasets.	Reduces processing time by up to 86% with high resolution data.	High computational resource requirements for large dataset processing.
[110]	DL, UAVs	Creates a self-supervised DL method for rapid post-earthquake building damage detection using UAV data.	High accuracy in detecting building damage. Supports real-time application.	Requires high-resolution UAV data.
[111]	AI, drones	Explores the potential of drones and AI for enhancing disaster response.	Faster response, improved decision-making.	Technical, regulatory, and ethical challenges in deployment.
[112]	MANETs, robotics	Presents an architecture for rescuer robot team based on multihop MANETs for efficient search and rescue in disasters.	Operates autonomously and reduce risk to rescuers.	Relies on complex MANETs, unstable and challenging to maintain in disaster scenarios.
[113]	AI, audio recognition, drones, thermal imaging	Explores drone-assisted human detection using AI to locate survivors through thermal imaging and audio signals in hard-to-reach disaster areas.	Enhances detection capabilities and efficiency. Effective in remote, inaccessible disaster areas.	Limited flight duration for drones and high battery dependency.
[114]	CNN, drones, sorting algorithms	Combines CNNs and sorting algorithms for flood severity detection using drones, aiding in relief prioritization.	Improves speed of relief efforts using drones.	Limited performance in complex terrains and under adverse weather conditions.
[115]	advanced path planning algorithms, IoT, robotics	Proposes a robot-guided evacuation system for crowd evacuations in dynamic environments using elliptic tangent graph approach to optimize evacuation routes.	Finds shorter evacuation path to reduce evacuation time compared to other methods.	Challenged by complex dynamic environments which limit real-time adaptability.
[94]	Leap Motion, robotics, wireless sensors	Designs a remote control system for search and rescue robots using Leap Motion and analog sensors for effective remote operation.	Increases rescuer safety by enabling remote robot control in hazardous areas.	Limited by sensor precision and Leap Motion’s sensitivity in cluttered environments.
[116,117]	AI, knowledge-based systems robotics	Develops a knowledge-based system for managing robotic assets during urban search and rescue in collapsed structures, addressing confined-space challenges.	Increases access to confined spaces, reducing rescuer exposure to hazardous areas.	Limited by robots’ durability in extreme environmental conditions.
[118]	DL, semantic segmentation UAV	Evaluates and compares real-time semantic segmentation models for UAVs to identify flood damage in aerial images.	Efficient segmentation models suited for real-time analysis.	Requires high computational resources for real-time processing.
[119]	UAVs, USVs	Cooperative use of unmanned sea vehicles (USVs) and unmanned air vehicles (UAVs) for hurricane damage detection.	Increases situational awareness in complex environments.	Requires significant human oversight.
[120]	3D modeling, UAVs	Deploys UAVs for high-resolution imaging, flood monitoring, 3D model creation, and rescue mission assistance.	Provides quick 3D modeling for real-time decision-making.	Depends on clear visibility and favorable operational conditions for optimal results.
[121]	Bayesian logic, robotics	Proposes a semi-autonomous robot for urban search and rescue, using Bayesian logic for visual detection.	Effective in locating victims, accurate mapping.	Requires manual operation.
[99]	Robotics	Builds a multi-terrain surveillance robot with a robotic arm for remote-controlled rescue operations.	Remote operation reduces risk to rescuers. Cost-effective.	Not autonomous navigation.
[122]	GPS, IoT, robotic sensing	Designs a human detection system using robotic platforms equipped with sensors to locate victims under earthquake rubble using radar, GPS, and IoT.	High accuracy in identifying living individuals under debris.	Limited by sensor range and signal interference.
[123]	HMI, robotics	Designs a human-machine interface (HMI) for search robots in disaster response to replace humans in hazardous areas.	Improves search accuracy and rescuers’ safety through remote operation.	Human supervision required for effective operation.
[95]	Infrared cameras, LIDAR, UAVs	Combines infrared cameras and LIDAR in drones to detect victims in low-light disaster scenarios.	Reliable in low-visibility conditions, fast victim localization.	Not fully autonomous.
[124]	AI, UAVs, YOLO v8	Uses AI-powered drones for accurate object detection and rescue operations to improve response efficiency.	High accuracy in victim detection from aerial imagery.	Depends on the quality of pre-trained models and data availability.
[125]	Dynamic mapping, HMI, robotics	Proposes a robotic system for dynamic mapping in disaster scenarios like earthquakes and structural collapses, offering rescue teams autonomous path feedback to locate victims in inaccessible zones.	Provides dynamic mapping and autonomous navigation for effective rescue.	Signal disruptions may affect real-time data transfer and control.
[102]	FANETs, UAVs	Proposes FANETs to establish real-time communication in remote disaster areas with limited connectivity.	Reliable real-time communication in disaster zones.	Highly dependent on stable flight and connectivity.
[126]	Remote sensors, robotics	Develops a ground-based mobile robot platform for search and rescue in rubble.	Designed for high mobility and rugged conditions.	Constrained by battery life.
[127]	Hyperspectral imaging, robotics, sensor fusion	Explores the use of hyperspectral imaging for detecting victims in post-disaster environments.	Accurately detects victims under challenging, low-visibility conditions.	Sensitive to environmental interference and requires specialized equipment.
[128]	IoT, UAVs, UGVs, data fusion	Presents a data management system for coordinating UAV and UGV fleets to support search and rescue efforts, with data processed in real-time for situational awareness.	Increases efficiency in search and rescue operations through real-time data centralization.	Depends on communication networks and can suffer from data overload in high-density scenarios.
[129]	3D mapping, RS, robotics, sensors	Introduces a rescue robot system for gathering information in rubble piles during earthquake search and rescue missions.	High adaptability and durability in rubble environments.	Teleoperation relies on skilled operators and 3D mapping accuracy is reduced in cluttered environments.
[96]	Robotics, wireless communication	Describes a prototype of a versatile, low-cost four-legged robot that can walk, climb, and fly in disaster recovery missions.	Adaptable to multiple terrain, cost-effective.	Navigation accuracy may be impacted in extreme terrains.
[130]	SOMs, UAVs	Uses UAVs with self-organizing maps (SOMs) to restore network coverage during disasters.	Quickly establishes communication in affected areas.	Relies on UAV battery life and is affected by extreme environmental conditions.
[104]	Genetic algorithm, UAVs	Uses genetic algorithms to optimize tethered UAVs deployment, ensuring connectivity in disaster-affected areas.	Optimizes UAV placement for communication stability.	Limited by tethered constraints and operational deployment challenges in adverse conditions.
[65]	PSO, UAVs	Integrates PSO techniques with face detection and object tracking algorithms to develop UAVs for aiding in disaster scenarios.	Fault-tolerant system, ensuring continuous operation despite individual UAV failure.	High complexity in swarm management and coordination under dynamic conditions.
[131]	IoT, UAVs	Proposes a drone-based multihop ad hoc network to reestablish connectivity in disaster zones, using drones to communicate data between rescue teams and affected areas.	Enables communication in disaster areas, adaptable to multiple disaster scenarios.	Relies on stable drone positioning and limited by drone battery life.
[132]	AI, RL, UAVs	Uses UAVs with RL for real-time flood navigation and path planning for waterborne vehicles.	Efficient route planning under dynamic conditions.	Battery limitations impact operational continuity.
[133]	Drones, IoT	Develops a model to use drones equipped with radio frequency (RF) pose sensors to locate survivors in collapsed structures, integrating IoT devices to improve search and rescue.	Enhances survivor detection in challenging conditions.	Expensive to use and maintain RF-pose technology and drones. Limited by sensor range.
[134]	DL, UAVs	Develops a you only look once (YOLO)-based DL model for identifying and counting flood survivors (humans and animals) in UAV images to assist in rescue operations.	High precision (98%) in identifying survivors.	Depends on high-quality drone images and high computational demand for real-time data processing.
[135]	UAVs, WMN	Introduces wireless mesh network (WMN) using UAVs and ground nodes to establish communication in post-disaster scenarios.	Provides robust connectivity over large areas.	Limited by UAB flight time and requires high UAV count to cover large areas.
[136]	AI, blockchain, drones	Proposes a blockchain and AI-based flood detection system using UAVs for real-time data collection and secure information sharing.	Secure data sharing ensures information integrity during rescue operations.	High computational demands for blockchain implementation on UAVs.
[137]	Robotics, thermal sensing	Proposes a snake-like robot equipped with thermal sensors and cameras to locate survivors in post-earthquake debris and collapsed buildings.	Navigates narrow spaces effectively, enhancing survivor detection.	Limited by battery life and thermal sensor range.
[138]	Robotics, sensors	Develops robotics for earthquake response, focusing on victim detection, structural assessments, and environmental monitoring.	Increased efficiency and safety for search and rescue.	High cost and technical complexity.
[139]	Advanced sensors, robotics	Introduces a multi-technology kit enhancing search and rescue operations with miniaturized robotics, sensors, and situational awareness tools.	Reduces victim detection time and increases rescue team safety.	Limited by technological infrastructure.
[140]	UAVs, UGVs, USVs, UUVs	Establishes a collaborative international framework using heterogeneous UAV/UGV/UUV/USV robotic teams for flood and landslide disaster management.	Enhances cross-border disaster response using collaborative robotics.	Requires robust coordination and collaboration across international teams and systems.
[97]	IoT, robotics	Develops a modular, semi-autonomous robot with articulating chasses to locate survivors in disaster environments, especially within debris.	Effective in navigating complex terrains and tight spaces.	Limited by battery life and signal range.
[98,141]	Robotics, sensor networks	Integrates a network of sensor-based mobile robots to support search and rescue, focusing on victim detection and disaster assessment.	Faster victim detection and area reconnaissance.	Challenges with real-time integration of all robots and sensors
[142]	AI, robotics	Explores using robotics in disaster response with simulations for urban search and rescue.	Provides real-time simulation and virtual reality (VR) training for responders.	Challenges with standardized communication and data integration.
[143]	AI, CNN, DL, UAVs	Uses UAVs with CNN-based DL models to detect flood-affected areas from aerial imagery in Pakistan.	High accuracy flood detection using UAV imagery and CNN.	Depends on high-quality imagery and robust CNN training.
[144]	IoT, robotics	Proposes a semi-autonomous rescue robot equipped with a wireless camera, digital signal processor (DSP), and GPS to aid in locating and rescuing victims in hazardous areas.	Enhances rescue efficiency with real-time GPS data and semi-autonomous functionality.	Limited to outdoor environments with strong GPS signals.

4.5. Information and Communication Technology (ICT) in Disaster Management

Robust information management systems are crucial to disaster relief, with information and communication technology (ICT) enabling precise, timely, and accessible information to support relief operations [49,55]. ICT enhances collaboration, decision-making, and damage assessment. Tools such as geographic information systems (GIS), online platforms, social media, unmanned aerial vehicles (UAVs), and AI accelerate the speed and coordination of relief operations [49,55].

Big data analysis plays a critical role in extracting valuable insights from vast datasets. It provides real-time information from sources like social media to improve response prioritization, recovery planning, and understanding situational awareness, which is essential for effective response [31,39,52,145]. Crowdsourcing offers an efficient, cost-effective method for data collection and analysis, leveraging AI and ML to generate high-quality structured data that strengthens disaster response [23,52,146,147]. Social media platforms, particularly Twitter, enable real-time crowdsourcing information sharing during disasters, raising awareness promptly [147]. Cloud computing offers quick access to resources, data storage, and applications, facilitating collaborative efforts and supporting real-time decision-making [31,148]. Additionally, cloud technology supports 3D simulation environments using data from wireless sensor networks (WSNs), which aid in training and response preparation [149]. The integration of AI with cloud platforms enhances two-way communication and strategic planning, ensuring operational agility and optimized resource allocation [64].

Intentional islanding, supported by AI controllers, offers efficient management of distributed energy resources (DERs) during grid disruptions caused by disasters [37,98,150]. This technology isolates power networks, identifies affected loads, prioritizes power distribution to critical facilities such as hospitals, regulates energy to prevent overloading and disruptions, and ensures efficient restoration [37,98,150].

Ad hoc networks (ANETs) are essential in disaster response due to their rapid deployment, fault tolerance, and reliability [58]. Combining UAVs with ground-based ANET nodes forms resilient air–ground ANETs (AGANETs), ensuring uninterrupted communication despite disruptions [58]. Hybrid ad hoc networks, integrating IoT devices and smartphones, enhance emergency communication, while mobile ad hoc networks (MANETs) enable essential wireless communication for rescue operations [56,58,89,102]. Wireless communication technologies, including ultrawideband radio, infrared-ultrawideband, Doppler radar, global system for mobile communication (GSM), and global positioning system (GPS), are key for disaster response management systems [50]. These technologies enable precise indoor location detection, support movement accuracy through radar systems, and facilitate survivor detection, with GPS proving especially useful in remote areas [50]. Delay-tolerant networks (DTNs) maintain communication for mobile nodes like rescuers, and wireless telemedicine systems enable real-time, cost-effective data transfer for survivor care [50]. Table 6 outlines ICT technologies in disaster management, highlighting each study’s focus, strengths, and limitations.

Table 6. Summary of ICT studies for disaster management.

Study	Technology	Study Focus	Strengths	Limitations
[46,151]	Data-based inductive reasoning, DSS	Develops a decision support system (DSS) using data-based reasoning to support humanitarian NGOs during disaster response and recovery.	High accuracy in disaster damage assessment and guides decision-making.	Requires historical data for accuracy.
[148]	Cloud collaboration, LIS	Application of distributed data centers and logistics information system (LIS) in cloud collaboration for efficient disaster relief management, demonstrated through a case study on the 2015 Nepal earthquake.	Enables quick, efficient information sharing and logistics.	Prone to network overload and data synchronization challenges in large-scale crises.
[64]	AI, Cloud computing	Develops an evidence-based framework, named 4-AIDE, integrating AI and cloud-based collaborative platforms for effective disaster and extreme weather response.	Improves cross-agency coordination and predictive accuracy in disaster response.	Highly dependent on data quality and availability, with challenges in integrating data sources across agencies.
[53]	AI, GIS, web application	Develops a web application with offline functionality for real-time disaster management, focusing on resource allocation, monitoring, and coordination.	Operates without internet and offers comprehensive support tools.	Limited by network infrastructure.
[150]	AI, DER	Proposes an AI-based intentional islanding algorithm to enable power distribution to critical loads like hospitals using DERs during grid unavailability caused by disasters.	Ensures power resilience for critical infrastructure.	Performance affected by varied renewable energy outputs.
[69]	AI, ML	Proposes a framework to automate the identification of communities in need during crises using data-driven techniques and AI.	Improves aid targeting through automated identification.	Data scarcity may affect detection accuracy.
[152]	AI, DL, edge computing	Uses Ai-driven edge computing and virtual simulations to enhance real-time decision-making in disaster scenarios through aerial data analysis.	Reduces latency in data processing, enabling quick response.	Relies on a stable network for optimal performance.
[153]	Crowdsourcing, DTN	Proposes a framework using crowdsourcing and delay tolerant network (DTN) for information sharing during emergencies when connectivity is limited.	Provides real-time information exchange in low connectivity environments.	Depends on the availability and reliability of crowdsourced data.
[154]	ML, semantic web	Introduces a semantic model for improving data interoperability and advanced analysis for disaster responders.	Improves data interoperability and enhances first responders’ decision-making.	Challenges in harmonizing heterogeneous disaster-related data.
[37]	ML, SCADA	Proposes an ML algorithm for intentional islanding of DERs in low voltage direct current (DC) distribution systems post-disaster, using live data from supervisory control and data acquisition (SCADA) systems.	Enhances decision-making accuracy in energy management during disasters.	Depends on data quality from SCADA systems.
[155]	ML	Uses ML to optimize contraflow usage based on real-time traffic for improved hurricane evacuations.	Enhances traffic flow efficiency during evacuations.	Relies on extensive, accurate real-time traffic data and processing.
[156]	Ant colony optimization	Proposes an algorithm for efficient route planning during disaster relief distribution, incorporating real-time disaster information updates.	Optimizes route planning for timely disaster relief delivery.	Accuracy depends on timely information updates.
[157]	Multi-objective optimization, NSGA-II algorithm	Develops a multi-objective model using NSGA-II to optimize emergency supply distribution post-disaster, considering transport constraints and victim satisfaction.	Optimizes distribution of emergency supplies.	Relies on extensive and accurate data to model real-time needs and requires high computational resources.
[158]	AI, satellite IoT	Presents a framework for scalable, deployable, and cost-effective satellite communication infrastructure for disaster zones.	Increases communication resilience during disasters.	High cost of satellite communication infrastructure.
[159]	AI, cloud-edge computing, federated learning	Proposes a cloud-edge architecture for real-time disaster decision support using federated learning and distributed intelligence.	Scalable and efficient real-time decision-making.	Complexity in implementing federated learning models.

The next two sections discuss geospatial analysis and social media and smartphone applications, both integral components of ICT.

4.5.1. Geospatial Analysis in Disaster Management

A comprehensive disaster management strategy incorporates a systematic approach in planning, including risk reduction measures, recovery plans, and a well-trained response team that engages the community [23,160]. This approach is key for making timely and accurate decisions, minimizing damage, and saving lives [23,160,161]. Geospatial analysis plays a fundamental role in these efforts, providing rapid damage assessment maps and enabling informed decision-making based on spatial data [23,160]. Sharing geospatial data is critical for establishing a centralized disaster data platform, and its absence adversely impacts public services [160].

Technologies such as global navigation satellite systems (GNSSs), geographic information systems (GIS), and remote monitoring management (RMM) lay the groundwork for climate and disaster modeling, while wearable devices enhance location tracking and health monitoring [59]. This integration allows emergency management services to devise diverse strategies for disaster preparation and response, with AI helping in identifying, communicating, and potentially predicting disasters, thus enhancing public warning systems [59]. Additionally, GIS supports strategic planning and real-time decision-making for effective disaster response, enabling the storage, analysis, and visualization of spatial data [23]. As a result, GIS is widely utilized for risk assessment and estimating damages and losses [23,49,162]. Maintaining accurate local GIS data and quickly estimating damage severity and extent through RS imagery are crucial. Combining spatial data with GIS-based multi-criteria evaluation techniques enhances decision-making by creating detailed maps [162,163]. Satellite imagery in RS offers high-resolution data critical for assessing and monitoring disaster impacts, providing an objective means to evaluate potential scenarios. Moreover, integrating satellite RS with GIS further enhances planning, situational awareness, and recovery efforts [23,57,101,163].

Inadequate traffic management on road networks often leads to significant disruptions and safety challenges after disasters. Integrating datasets, including road networks, traffic patterns, and geography, along with satellite images, meteorology, and disaster models, is valuable for identifying affected areas, improving road network management, and defining evacuation routes [52,68,145,164]. Systems designed for aggregating, analyzing, visualizing, and optimizing heterogeneous data enable comprehensive disaster management and facilitate crucial decision support [151]. Decision support systems integrating GIS-based data management and visualization improve communication among local authorities, affected populations, and stakeholders [160,162]. These systems also offer adaptive tools for resource distribution [160,162]. Enhanced decision support systems are crucial for assessing vulnerabilities and aiding in the development of emergency plans and evacuation routes [46,68,151,164]. The integration of GIS and RS facilitates early warning, monitoring, and damage assessment, supporting effective decision-making and disaster response management [53,56,89,164]. Table 7 presents findings on geospatial analysis in disaster management, highlighting each study’s focus, strengths, and limitations.

Table 7. Summary of geospatial analysis studies for disaster management.

Study	Technology	Study Focus	Strengths	Limitations
[59]	AI, GIS, GNSS, RMM, RS	Examines AI in emergency planning, focusing on GNSS for precise data collection (GIS, RS, RMM) and algorithm frameworks to improve climate modeling and traffic control for disaster response.	Increases decision-making speed and accuracy, with GNSS providing real-time traffic insights.	Integration with current infrastructure may be challenging.
[165]	DEM upscaling, DL, GIS	Develops a high-resolution flood mapping system using DL with digital elevation model (DEM) upscaling, and GIS-based floodwater estimation.	Accurately estimates flood extent and depth, improving urban flood response.	Relies on accurate DEM availability and may not perform well in densely vegetated areas
[68]	DL, ML	Develops a model to assess damage and detect safe evacuation routes using satellite images and DL.	Improves route accuracy, adaptable to real-time images and dynamic conditions.	Depends on satellite image availability
[166]	AI, RS, TL	Applies TL to high-resolution imagery for detecting post-disaster building damage.	High accuracy for damage detection without pre-event images.	Limited by availability of high-resolution post-event images.
[167]	Conditional copulas, data fusion, SAR	Examines spatial multi-sensor data fusion using synthetic aperture radar (SAR) and optical datasets to assess tornado damage.	Provides high Kappa accuracy (75%) in change detection, useful for immediate response. Cost-effective.	Challenges with model generalization due to data heterogeneity.
[47]	CNN, DL	Uses DL to assess building damage severity from satellite images pre- and post-disaster.	Accurate damage detection and efficient at classifying various damage levels.	Requires high-quality satellite imagery for precise analysis.
[168]	GIS, RS	Creates an RS-based earthquake response system using digital earth technology to quickly assess damage, share seismic data, and expedite disaster response.	Efficient data management and real-time sharing across user levels.	Requires stable digital earth platform infrastructure.
[169]	GIS, queuing theory	Develops a GIS-based DSS using queuing theory to efficiently allocate resources for post-earthquake search and rescue operations.	Improves resource allocation speed and precision.	Limited by real-time data availability.
[170]	RS, SAR imaging, classifier algorithms	Utilizes Sentinel-1 satellite images and classifier algorithms to detect and assess flood zones accurately.	High accuracy in flood detection using satellite data.	Depends on satellite data. Data processing can be time-consuming.
[171]	GIS, ML, RS, SAR	Develops a spatial framework integrating ML and survey data for flash flood assessment and fund disbursement in Pakistan’s flood-affected regions.	High accuracy in mapping flood impact zones and improved funding transparency.	Relies on robust data collection, which can be challenging in crisis scenarios.
[172]	DSS, GIS	Introduces a GIS-based DSS for cyclone emergency response in Taiwan.	Effective for real-time hazard assessment and decision-making support.	High dependency on data accuracy for effective response.
[173]	Big data analytics, GIS	Develops a CyberGIS framework to combine social media and GIS data for disaster monitoring. Tested using Hurricane Sandy.	Effective integration of multi-sourced data for real-time mapping and event tracking. Scalable.	Handling high-data volumes poses computational challenges.
[162]	Distributed computing, GIS	Uses distributed computing for GIS image processing to develop a DSS for disaster damage assessment and resource allocation	Speeds up large-scale image processing for disaster response.	Requires high computational resources.
[174]	GIS, ML, RS, SAR	Integrates ML and RS with Google Earth Engine for real-time flood impact assessment, using SAR to map affected areas.	High accuracy in flood impact assessment and land use/land cover mapping.	Relies on Google Earth Engine, high-quality data, and data processing infrastructure.
[175]	RS, SDI	Develops a web-based system for flood information delivery using spatial data infrastructure (SDI) services to improve decision-making.	Provides automatic mapping and real-time resource delivery.	High dependency on geospatial data availability.
[176,177]	Crowdsourcing, edge detection, DL, DNN	Utilizes image processing and DNN to estimate flood depth from street photos by detecting submerged stop signs, aiding in real-time flood assessment.	Provides accurate depth estimation to support flood response and evacuation planning	Depends on the availability of crowdsourced data and clear, correctly angled photos.
[178]	CNN, dense neural networks	Leverages satellite imagery, dense neural networks, and CNNs to identify flooded areas and suggest safe evacuation routes in real-time.	Offers high flood detection accuracy and real-time safe pathfinding during disasters.	Relies on continuous data updates from satellite images and requires high computational resources.
[179]	DL, RS, SAR	Proposes flood mapping using synthetic Sentinel-1 radar data and optimized U-Net models for efficient flood detection with fewer parameters.	High detection accuracy and fast processing.	Synthetic data may impact real-world application. Requires large datasets for training DL models.
[180]	CNN, DL, RS	Proposes a CNN-based model for multiclass damage classification of buildings post-Hurricane Michael using high-resolution satellite images.	High accuracy in multi-class damage classification.	Requires high-quality satellite imagery for accurate damage assessment. Data imbalance affects certain damage classes.
[181]	AI, ML	Proposes a multi-layered AI emergency tool that integrates geospatial data and ML for enhancing community resilience and aiding emergency response, tested during Hurricane Florence.	Provides timely, relevant information to optimize evacuation routes.	Data handling challenges due to multi-layer input complexity and reliance on accurate data sources.
[182]	ICT, DSS	Uses China’s national spatial data infrastructure (SDI) to provide geospatial information services for disaster response after the Wenchuan earthquake.	Enhances situational awareness and optimizes disaster response using updated spatial data.	Limited by SDI data update speed and infrastructure requirements.
[183]	ML, RS	Proposes post-event very high-resolution SAR imagery with ML to assess individual damage after earthquakes, tested in Haiti and Turkey.	Achieves high accuracy without needing pre-event images, suitable for rapid damage assessment.	Requires high-resolution SAR data.
[184]	DL, RS, SAR	Uses multimodal RS data to map building damage post-disaster, employing CNNs.	Provides high-resolution and accurate building damage assessment.	Impacted by data availability and weather conditions.
[185]	RS, social sensing	Combines social media and RS data to assess flood exposure, damage, and population needs in Pakistan.	Integrates social and RS data for effective crisis mapping.	Requires complex integration of remote and social data sources, limiting scalability.
[186]	ML	Uses random forest algorithms on satellite imagery to map flood extents in urban areas.	High accuracy classifying flood extent.	Depends on quality and availability of satellite imagery.
[187]	AI, CNN	Uses CNNs on satellite and aerial images for automated building damage classification post-disaster to guide recovery.	Enhances damage assessment accuracy by 4% using high-resolution imagery.	Image resolution variability affects classification performance.
[188]	AI, transfer learning	Presents a flood detection methodology using Vision Transformer models on Sentinel-1 and Sentinel-2 satellite images.	High accuracy across multiple image types, outperforming traditional CNN models.	Depends on high-quality satellite imagery.

4.5.2. Social Media and Smartphone Applications in Disaster Management

The widespread use of smartphones and social media generates extensive data, offering insights for post-disaster research into health, safety, and individual locations [54,145,189,190,191,192,193,194,195,196]. AI-integrated applications serve as central platforms for disaster management, facilitating information dissemination, damage evaluation, aid coordination, and support services, enabling swift and effective responses [53,60,89,195]. These applications allow users to send texts, SOS messages, images, and location data, updating their status and communicating with emergency responders for assistance, even in areas with limited internet connectivity [53,56,89]. Integration with GIS and RS enhances planning, situational awareness, and recovery activities, enabling users to access live maps, mark affected areas, and plan rescue operations [31,53,56,89,190,197]. Smartphone applications further streamline communication between affected individuals and rescue teams, reducing response times and minimizing damage during disasters [23,53]. These applications offer real-time alerts, manage resources, and provide access to essential supplies [23,53]. Additionally, they record victims’ medical conditions, facilitating efficient medical response and evacuation planning for rescue teams [198].

Social media platforms, such as Twitter and Facebook, are increasingly utilized during disasters, offering valuable real-time data for response efforts and facilitating volunteer mobilization and information dissemination to affected communities [23,32,39,48,49,50,52,54,57,61,70,100,147,189,190,191,192,193,194,195,196,199,200,201,202,203,204,205,206]. They provide ground-level insights, allowing a comprehensive understanding of disaster impacts [39,70,201,202,206]. To analyze social media data, AI techniques like ML, data mining, DL, NLP, sentiment analysis, computer vision, and CNNs are used to process and categorize textual and multimedia content [23,48,52,54,55,61,70,147,189,190,191,192,193,194,195,199,200,201,203,204,205,206]. Supervised and unsupervised ML, sentiment analysis, and topic modeling are crucial for filtering and summarizing social media data [48,69,147,189,190,199,200,201,204,205,206,207]. Sentiment analysis helps understand public sentiment, including panic and concerns, while multimedia content analysis enhances situational awareness [48,100,147,197,199,201,204,205,207]. This information aids crisis managers and responders, supporting the development of automated disaster response management systems [48,147,199].

Social sensors, integrating social media platforms with data analysis, play a pivotal role by transforming these platforms into data collection channels, allowing the extraction of valuable insights [146]. These sensors contribute to situation awareness, event detection, damage assessment, and information dissemination, enabling communities and authorities to respond effectively to the challenges posed by the disasters [146]. Table 8 summarizes findings on social media and smartphone applications in disaster management, underscoring each study’s focus, strengths, and limitations.

Table 8. Summary of social media and smartphone applications studies for disaster management.

Study	Technology	Study Focus	Strengths	Limitations
[208]	Geo-coding, geo-parsing, ML	Uses social media data for real-time flood mapping and crisis response during the Chennai floods, leveraging ML	Achieves 89% accuracy in real-time crisis mapping using Twitter data.	Limited to Chennai floods and Twitter data, with retrieval restricted to only 7 days.
[209]	AI, NLP, sentiment analysis	Explores open-source AI tools for analyzing social media disaster data.	Open-source tools provide insights with low effort.	Inconsistencies in sentiment analysis results.
[210]	Domain adaptation, genetic algorithm, ML	Presents a genetic algorithm-based domain adaptation framework (GADA) for classifying disaster tweets in large datasets.	GADA improves classification accuracy and reduces training time.	Restricted to English-language datasets and binary classification.
[211]	ML	Develops an ML method to identify disaster-related tweets, supporting response efforts with geospatial social media data for search and rescue, damage assessment, and monitoring.	Efficient and quick identification of disaster-related tweets, enabling faster response.	Social media data can include noise and misinformation, as well as miss relevant details.
[212]	Geo-parsing, ML	Analyzes Twitter data for disaster insights, using ML and geo-parsing to support flood response and situational awareness.	Automates data processing and visualization.	Probability of misinformation and lack of precise location in most geo-referenced data.
[213]	ML, NLP	Proposes a framework to capture different levels of situational awareness from Twitter data during disasters. Tested with geo-tagged data from Hurricane Michael.	Provides real-time situational insights for different stakeholder needs.	Geo-tagged tweets are less than 10% and disadvantaged groups may not have access to Twitter.
[48,199]	AI, computer vision, NLP	Analyzes Twitter data during three hurricanes to provide real-time insights for disaster response using NLP, AI, and computer vision.	Enhances situational awareness and actionable insights for responders.	Needs real-time access to social media data. Challenging to filter high-volume data.
[214]	ML, sentiment analysis	Analyzes disaster sentiment on Twitter in English and Japanese during hurricanes and earthquakes to aid in planning communication strategies.	Bilingual sentiment analysis, providing insights into cultural differences in disaster responses.	Scarcity of data.
[215]	DL, latent Dirichlet allocation (LDA) modeling	Develops a framework for flood data extraction from social media, validated in the Shougang flood case, combining DL and regular expressions to achieve high data accuracy.	Achieves 83% accuracy in data extraction and provides comprehensive flood analysis.	Limited by data density in social media, which may not cover critical areas sufficiently.
[145]	Android, EMS, GIS, web platform	Implements an emergency management system (EMS) using GIS, Android, and a web platform to enhance emergency response and resource allocation in real-time.	Provides quick response capabilities, improving resource allocation accuracy.	Limited by network dependency for real-time updates in areas with weak connectivity.
[216,217]	ML, NER, NLP, RNNs, sentiment analysis	Uses RNNs for named entity recognition (NER) on Twitter data, focusing on identifying key locations and resources during earthquakes.	Enhances situational awareness through real-time location tracking.	Dependent on Twitter data and geo-tagged tweets. Accuracy may be impacted by informal language and grammar.
[146]	Data mining, sentiment analysis, social sensors	Reviews social sensor application for real-time disaster monitoring, information dissemination, and public sentiment analysis.	Cost-effective, real-time coverage of disaster areas through social media.	Data quality and relevance can vary, with potential for misinformation.
[218]	NLP, topic modeling, TL	Uses TL, topic modeling, and georeferenced data from crowdsourced applications to assess emergency needs and enhance community-level disaster response.	Enhances situational awareness and response speed.	Limited geolocation data and accuracy in social media data.
[219]	ML, NLP	Automates actionable information extraction from microblogs, using ML to retrieve and match posts on resource needs and availability.	Increases retrieval accuracy for aid coordination.	Dependent on available social media data and language consistency.
[220]	BERT, DL, ML, TL	Uses bidirectional encoder representations from transformers (BERT) and ML models to classify tweets from cyclones, enhancing actionable insights.	High classification accuracy for disaster tweet categories.	Dependent on large, labeled datasets.
[221]	Bi-LSTM, DL, ML, neural network, SVM,	Applies support vector machine (SVM) and bidirectional long short-term memory (Bi-LSTM) to categorize tweets by disaster type for crisis management.	Improves efficiency in tweet categorization.	Informal language and ambiguity reduce model performance.
[222]	ML, NLP, random forest	Uses random forest to classify disaster-related tweets to identify locations during crisis.	High accuracy in classifying tweets, enhancing situational awareness.	Limited by tweet length and informal writing styles.
[207]	LDA, ML, semantic topic analysis, spatiotemporal analysis	Uses ML, integrating topic modeling with spatiotemporal analysis, to assess damage through social media data.	Real-time data enables mapping of disaster impacts, enhancing damage assessment.	High data dependency for accurate mapping.
[223]	CNN, data mining, DL	Uses CNN-based DL model to classify tweets from hurricanes Sandy, Harvey, and Irma for real-time situation awareness.	Real-time social media data classification with high accuracy for cross-event disaster scenarios.	Cross-event classification accuracy is impacted by varying language consistency and format.
[70]	CNN, DL	Employs CNNs to segment and classify real-time earthquake images from social media to assess damage.	Increases rapid response through real-time visual data.	Low accuracy in identifying relevant images due to non-uniform social media content.
[224]	AI, drones	Uses AI-driven person detection from drone-based aerial viewpoints and social media analytics to improve disaster response.	Improves detection accuracy, supporting faster victim location.	Dependent on high-quality images and consistent social media data.
[225]	ML	Uses ML to categorize tweets for situational awareness in disasters, improving response efforts.	Enhances real-time situational awareness for rapid response.	Limited by the quality and relevance of social media content.
[226]	BERT, DL, ML, LSTM	Classifies disaster-related multilingual tweets using ML and DL models like BERT.	High accuracy in multilingual tweet classification.	Performance varies across languages due to complexity.
[227]	DSS, NLP	Uses social media and NLP to develop a DSS for managing earthquake emergencies.	Quick identification of relevant disaster information.	Inconsistent data quality can reduce accuracy.
[228]	ML	Develops an emergency situation awareness (ESA) platform to monitor real-time Twitter data, using ML for event detection.	Real-time event detection and monitoring capabilities.	Relies on social media data accuracy and timeliness.
[229]	Crowdsourcing, ML, NLP, support vector machine	Identifies actionable information from tweets for better disaster management using ML.	SVM achieves over 74% accuracy in extracting actionable information.	Relies on social media data, which may have quality issues.
[189]	ML, NLP	Classifies tweets as informative or non-informative to help emergency manager filter relevant data.	Enhances situational awareness by reducing data clutter.	Data quality issues due to informal tweet language.
[230]	ML, NLP	Uses NLP and ML to classify disaster-related social media posts to direct relevant information to responders.	Effective filtering of relevant posts to improve response time.	Limited by class imbalance and noisy data.
[231]	AI, GPT models	Uses GPT models guided by geo-knowledge to enhance the extraction of location data from social media during disasters	Improves location extraction accuracy by over 40% compared to named entity recognition (NER) tools.	Geo-GPT-4 model has difficulty distinguishing certain location categories. Limited generalization across events.
[232]	AI, NLP, sentiment analysis, spatial analysis	Analyzes social media data for flood management through sentiment analysis, topic modeling, and spatial-temporal analysis.	High accuracy (91%) in evaluating public opinion.	Variability in data quality due to informal language.
[233]	DL, LSTM, NLP	Uses DL and spatial analysis on tweets from Hurricane Irma to track disaster impacts and categorize affected areas.	Improves spatial awareness during disasters.	Limited by quality of social media data and noise.
[234]	ML, smartphone application	Develops a mobile application using fuzzy interference and ML to improve survival rates by prioritizing victims and safety guidance.	Enhances victim-rescuer communication, leading to quicker response times.	Relies on mobile infrastructure, which can be compromising in severe disasters.
[235]	RS	Merges social media data with RS for flood impact analysis, offering real-time assessment of disaster-affected areas.	Provides real-time public sentiment and situational data.	Data variability due to informal language, impacting accuracy.
[236]	ML, NLP	Uses ML to analyze word usage patterns in social media for automated early emergency detection and real-time alerts for faster response.	Provides timely, automated emergency identification.	Depends on data availability. Accuracy may be impacted due to language inconsistency.
[237]	ML, NLP	Presents ML framework for analyzing and categorizing disaster-related tweets to provide real-time situational awareness.	Accurate tweet classification using multiple ML algorithms.	Varying language styles may affect accuracy.
[238,239]	ML, NLP, sensors, sentiment analysis	Presents multimodal ML models to analyze Twitter and physical sensor data to assess physical and social impacts before, during, and after hurricanes.	Integrates social media and physical sensor data for a comprehensive impact assessment.	Requires extensive data cleaning and consolidations, adding complexity.
[240]	ML, NLP	Proposes a model for classifying disaster-related tweets using multimodal approach, including text and image.	High accuracy in tweet classification.	Depends on multimodal data and may require significant computational resources.
[241]	ML, NLP, sentiment analysis	Focuses on understanding human behavior and factors that drive tweet popularity during disasters.	Provides insights into factors that increase tweet impact during disasters.	Relies on tweet popularity. Limited to social media engagement.
[200]	DL, ML	Develops a multi-task learning framework to classify social media images across various disaster-related tasks.	Multi-task learning improves processing efficiency and performance across tasks.	High computational requirements for multi-task learning.
[242]	ML	Uses ML to filter and classify disaster-related images from tweets to help responders prioritize emergency aid.	Improves efficiency by filtering non-essential data.	Accuracy may vary based on available training data.
[243]	Forest optimization, ML	Classifies disaster tweets for resource management using forest optimization.	Effective prioritizing resource needs based on optimized classification.	High computational resources and relies on large, high-quality datasets.
[198]	GPS, smartphone application	Presents an Android application for victim assessment using GPS and mass casualty triage, recording medical conditions, storing data offline, and syncing to a server when connected.	Enables quick victim assessment and data transmission.	Depends on network availability. Without connectivity, data transmission is delayed.
[244]	DSS, GIS, smartphone application	Proposes a centralized DSS with smartphone applications for real-time data sharing, notifications, and coordination.	Real-time data sharing and DSS enhance coordinate response.	Relies on stable telecommunications, which may fail in severe disasters.
[245]	CNN, ML, TL	Uses TL to classify disaster images posted on social media by urgency and relevance to assist responders.	Enhances response prioritization using image-based urgency classification.	Relies heavily on labeled data and initial training, limiting real-time scalability.
[246]	ML, sentiment analysis	Uses ML and social media data for a comprehensive flood impact assessment in Guangzhou.	Enables real-time flood impact assessment.	Depends on availability and quality of social media data.
[247]	CNN, SVM	Develops a CNN-based system for classifying disaster data from Twitter to analyze panic levels, helping to inform response strategies.	Provides a real-time understanding of public sentiment.	Social media data can include noise and misinformation.
[248]	ML	Proposes a method to detect disaster-impacted areas in near-real-time using semantic and geospatial ML applied to social media data.	Provides timely, high-resolution insights, enhancing situational awareness.	Relies on social media data availability and accuracy.
[249]	ML	Combines social media data with physical and socioeconomic variables to predict hurricane damage and prioritize response.	Improves damage prediction accuracy with multi-source data integration.	Limited by data quality and availability of socioeconomic variables.
[250]	ML	Combines social media data and weather information to increase flood situational awareness in urban areas, improving response precision.	Improves situational awareness by combining multiple data sources.	Relies on real-tome social media data, which may contain noise.
[251]	ML, sentiment analysis	Uses sentiment analysis of social media data to analyze public emotions during floods.	Real-time sentiment data enhances situational awareness.	Depends on accuracy of sentiment data from social media. Requires extensive filtering.
[252]	CNN, ML, TL	Uses Twitter to extract disaster-related information using CNNs and TL to classify images and categorize disaster types.	Achieves high classification accuracy (98%).	Relies on social media data availability and accuracy.
[253]	ML, sentiment analysis	Classifies sentiments in disaster-related tweets to improve information dissemination and public response to disasters.	Provides real-time situational awareness and targeted relief recommendations.	High volume of data makes it challenging for real-time processing.
[254]	Geo-parsing	Integrates social media crisis mapping and decision-making support to assist in real-time tsunami and earthquake management.	Real-time crisis mapping, allowing rapid response.	High false-positive rate and data quality variability.
[255]	ML, NER	Assesses urban flooding using social media data, applying named entity recognition (NER) to detect locations and times of flooding events.	Near real-time flood hazard mapping.	Lack of geolocation data in social media and reliance on data accuracy.
[202]	Crowdsourcing, ML	Develops a French-speaking online platform for semi-automatic analysis of Twitter posts during and after disasters.	Enhances situational awareness with public participation.	High data volume and variability in data quality.
[256]	Geo-referencing, ML	Explores the use of social media data in disaster management during the Ianos cyclone, focusing on classification and geo-referencing.	Improves situational awareness through real-time data.	Lack of accurate geo-referenced data in social media and classification complexity.
[203]	NLP	Uses NLP for analyzing unstructured social media data to monitor disasters in real-time.	Real-time monitoring and multi-language support.	High noise in social media data.
[257]	AI, CNN, computer vision, DL, NLP	Uses AI to analyze social media for flood tracking and establishing a passive hotline to guide rescue operations.	Provides situational awareness through real-time data.	Noise in data requires extensive filtering for actionable insights.
[258]	Data mining, ML, NLP	Uses data mining techniques to analyze Twitter data, identifying help requests and resource availability during floods and cyclones.	High accuracy in filtering relevant data.	Difficulty managing high data volumes.
[259]	Geo-tagging	Analyzes urban response to Tropical Cyclone Cempaka in Indonesia through geo-tagged disaster-related Twitter content.	Provides insights into public sentiment and community awareness during disasters.	Dependent on Twitter data and geo-tagged tweets. Difficulty representing remote areas.
[260]	AI, DL	Compares human-coded and DL methods to analyze social media images and improve real-time disaster communication during Hurricane Harvey.	The combination of human insights and DL improve filtering noise and extracting relevant content.	Class imbalance leads to misclassification, and reliance on extensive human labeling affects scalability and efficiency.
[261]	AI, BERT, DL, NLP	Uses BERT and social network analysis to manage social media data from the Nebraska floods, identifying critical information sources.	Enhances situational awareness with improved classification accuracy.	Data limitations and class imbalance.
[262,263]	Computer vision, crowdsourcing, ML	Leverages multi-crowdsourced images from social media for rapid disaster assessment using five ML classifiers.	Real-time assessment and damage prediction from multiple data sources	Limited by data quality and availability.

5. Discussion

Developing countries, particularly low-income communities, face significant challenges in managing vulnerabilities to natural disasters, resulting in extensive and long-lasting infrastructure damage, high mortality rates, and inadequate and delayed disaster response [2,3,264]. These vulnerabilities stem from a combination of factors, including limited resources, lack of education and awareness among the population, inadequate design and construction of buildings and infrastructure, as well as physical, social, and economic inequities [2,19,102,265,266]. The limited golden relief time for rescuing survivors after a disaster, lasting up to 72 h, highlights the need for timely and targeted disaster response measures [102]. These include early warning systems, effective decision-making processes, and swift and safe rescue operations, which remain challenging in these contexts [16,31,69,102].

Disruptive technologies such as AI, ML, and robotics and their applications in geospatial analysis, smartphone applications, and social media hold significant potential for addressing these challenges by accelerating processes, increasing effectiveness and efficiency, and ensuring safety. However, despite their promise, several barriers hinder their adoption in low-income communities, including:

• Social barriers—Social factors play a key role in limiting the adoption of disruptive technologies. These factors include: (1) the low education levels in low-income communities, which affect behavioral intention and are critical for preparedness, prevention, and adequate response [267,268]; (2) a lack of public training and awareness of the benefits of disruptive technologies, complicating response efforts to engage the community, disaster managers, and responders in technology-driven initiatives [267,269]; (3) distrust among stakeholders, including government agencies, NGOs, private industry, local communities, and all parties involved in disaster response efforts, leading to reduced collaboration and decision-making delays [267,270]; (4) the absence of clearly defined roles, responsibilities, and coordination mechanisms, as well as a lack of engagement with technical expertise [267]; and (5) distrust and reluctance to adopt and use new technologies [271].
• Economic barriers—One of the major challenges facing low-income communities is financial constraints [2,19]. Financial factors hindering the adoption of disruptive technologies for disaster response include: (1) high levels of unemployment and poverty, as well as lack of insurance, which impedes access to resources to prepare for and effectively respond to disasters [3,19,20,267,270,272]; (2) reduced local government revenue, limiting the ability to invest in new technologies that are often expensive [267]; and (3) uneven access to financial resources, along with the delayed allocation of funding, which impacts equitable recovery and timely response [267].
• Physical barriers—The physical damage in low-income communities, which often live in informal settlements, exacerbates response and recovery difficulties [2,267,273,274]. These communities experience extensive damage to buildings, transportation systems, and other critical infrastructure, such as water, electricity, and communication networks [2,267]. Furthermore, slow debris removal and contamination hinder quick recovery [267]. These physical conditions present significant challenges for deploying and effectively implementing diverse disruptive technologies for disaster response, which often depend on stable infrastructure and reliable communication networks [56,59,90,92,139,182].

Addressing these barriers requires community engagement and policies that foster equity and inclusivity. These approaches ensure diverse stakeholder participation in disaster response initiatives, enhancing collaboration and leveraging unique perspectives for more equitable and effective outcomes. Moreover, affordable technology solutions tailored for resource-limited communities are essential. Consequently, equity and fairness must be prioritized to promote the adoption of disruptive technologies in low-income communities. Therefore, the disruptive technologies proposed should consider affordability and accessibility, enabling widespread use among individuals with limited income and resources.

Robotics (e.g., drones) and several ICT tools (e.g., social media and smartphone applications), which are both affordable and efficient, can significantly enhance the speed and effectiveness of disaster response [49,55]. While this study synthesized existing knowledge, it assessed the practical applicability and provided actionable insights for integrating disruptive technologies into disaster response strategies, particularly in low-income communities.

Effective disaster management systems rely on accurate data, reliable communication networks, and collaboration among diverse stakeholders [23,49,53,56,57,58]. ICT facilitates the timely collection and dissemination of real-time data, aiding in victim identification, enabling communication with emergency relief services, allowing for the dissemination of alerts and notifications, facilitating damage assessment, and improving decision-making [28,49,55]. Technologies like smartphones and social media networks, such as Twitter and Facebook, are widely utilized during disasters [54,145,189,190]. They enable real-time data collection, fostering a comprehensive understanding of disaster impact and facilitating communication and coordination [39,54,70,145,189,190,201,202].

Additionally, robotics plays an essential role in disaster response management. They can conduct surveillance, access hazardous areas, assess damage, and perform search and rescue operations [32,50,57,72,87,94,96,97,98,99,100,101]. Drones, which are cost-effective and efficient, can survey disaster-affected areas, deliver essential supplies, locate victims, and identify safe routes for both rescue operations and evacuation [50,52,65,95,98,102].

Leveraging these affordable technologies in low-income communities can significantly enhance the efficiency and promptness of disaster response efforts. By prioritizing equitable access to these technologies and involving local communities in the planning and implementation processes, disaster response can become more inclusive and effective in meeting the needs and challenges of low-income communities. Furthermore, future innovations and efforts should aim to reduce costs and maximize efficiency, potentially through partnerships with technology developers who can provide low-cost solutions tailored to low-income communities’ needs.

6. Limitations and Future Work

This research acknowledges certain limitations: (1) it restricts its article search scope to one database, Scopus, potentially overlooking valuable articles from other databases; and (2) subjective factors influence the selection and interpretation of articles. Future studies could delve deeper into the literature review by including additional databases. Furthermore, future work could evaluate the limitations of all the discussed disruptive technologies. The recommendations of technologies for low-income communities are preliminary, and future research endeavors should evaluate the proposed technologies to determine their effectiveness and feasibility in these communities. Future research should focus on translating these findings into practical implementation frameworks and pilot programs to assess the real-world applicability and scalability of identified technologies in resource-constrained settings. Additionally, future research could investigate collaborative opportunities with technology developers to design affordable and scalable solutions that address the needs and challenges of low-income communities, fostering a more effective and timely disaster response.

7. Conclusions

Natural disasters cause extensive damage and economic losses and hinder sustainable development, posing threats to lives and endangering community well-being. With their frequency increasing and recovery efforts often delayed, effective disaster management is of the utmost importance. Disruptive technologies, such as AI, ML, robotics, social media networks, and smartphone applications, offer significant potential to enhance disaster management efficiency. However, their utilization in low-income communities, which are particularly vulnerable, remains underexplored.

Several barriers impact the effective adoption of disruptive technologies in low-income communities, including (1) social barriers, such as low education levels, lack of public training and awareness of the benefits of these technologies, distrust among stakeholders, and a reluctance to adopt new technologies; (2) economic barriers, such as high levels of poverty and unemployment, uneven access to resources, and delayed allocation of funding; and (3) physical barriers, such as extensive damage to infrastructure and transportation systems, along with slow debris removal and contamination, which hinder the deployment and implementation of these technologies. To effectively address these barriers in low-income communities, it is critical to prioritize equity and inclusivity. Affordable and accessible solutions tailored to the needs of resource-constrained communities are fundamental for disruptive technologies to achieve widespread adoption. To this end, this study conducted a comprehensive review of existing literature on disruptive technologies to understand how they can be leveraged to improve the efficiency, effectiveness, and speed of disaster response management. Subsequently, the research explored which of these technologies are the most effective and feasible for enhancing resilience and expediting response in low-income communities, considering the adoption barriers and limited resources of these communities. This review highlights potential opportunities for leveraging disruptive technologies in disaster response, offering insights that can guide future research and practical interventions to address critical challenges in low-income communities.

Evaluating and proposing practical measures for the implementation of disruptive technologies in low-income communities is essential, given their heightened exposure and vulnerability. Such measures could mitigate damages, enhance community well-being, and, most importantly, reduce loss of life. The authors preliminarily propose leveraging three cost-effective technologies, including smartphone applications, social media, and drones, in low-income communities to enhance the efficiency and promptness of response efforts. The findings of this study benefit communities and community stakeholders by addressing disaster management challenges and providing knowledge about disruptive technologies that can be seamlessly integrated into disaster response management, thereby enhancing efficiency and effectiveness.