Enhancing Predictive Urban Planning in European Smart Cities Through AI-Driven Digital Twin Technology: A Case Study of Greece

Abstract

This research aims to assess the contribution of artificial intelligence (AI)-driven digital twin technology in improving the predictive planning of European smart cities, particularly in Greece. It considers the effect of specific elements including simulation accuracy, real-time data processing, artificial intelligence tools, and system readiness on the urban planning process. Structured questionnaires were administered to 301 urban professionals working in smart cities across Greece, focusing on their perceptions of the impact of digital twin features on predictive urban planning effectiveness. Respondents were asked how crucial they found the different features of digital twins in actually improving predictive urban planning. Measurement data were described using the arithmetic mean, standard deviation, and coefficient of variation, while categorical data were described using frequency distribution tables and percentages. This study revealed that the simulation fidelity, available real-time data integration, artificial intelligence analytics, and results- oriented monitoring system maturity have a positive impact on the accuracy, speed, and flexibility of urban planning. Some of the respondents noted these features as very useful for the prediction of urban conditions and decision-making purposes. Nevertheless, some drawbacks related to the computational load and data flow were also revealed. AI-driven digital twins are useful for improving the effectiveness of urban planning. However, they encounter technical issues; therefore, seeking to focus on system maturity and data integration is necessary for their successful implementation. Cities should adopt advanced digital twin technologies and enhance the compatibility of data and maintain AI transparency for better urban planning results.

1. Introduction

Globally, urban centers, especially those in Europe, are experiencing population growth pressure, obsolete infrastructures, and climate change vulnerability which have culminated in the need for advancements in the management of cities [1,2]. Some of the challenges that arise as a result of urbanization include traffic congestion, solid waste management, energy utilization, and environmental concerns [3,4,5]. In response, many cities are adopting smart technologies—AI and digital twin technology—as the leading innovations for the introduction of advanced urban governance and planning [6,7].

Real-life city models or digital twins (DTs) are a relatively new yet powerful concept that enable scientists and policymakers to experiment with physical environments in a simulated environment that replicates a city [8,9]. Across the world, the use of digital twin technology is gradually increasing as governments work towards sustainability objectives, for instance in building smart and resilient cities [10,11]. Currently, cities such as Helsinki and Amsterdam are using the digital twin model, and the recorded outcomes demonstrate the efficiency of spatial twins in urban planning [12,13]. Nevertheless, there is still a lack of attention towards the effectiveness of incorporating digital twins for predictive urban planning while analyzing the practical outcomes of unique digital twin characteristics in real contexts [14,15,16].

Regionally, Europe is ahead regarding the adoption of AI and digital twin technologies to make cities more resilient, sustainable, and smarter [17,18]. Many cities in Europe consider digital twins as revolutionary tools for the better management of cities, although an apparent lack of knowledge about the functions that have the most significant impact on urban planning still exists [19,20]. In this context, it is essential to assess the correlation between the features of digital twins and the urban planning outcomes, especially as cities such as Athens and Thessaloniki are starting to consider the adoption of digital twin technology [3,12].

In the local setting, Greece is a quintessential candidate for the adoption of digital twin technology aimed at addressing the issues of urban transformation, protection of cultural heritage, and resilience to climate change because it is grappling with issues inherent to coastal megacities [3,21]. As Greece implements smart city practices across its population, this paper focuses upon Greece, and more specifically Athens, to evaluate the efficiency of digital twin systems within the prospective urban planning [12,18]. Focusing solely on Greece, this paper seeks to address the aforementioned gap by presenting empirical findings about the performance of digital twins in enhancing urban planning through simulation fidelity, real-time data integration, AI analysis, and system usability as evidenced by the interactions within the end user interface [15,22].

To be precise, the concept of system readiness implies the ability of digital twin systems to be effectively deployed and operated in an urban environment. Some of these factors include the infrastructure compatibility, the availability of data, and the technical ability of the system and the users. Indicators used to measure the aptitude of the systems include the system stability, benchmarking of performance, and the comfort and familiarity of the user with the technology. These indicators assist in finding out whether the digital twin system is functional and in a position to support urban planning processes [19,20].

Thus, the present study is crucial to advance knowledge regarding the application of the digital twin systems to improve the predictability and resilience of the urban planning of smart cities in Europe with an emphasis on Greece [20,23]. Thus, this study will enhance the understanding of the effects of specific DT functionalities and complement the existing body of literature on smart cities; it will provide valuable best practices and guidelines for city planners and policymakers who are considering the use of AI integrated technology in their cities.

1.1. Problem Statement

The rising population density, complexity of infrastructures, and dynamics that characterize most modern cities have put a lot of pressure and strain on existing conventional urban planning procedures [24,25]. Consequently, cities across the globe have begun adopting smart infrastructure technologies with a special emphasis on artificial intelligence-based digital twin solutions for planning and decision making. A digital twin is the creation of an actual model of a city with data streams, simulations, and analytics for modeling and forecasting urbanization processes, further management, and planning [7,8].

Nevertheless, there is still a lack of knowledge about which aspects of the digital twin systems directly help to improve the efficiency of the prediction of urban development. Even though currently cities like Helsinki and Amsterdam are implementing the use of digital twins, there are a lack of case studies to support the application of such systems in urban planning. In reviewing the existing research, we also noted that prior work relied on conceptual models or specific case studies, which are not as statistically valid or generalizable to other urban environments, as these studies by Bibri et al. [26] and Evangelou et al. [3] demonstrate.

Consequently, studies have not been able to identify which key characteristics, including accurate simulation, real-time data integration capability, the artificial intelligence-based analytics feature, maturity level, and the user interface interactivity feature, most significantly contribute to the enhanced results of urban planning [27,28,29].

This gap has resulted in cities adopting and developing complex digital twin systems; however, it is not known which aspects will have the biggest positive impact on the overall performance of the prediction models. This prevents the municipal authorities from improving urban planning and management, ensuring resource optimization, and avoiding poor decision-making using the developed digital twin functionalities. This study thus seeks to address this research gap by assessing the features of digital twins that offer the most optimized outcomes for predictable urban planning, especially in the European region, including Greece.

Greece was specifically chosen due to its unique urban challenges, including coastal vulnerabilities, rapid urbanization, and significant pressure on the infrastructure. These challenges make Greece an ideal case for examining the potential of AI-driven digital twin technologies in predictive urban planning.

1.2. Purpose of the Study

The goal of this research is to assess the effectiveness of using AI digital twin technology as a tool for improving predictive urban planning in several smart cities in Europe with an emphasis on Greece. European cities are now battling issues including increased urbanization, climate change, and the management of infrastructure, and hence there is a need to adopt advanced technologies which enable the simulation, analysis, and forecasting of the urban environment.

1.3. Research Objectives

The purpose of this study is to identify how some of the key attributes of digital twin technology impact the predictive urban planning capabilities within different smart cities in Europe. Specifically:

• To evaluate the impact of increased fidelity on simulation and the impact of simulation on the ability to predict urban planning.
• To assess how real-time data integration affects the effectiveness of predictive urban planning.
• To examine the correlation, if any, between the use of AI-based analytics and the efficiency of predictive urban planning.
• To assess how system maturity affects the ability of predictive urban planning to achieve its aims.

1.4. Research Hypotheses

Hypothesis 1 (H1).

There is a positive relationship between the simulation fidelity level of digital twins and the effectiveness of predictive urban planning.

Hypothesis 2 (H2).

Real-time data integration highly improves the effectiveness of predictive urban planning.

Hypothesis 3 (H3).

AI introduced into digital twins augments the efficiency of predictive urban planning.

Hypothesis 4 (H4).

There is a positive association between the system maturity of digital twins and the effectiveness of predictive urban planning.

2. Literature Review

2.1. Theoretical Framework

The theoretical frameworks used in this study include complex adaptive systems (CAS) theory and the technology–organization–environment (TOE) framework. These theories offer a sound framework for the complex relationships that underpin smart cities and how digital twins can be a component of the solution. The theory of CAS states that cities are made up of several entities that are in constant interaction with each other. It is posited that these components engage in reciprocative dynamics and feedback patterns, culminating in complex behaviors in the future [26,30]. This theory is particularly useful when applied to digital twins, as they are implemented as tools that recreate these complex urban interactions and illustrate how transport, energy, and environmental systems relate [4,31]. In this way, digital twins help the city planners to predict the state of the urban systems and improve the management processes [23,32,33]. For this reason, the CAS framework underscores the relevance of digital twins in responding to evolving urban issues in a dynamic and timely manner [25,34].

The technology–organization–environment (TOE) framework, on the other hand, provides another theoretical foundation, facilitating the analysis of using technologies such as digital twins in smart cities [23,31]. The TOE framework categorizes the factors affecting the extent of implementation as technological factors, organizational factors, and environmental factors [7,23]. The framework is especially useful for explaining how digital twins are implemented in urban contexts, where technology has to interface with organizational and environmental factors such as policies and requirements [26,35]. The analysis of the TOE framework shows that the assessment of external factors, including governmental decisions and market demands, in adopting digital twin systems in the cities is important [3,24]. The TOE model mitigates the shortcomings of CAS because it considers three determinants of technology adoption, namely technological, organizational, and environmental. The framework can be particularly useful for studying the urban context in the adoption of digital twins, as the technology should be in line with organizational processes and environmental factors like policies and market requirements [7,23]. The TOE analysis upholds the role of external effects considerations, such as governmental policies and market tendencies, in the implementation of digital twin systems in cities [3,24].

These two theories can be applied together since they seek to understand cities as complex adaptive systems and consider the internal and external environments through which digital twin technologies are adapted and used [30,36]. Therefore, the CAS and TOE frameworks examine how digital twin characteristics, including simulation realism, real-time data integration, artificial intelligence, and system development, can help improve urban planning in the context of constant technological and organizational change [19,31]. Furthermore, the theoretical analyses reveal that AI and DTs are revolutionary assets in the field of urban planning from different aspects, including sustainability, efficiency, and resilience [4,34]. Based on the findings of Shulajkovska et al. [31] and Zarrabi & Doost Mohammadian [7], digital twins enable enhanced intelligence and timely decision-making for managing cities.

Nevertheless, the question arises as to how these digital twins can be utilized to produce tangible enhancements in the conceptualization and management of the city. To this end, this research aims to contribute to the current body of scholarship regarding how certain features of the digital twin matter to predictive urban planning in European smart cities, especially in Greece [3,9]. As such, adopting the CAS/TOE framework helps to provide an understanding of how digital twin technology can be effectively applied to improve planned and existing cities based on the various issues arising from modern day metropolises [30,35]. The use of these theories in this study will assist in closing the gap between technological possibilities and enactment to assist the cities in embracing sustainable and practical methods of urbanicity.

2.2. Review of Key Constructs of the Study

2.2.1. Simulation Fidelity

Simulation fidelity is the ability of a digital twin to emulate the constitutive physical environment in which it exists. Essentially, the importance of simulation fidelity on the overall precision of predictions and decisions drawn based on the digital twin models. Simulation models improve the effectiveness of urban planners in anticipating many variables including the traffic, energy consumption, and climate within cities [9,31]. Thus, the utilization of high fidelity simulation digital twins to anticipate future city conditions enables the projection of future issues like climate change, infrastructure deterioration, and population increases in smart cities [7,37,38]. Some research has shown that the enhancement of simulation realism leads to better predictions, and at the same time promotes more efficient urban growth through the detection of problematic areas in the system’s efficiency [23,34].

However, realizing such high fidelity becomes increasingly difficult as cities grow more dynamic and integrate with one another. As per Sheraz et al. [39] and Zong and Guan [30], the present state of developing highly accurate models specifying complex urban systems demands improved complex models alongside vast computational power. This results in a compromise between model realism and computational cost, where more realistic models are expected to take a longer time for simulation. Therefore, the analysis of simulation fidelity becomes paramount in determining precise approximate models that are also relevant to real world applications [3,19].

2.2.2. Real-Time Data Integration

Real-time data integration is crucial for the effective operation of digital twins in urban planning as it constantly produces the data that increases the usefulness of the digital twin and improves the response time of the digital system. Smart cities that incorporate IoT devices as well as sensors and other data feeds into the digital twin use the data in real time to control and adapt the urban systems [25,40]. For example, real-time traffic flow data can be used to control traffic signals and predict congestion, or even indicate the best time to use public transport [13,41]. Such real-time integration helps to make the urban planning models dynamic and not only accurate for the existing conditions but also sensitive to changes in the inner city conditions [7,31].

However, there are difficulties related to incorporating real-time data into the models of digital twins. First, the increasing volume of data in the current digital world requires efficient methods of acquired data management and processing to allow for a timely decision making process [23,34]. Furthermore, challenges around data quality and compatibility will arise as disparate data sources may vary in quality, accuracy, and reliability [19,42]. To make real-time data integration into digital twins as useful as possible in cities that are integrating new technologies, these will have to be addressed [4,43].

2.2.3. AI-Based Analytics

AI analytics are another capability of digital twins of cities that enable the use of machine learning and other artificial intelligence methods to analyze large datasets and make predictions about future urban environments [23,30,44]. AI tools reflect the historical and/or current data with the help of digital twins and predict potential urban tendencies, including growth and resource consumption, or the outcomes of emergency situations [4,35]. AI analytics can enhance resource management and public services, and adjust the effects of the environmental and socio-economic changes impacting urban infrastructure [26,31,45].

Nonetheless, incorporating AI-driven digital twins into the planning and monitoring processes of cities is still a challenge because the actual AI models still have to be transparent, explainable, and properly directed towards the requirements and goals of urban planners and designers. These AI algorithms, especially those involving deep learning, are usually labeled ‘black box models’, meaning that it is complex or nearly impossible to decipher the rationale behind some of the predictions or recommendations provided [13,39]. Also, the preparation of datasets for training the AI models poses a challenge, as the data must be of high quality and capture real-world scenarios [25,30]. These aspects must be considered to ensure that AI analytics can be adopted efficiently in the context of the real programming of cities [34,41].

2.2.4. System Maturity

Regarding digital twin systems, the idea of system maturity measures the effectiveness of a system in terms of how well it can continue to operate under new data interfaces and new technology trends, as well as with the ever-changing needs of a city [23,30,34]. Technology readiness and the institutional conditions of the city must be sufficient to support the continued application and development of the digital twin systems long-term [3,46]. A complex and sophisticated system can readily respond to the current dynamics in urban management and is capable of integrating the future changes in existing and/or new technologies as well as changes in the policies or goals for urban development [23,35].

Nonetheless, achieving a high level of system maturity is a long and slow process in which various factors interrelate. Some of the issues are related to the scalability of the digital twin system as well as the continuous incorporation of new data types and technologies [13,42]. Moreover, due to the dynamic environment, new features for system upgrades and incorporating digital twin technology in urban planning may demand massive capital investment and political will [6,7]. This underpins the need to understand that while digital twin systems are technologically sound, their implementation requires proper support from the organizational and legal structures within the city [19,46].

2.3. Empirical Review

2.3.1. Digital Twins in Urban Planning

Research carried out at the European and global levels reveals the ability of digital twin frameworks to improve planning accuracy, robustness, and decision-making speed [3,4,9,47]. For instance, through energy management and traffic simulation applications, the cities can simulate the environment, analyze the consequences of their actions, and modify their plans [19,22,39,48]. It is important to note that there are many examples of digital twin and smart mobility implementations in practice, such as in Greece, Poland, and Finland, to address the challenges in infrastructure and the environment [8,12,13,49].

The digital twin is a popular concept in urban planning that helps cities to build virtual models of their physical infrastructure for testing, assessment, and prognostication. Experience indicates that incorporating digital twins increases the spatial understanding or knowledge concerning the use of land, energy, and transport [3,9,19,25,39]. For example, in Greece the utilization of digital twins has enhanced the organizational structure and management of city construction and emergency strategies [3,22]. In the same way, refs. [50,51,52,53] show that the use of digital twin technology to address resource management in cities has improved the sustainability and livability of cities. These implementations thus depict a move from reactive to proactive urban management—this is in tandem with other global interventions under SDG 11 [10,49].

However, there is a disjointed and technologically incoherent application of digital twins as trends furthering their utilization continually grow. Challenges include a shortage of general reference models and integration with other systems, which limits the application and implementation of digital twins in urban planning [15,35,47,54]. For instance, variations in data collection procedures and model accuracy reduce the applicability of solutions between cities [40,43,55]. These issues are especially seen in emerging economies and in small municipalities where a lack of resources hampers the level of adoption [13,23,56]. Additionally, many of them are pilot-scale projects, with limited funding for sustainable long-term sustenance or sound evaluation strategies [6,52,57].

Past studies also highlight challenges with the actual flow of building digital twins for complex systems, such as urban environments with high emergence and velocity data [2,7,58]. An analysis of cities with an established IoT network system shows that such cities have better model performance in areas such as traffic control, air pollution prediction, and energy supply [42,59,60]. However, the real-time integration of data faces challenges due to the existing system architecture and information security problems [31,61,62]. In addition, a lack of integration between departments and their datasets and poor stakeholder coordination can slow down decision-making, thereby negating some of the advantages of the technology [4,8,63].

The adoption of the concept is also premised on human-centric design and the willingness of end users to embrace the technology. Thus, several case studies point to the need to focus on user interfaces that allow any planner, engineer, or decision-maker to engage with the digital models in an interactive fashion [9,13,64]. In this respect, we have seen that training and capacity-building are known to be crucial enablers in system adoption, but they are frequently neglected during the implementation stages [22,36,65]. Failure to adequately engage with digital twin technology expertise and collaborate with other sectors may lead to underinvestment or a poor understanding of simulations [23,34,66]. Therefore, to understand its potential and the potential opportunities for improving the environmental resilience of cities, it is crucial to consider the social, organizational, and technological components together [18,49,67].

2.3.2. AI Integration and Predictive Urban Planning

AI implementation in urban planning activities has changed the way cities are run, managed, and expanded through models that project infrastructure, traffic, and even environmental policies. Studies have evidenced that AI-based predictive models help a city to identify patterns and to prepare for different phenomena which occur in the urban environment with less reliance on a post hoc approach [23,25,30,34]. For instance, city officials employing ML techniques within their digital twin systems have noted enhanced efficiency in strategic planning, prediction of traffic conditions, and real-time service provision [35,39,54]. AI also plays an important role in analyzing the historical data, which helps in further extending better predictions about the growth of a city to plan for better utilization of the land and resources [7,47,57].

However, there is still quite a large gap between science fiction and reality in terms of AI and its use in daily life. Several works indicated that cities are still facing problems when adopting AI tools into their planning environment due to issues of compatibility, lack of human resources, and funding [6,19,24,52]. This is especially true for the older generations of urban system regions with legacy systems, as the updating of physical structures becomes a major hindrance to the integration of new AI modules, leading to scalability issues [3,9,40]. Furthermore, while AI is very effective in processing large amounts of data, the quality of the output depends on the quality, availability, and consistency of the data used; in many European urban jurisdictions, these are still a mixed bag [10,13,43]. This highlights the importance of developing rich datasets to support AI adoption in planning activities [8,49].

Another empirically observed issue is that many AI systems are ‘black boxes,’ and this hinders policy and explainability. Urban planners have insufficient insight into how AI-generated forecasts are produced, therefore impeding the conversion of such forecasts into planning policies [61,62,66]. This unliveliness leads to a lack of trust and thwarts the integration of AI into governance processes [55,64,68]. In addition, the biases that are introduced through the design of the algorithm and the input data may serve to perpetuate existing inequalities in the urban environment, especially in those areas where data are scarce [18,21,36]. Therefore, while promoting equity in urban economies through the use of AI is effective, it should be augmented with ethical and regulatory solutions [4,22,67].

Overall, AI-guided urban planning can be successful when it is incorporated within the digital twin context. Experiences in Amsterdam and Seoul showed the effectiveness of AI in designing effective trajectories for public transportation and energy forecasting and the simulation of disasters with high accuracy, as described in [41,59,60]. As presented in the works of Pereira et al. [42], Evangelou et al. [3], and Moustaka et al. [11], increased urban resilience and sustainability result from the integration of AI into a systemic planning approach. Finally, AI enhances real-time decision making through dynamic dashboards and interfaces that enable cities to effectively and quickly adapt to the dynamic conditions [20,58,69]. However, maintaining these improvements requires further funding for digital platforms, the renewal of planner certifications, and international work on model compatibility [35,56,70].

2.3.3. Real-Time Data Integration in Smart Cities

Real-time data integration is one of the basic tenets of a smart city, since it enables the constant monitoring, processing, and management of urban systems via interconnected sensors and smart structures. Studies have shown that real-time feeds enhance traffic conditions, security, and the proper functioning of infrastructures [23,25,34,39]. For instance, Van Den Berghe [43] explains how real-time data integration in the digital twin guarantees an accurate simulation of the city and dynamic modeling to improve the response times and planning. Some cities, including Athens and Amsterdam, have integrated real-time data united with digital twins for managing energy usage and studying the urban environment [3,9,60]. Such applications illustrate the need for integrating real-time data to provide anticipatory urban management and precise service delivery [35,47,62].

Nevertheless, the real-time integration process remains a challenge due to the technological disparities, data isolation, and dissimilarities between municipal departments. Too many dispersed, integrated, and disparate sources of data are still an issue and have been reported in developed and developing countries where legacy systems continue to dominate [13,24,36,52]. This issue is made worse by the absence of methodologies in data governance that could work in unison to establish format harmonization, security regulations, and the compatibility of platforms [19,21,40]. These issues are challenging even in technologically developed cities and can impede the usefulness and timely arrival of the incoming data streams, preventing real-time decision making [6,7,68]. In addition, due to limited resources and technical capabilities, small- and medium-sized cities cannot install updates that support real-time integration [15,31,49].

Another issue that requires attention is the security and ethical use of real-time data during the process of collection and utilization. Scholars have argued that as cities become more integrated, deposit more data, and become more reliant on technology, their security is at risk from cyber-attacks [8,39,61,66]. Technological applications, such as real-time surveillance for instance, have attracted concerns around privacy invasion and data ownership, especially in the contexts of low transparency and citizen trust [4,10,23]. In addition, the processing of real-time inputs through AI results in algorithmic bias due to the training datasets being either incomplete or not fully representative of the wider population [6,57,62]. To mitigate these risks, scholars have called for increased regulation, whereby innovation is addressed alongside data protection and the public interest [22,55,67].

Nevertheless, these highlighted cases of real-time integration show that it is possible to achieve the set objectives. Pi et al. found examples from Singapore, Dubai, and Helsinki that suggested implementing IoT sensors and digital twins creates real-time dashboards for dynamic service adjustment, congestion easing, and energy efficiency [35,58,65]. These cities have been able to integrate transportation, waste, water, and emergency response data to control the central portals for decision making and the automation of repetitive tasks [31,54,56,71]. It hence calls for the integration of architectural elements, intersectoral cooperation, and a constant investment in technology enablers consistent with the sustainable development foreseen by Lucchi [69], Adibi et al. [64], and Herath et al. [20]. Hence, past research provides substantial evidence for the proposition that real-time data integration is not just a technical functionality but an implementational tool for smarter, safer, and more responsive cities [3,34,49].

2.3.4. System Maturity and Urban Resilience

The concept of system maturity in relation to digital twin and AI-based urban platforms can be defined as the extent to which a city’s stakeholders, regulations, and organizational setup is able to foster sustainable and data-informed urban management [35,43]. Past studies show that cities which have advanced systems with integrated long-term digital plans, formal practices, and constant technological advancements are able to achieve superior levels of success in terms of predictive analytics in urban planning and urban resilience [3,15,19,56]. For instance, Kljaić et al. [55], while studying the application of digital twins and environmental, social, and governance (ESG) frameworks, suggest that only when digital twins are positioned inside robust communal systems do they create enduring worth. Similarly, more established smart cities, for instance, Amsterdam, Singapore, and Vienna, show that sustained prolonged efforts to advance smart city initiatives and policies result in better resilience against disturbances, including epidemics and climatic disasters [10,20,25,34].

On the other hand, in cities where digital systems are still in development, the digital effects of predictive urban planning are considerably smaller. Some research illustrates that pilot-scale or one-off digital twin projects cannot generate significant planning impacts because of scalability, funding, and institutional sustainability problems [3,31,47,49].

A lack of system maturity is also reflected in low inter-departmental integration, unstructured data capture approaches, and limited real-time integration, which hinder the operational functionality of digital tools [6,39,42,64]. Similarly, literature surveys observe that research reveals that cities with low system maturity still do not have well-established measures for defining the metrics of performance for digital twins, which hampers the assessment and improvement of planning approaches [8,13,40]. This lack of feedback erodes the iterative learning, which is a core concept in designing resilient cities [9,35,62].

Notably, the role of governance models and stakeholder engagement also defines system maturity. Research evidence affirming participatory governance—where citizens, governing authorities, and technical specialists jointly develop planning strategies—shows that such cities are both digitally advanced and more antifragile [4,13,49,61]. As such, these cities integrate digital literacy for citizens, integrate ethical principles in the deployment of AI solutions, and provide open information about data usage, all of which bolster public confidence and system longevity [10,24,66].

Furthermore, an advanced maturity level is expected to leverage layered digital architectures comprising cloud, edge computing, and 5G networks to enable distributed processing and real-time decision making [22,39,58]. Namely, a lack of effective governance and technical capacity can lead to stagnation, even if the city possesses state-of-the-art tools [7,52,54].

The relationship between system maturity and urban resilience is mirrored in crisis response performance. Past research proves that the cities that are part of a digital environment are resilient and able to adapt more quickly during disasters, pandemics, and infrastructural breakdowns [3,6,46,60]. For instance, Herath et al. [20] established that the cities implementing modular traffic and emergency management digital twin platforms could flexibly redeploy resources each time there was a traffic jam or other disruption. In addition, post-event data reviews and performance audits are common in mature systems, which facilitate learning and system improvement [8,23,68]. Hence, based on quantitative research, the long-term efficiency of predictive urban planning is tied not only to extensive digital systems but to mature systems and governance surroundings [9,30,43,49].

2.3.5. Current Application of AI and Digital Twin Technology in Global Urban Planning

Artificial intelligence (AI) and digital twin technologies have found extensive applications in global urban planning, significantly enhancing the ability of cities to model, simulate, and optimize urban environments. Digital twins are AI-generated recreations of the physical cities, which combine the data feeds of multiple sources such as the IoT devices and sensors. These technologies have been deployed in a wide array of global smart city projects, such as those in Singapore, Amsterdam, and Barcelona, where they are applied to optimize traffic flow, energy consumption, waste management, and urban mobility [1,6,13]. Through real-time data processing and predictive analytics, digital twins enable cities to proactively address urban challenges such as congestion, pollution, and climate change [2,7].

In particular, AI has transformed the functioning of digital twins by enabling them to learn from past data and make autonomous decisions based on real-time inputs [4,19]. As a simple example, the future of urban transportation systems can be improved and made smarter by incorporating AI to manage more intelligent traffic systems, mitigate congestion, and generally enhance commuter’s lives. Moreover, AI-powered digital twins are also utilized for energy optimization in urban settings, helping cities to decrease their carbon footprint while maintaining the efficiency of energy use in buildings and industries [8,14,17]. The potential of these technologies to predict future trends based on past data makes them invaluable in urban planning, ensuring that cities can anticipate and adapt to changes in population density, energy needs, and environmental conditions [9,13]. However, as much as the globalization of these technologies comes with many benefits, there are some problems as well. The integration of AI and digital twins into the existing infrastructure often requires substantial investment, as well as overcoming barriers related to data privacy, security, and governance [5,20]. Moreover, it is still difficult to standardize AI algorithms across regions and cities. Despite these issues, the role of AI and digital twins in urban planning continues to grow, offering cities the tools to become more efficient, resilient, and sustainable in their approach to urban development [3,9,19].

2.3.6. Characteristics and Challenges of Smart City Construction in Europe

Europe has been leading the smart city revolution, and some cities in Europe have already adopted the use of AI and digital twin technologies in their city development plans. Features of these intelligent cities include high levels of environmental protection, energy savings, and the introduction of digital technologies to build better conditions in cities. Cities such as Amsterdam, Berlin, and Copenhagen are prime examples of European smart cities where AI is used to enhance public transport systems, manage energy consumption, and improve waste management, contributing to the region’s commitment to sustainable development [5,8,17].

Nevertheless, the construction of smart cities in Europe has various challenges. One major challenge is the high cost of the technology infrastructure, including the deployment of IoT devices, sensors, and AI algorithms, which can be a significant burden on cities with limited budgets [12,16]. Furthermore, the complexity of integrating the new technologies within the existing infrastructures, especially in older European cities, presents considerable challenges for urban planners and developers [14,19]. The lack of a standardized approach to data collection, sharing, and processing also hinders the seamless implementation of AI and digital twin technologies across the different cities in Europe [9,18]. The diverse regulatory environments in different European countries create additional complexity, as urban planning standards and data privacy regulations can vary significantly [13,18].

Additionally, social inclusivity and digital divides are also urgent issues to address as the smart cities are developing. Not all citizens have equal access to the benefits of AI and digital twin technologies, particularly those in rural or economically disadvantaged areas [10,19]. It is essential to address these inequities to guarantee a fair distribution of smart city growth advantages to all the residents. Additionally, concerns regarding data privacy and cybersecurity in smart city projects are heightened as the amount of data being collected continues to increase [6,13]. Despite these challenges, the integration of AI and digital twin technologies remains essential for the future of European cities, helping them to meet sustainability goals while improving urban management and governance [12,20].

2.3.7. The Case of Greece

The Greek context offers a unique perspective on the integration of AI and digital twin technologies into urban planning, due to the country’s distinctive urban and economic landscape. Greek cities, such as Athens and Thessaloniki, are grappling with rapid urbanization, which places significant pressure on the existing infrastructure, resulting in challenges in traffic management, waste disposal, and energy consumption [5,12]. However, Greece’s commitment to becoming a smart city leader in Europe is evident, with several initiatives focused on incorporating AI and digital twin technologies into urban planning [3,19]. These efforts are supported by the EU’s green and digital transformation goals, which have been integrated into Greece’s urban development strategies [7,18].

The complexity of the geography is one of the main features of the Greek experience in the development of the smart city. The presence of numerous islands and diverse urban settings creates unique challenges in implementing uniform smart city solutions across the country [5,13]. The need for flexible and scalable digital solutions is critical, as each urban area requires tailored interventions to address its specific needs, whether this involves managing tourist traffic in Athens or optimizing energy use in smaller rural communities [9,14]. Moreover, as in many European countries, Greece faces the challenge of balancing the adoption of new technologies with the preservation of its cultural heritage and traditional urban structures [18,19].

Despite the growing interest in smart cities in Greece, there remains a lack of comprehensive empirical studies on the application of AI-driven digital twins within the country’s urban planning practices. Existing studies have focused mainly on theoretical frameworks or small-scale pilot projects, with limited data on the actual effectiveness of these technologies within Greek cities [6,16]. This research aims to bridge this gap by examining how AI and digital twin technologies can be effectively integrated into Greece’s urban planning context, offering a roadmap for other Mediterranean and Southern European cities facing similar challenges [9,13]. By providing empirical evidence and actionable insights, this research will contribute to the broader understanding of how smart city technologies can enhance urban resilience and sustainability in Greece and beyond [7,18,19]. The implementation of digital twin technology in Greece provides a new opportunity to regulate the urban systems more effectively. This research seeks to highlight how these technologies can improve infrastructure management, resource allocation, and long-term sustainability in Greek cities, thus contributing to the global discourse on smart city development [9,19]. The findings will also provide valuable lessons for other regions, particularly those in Southern Europe, where urbanization rates are high and the demand for smart city solutions is growing [5,13]. This research may be used extensively in smart city projects in the Mediterranean area to examine the problems and opportunities surrounding AI and digital twin technologies in Greece.

2.4. Summary of the Literature and Research Gap

The literature shows that AI and digital twins can play a major role in enhancing the urban planning of smart cities. Research also reveals that using real-time data and AI analytics in digital twin systems are beneficial in managing cities with concerns about energy efficiency, traffic flow, etc. [25,34]. However, there is a lack of primary quantitative research that explores the relationships between digital twin features, including simulation fidelity and artificial intelligence-based analytics, and the planning outcomes for cities [30,46]. Despite several successful implementations, the literature regarding digital twins is rather limited and scattered and does not cover the general effects of deploying digital twins for urban planning [13,19]. More specifically, it is still unclear how and to what extent various attributes, such as the system maturity and real-time data integration, are complementary in supporting planning [31]. In addition, most of the studies contain only primary analyses without discussing how the digital twins evolve and are maintained in the long run [4,7]. This research directly addresses the gap in empirical evaluation by quantitatively examining how specific digital twin features—simulation fidelity, real-time data integration, AI analytics, and system maturity—affect predictive urban planning outcomes in Greece.

3. Methodology

3.1. Research Design

This study used a cross-sectional quantitative research approach to compare the various features of digital twins for predictive urban planning in the context of smart cities in Europe with an emphasis on Greece. The purpose of this study was to develop metrics to assess the level of simulation fidelity, real-time data integration, artificial intelligence capable of supporting decision making, system maturity, and the level of user interface interactivity that are directly correlated with the overall level of effective predictive urban planning. A structured questionnaire was used as the main data collection tool in this study. It was cross-sectional in its design, which allowed the researchers to capture the participants’ current beliefs concerning DTs and their relevance to urban planning.

3.2. Target Population, Sample Size Determination, and Sampling

This study involved urban planners, data scientists, and digital infrastructure managers in smart cities in Greece as the target population. Experts in these particular fields were selected because of their direct work in implementing, operating, and managing digital twin systems in urban planning. The participation of these hands-on individuals will ensure that the feasibility and difficulties of utilizing digital twin technology in urban environments are fully explored. With the orientation of such professionals, the study will document the insights of individuals with both technical expertise and practical knowledge in the urban planning and digital infrastructure industry.

The size of the targeted sample in this study was estimated using the Krejcie and Morgan [72] sample size estimation

Table 1. Table for determining sample size from a given or known population.

N	n	N	n	N	n
10	10	220	140	1200	291
15	14	230	144	1300	297
20	19	240	148	1400	302
25	24	250	152	1500	306
30	28	260	155	1600	310
35	32	270	159	1700	313
40	36	280	162	1800	317
45	40	290	165	1900	320
50	44	300	169	2000	322
55	48	320	175	2800	338
60	52	340	181	3000	341
65	56	360	186	3500	346
70	59	380	191	4000	351
75	63	400	196	4500	354
80	66	420	201	5000	357
85	70	440	205	6000	361
90	73	460	210	7000	364
95	76	480	214	8000	367
100	80	500	217	9000	368
110	86	550	226	10,000	370
120	92	600	234	15,000	375
130	97	650	242	20,000	377
140	103	700	248	30,000	379
150	108	750	254	40,000	380
160	113	800	260	50,000	381
170	118	850	265	75,000	382
180	123	900	269	1,000,000	384

that is commonly utilized in survey-based studies. The sample size was estimated as 301 respondents, based on an estimated population of 1350 urban professionals employed in the operation of digital twin systems within Greek smart cities. This sample size was evaluated as being adequate to provide sufficient statistical power and representativeness for drawing trustworthy conclusions from the sample data based on the survey. A bigger sample size would also mean that the data gathered would be strong enough to provide useful inferential analysis.

Purposive sampling was used to choose the respondents. This sampling method was particularly selected since it will involve the sampling of individuals with relevant knowledge and experience in the field of study. The purposive sampling will guarantee that the sample participants have direct experience in managing and implementing digital twin systems; thus, they can offer more insights into the difficulties, advantages, and practical uses of the systems. This sample selection method is frequently applied in studies that demand a level of expertise or special knowledge since this method guarantees that the information obtained is most relevant and resourceful.

Equation (1) shows the equation of Krejcie and Morgan.

$$\mathrm{n}={\displaystyle \frac{{\mathsf{\chi }}^2\mathsf{{\rm N}}\mathrm{P}\left(1-\mathrm{P}\right)}{{\mathrm{d}}^2\left(\mathrm{N}-1\right)+{\mathsf{\chi }}^2\mathrm{P}\left(1-\mathrm{P}\right)}}$$

(1)

where:

n = sample size

N = population size (1350)

χ² = Chi-square for specified confidence level at 1 degree of freedom (3.841)

d = desired margin of error (expressed as a portion = 0.05)

P = population portion (0.05)

3.3. Data Collection Instrument

The survey tool used in this research was a self-administered, 30-item scale using Likert-type questions. Based on these items, six factors were derived; five independent factors: simulation fidelity, integrated real-time data, artificial intelligence analytics, system maturity, and user interface interactivity; and one dependent factor: predictive urban planning effectiveness.

Before administering the questionnaire, the questionnaire underwent expert validation by three urban planning and data science specialists prior to data collection to ensure content validity. The items were reviewed in terms of clarity, relevance, and congruency with research objectives by the three specialists in urban planning and data science. They provided their responses that were considered to improve the questionnaire so that it could comprehensively capture the key information concerning digital twin technology in urban plans.

Most of the identified items had been developed through previous studies in digital twin systems and smart cities meaning the focus of the study was indeed well-anchored on previous studies. On the survey questionnaire, responses were based on a five point Likert scale where 1 = strongly disagree and 5 = strongly agree. This scale ensured that the perceptions of the respondents were measurable with a degree of uniformity, thus improving the credibility of the data obtained. Further reliability tests were also undertaken to determine the questionnaire’s consistency and strength before administering it to the respondents.

3.4. Operationalization of Variables

The key variables were measured or operationalized as follows:

Independent Variables (IVs):

• Simulation Fidelity: Defined as the extent to which simulation models are able to capture the realistic modes of urban conditions.
• Real-Time Data Integration: Assessed based on how the live data inputs affect the planning forecasts.
• AI-Based Analytics: Measured by the extent to which machine learning enhances the set-up of the scenario analyses.
• System Maturity: Assessed on whether the regular practice of digital twin systems over time increases its forecasting effectiveness.

Dependent Variable (DV):

Predictive Urban Planning Effectiveness: This is the extent to which the technological manipulation of digital twin systems affects the overall effectiveness of urban planning decisions or recommendations.

3.5. Data Analysis

Data analysis was conducted using descriptive and inferential statistical analyses. A brief description of the collected data outlined the age, gender, and experience of the respondents, as well as their position within the field of urban planning. To measure the degree of the positive or negative relationship between the independent and dependent variables, the Pearson coefficient was used. Also, a multiple linear regression analysis was carried out to determine the significance of the hypothesized relationships between the four independent variables (digital twin features) and the dependent variable (the effectiveness of predictive urban planning). Each statistical test used an alpha level of significance set at 0.05. The analysis of predictive variance in urban planning effectiveness involved the use of regression analysis whereby the independent variables were used to explain the result. The proposed multiple regression model for this research is Equation (2) [73,74].

$$\mathrm{Y}={\mathsf{\beta }}_0+{\mathsf{\beta }}_1{\mathrm{X}}_1+{\mathsf{\beta }}_2{\mathrm{X}}_2+{\mathsf{\beta }}_3{\mathrm{X}}_3+{\mathsf{\beta }}_4{\mathrm{X}}_4+\epsilon$$

(2)

where:

Y is the predictive urban planning effectiveness, which assesses the general efficiency of digital twin systems in enhancing urban planning accuracy and productivity.

X₁ = is the simulation fidelity (the first independent variable, IV), which reflects the capability of simulation models to accurately mimic the real environment.

X₂ = represents real-time data integration, which studies how the data collected from the sensors installed in cities affects the forecasting and planning of scenarios.

X₃ = represents AI-based analytics, which refers to the role of machine learning in enhancing predictive analysis and scenario planning.

X₄ = represents system maturity indicating the effectiveness of the digital twin system based on its age and the experience of the team and programmers involved in implementing the system.

β₀ = is the intercept of the model.

ε = represents the error term in the multiple regression model.

This model was used to test the amount of variance in the predictive urban planning effectiveness explained by the five independent variables (IVs). The assumptions of multiple regression such as normal assumption, linearity, multicollinearity, and homoscedasticity were tested and met as a condition of the analysis to be valid. The coefficients (β₁, β₂, β₃, and β₄) were estimated to provide an indication of the pathways of causality between the four IVs and the dependent variable. The hypotheses of the study were tested at the 5% (0.05) level of significance throughout the study.

3.6. Ethical Considerations

The rights of the participants were respected, and the following guidelines of ethical issues were considered: All the respondents provided informed consent and were informed that they were free to participate in the study without influence and could opt out of the study at any time. The data collected in this study were strictly used for academic purposes and stored in a secure manner to avoid identification of the individuals involved.

3.7. Limitations of the Study

The limitations of this study entail matters concerning sample representativeness and questionnaire design. The use of 301 respondents as the sample size was informed by the recommendations of Krejcie and Morgan, although the sample was limited to practitioners in Greek smart cities, which may influence the overall results as urban planners, data scientists, and managers in Europe could be more varied in their implementation strategies. Thus, one should not assume that the results obtained from Greece can be applied to other European smart cities, as they differ in their levels of urbanization, the penetration of a digital twin, and policies. Also, some sources for the scales used were not reported in their entirety; only the reliability and validity analyses of this questionnaire were mentioned. A lesser discussion around the importance of the reliability and validity of the scales used would make the study more transparent. These details could aid in providing a better insight into how the questionnaire might be applicable in various urban situations and enhance the authenticity of the results.

4. Results

This section presents the results obtained after analyzing the data collected from the selected respondents.

4.1. Demographic Characteristics

Table 2. Demographic characteristics of respondents.

Characteristic	Frequency	Percentage (%)
Gender
Male	189	62.8
Female	112	37.2
Age bracket (years)
18–25 years	44	14.6
26–35 years	85	28.2
36–45 years	98	32.6
46–55 years	55	18.3
56+ years	19	6.3
Professional Role
Urban Planner	138	45.8
Data Scientist	78	25.9
Digital Infrastructure Manager	85	28.3
Years of Experience
0–5 years	49	16.3
6–10 years	91	30.2
11–15 years	87	28.9
16+ years	74	24.6

Source: Survey (2025).

presents the demographic profile of the respondents. There were 301 participants in the survey; their details are captured in the following table based on age, gender, professional role, and years of experience in using digital twin technology in urban planning.

The data results in Table 2 showed that 301 participants responded, of which 60.8% had worked for 26–45 years, which indicates that the workforce is young and has exploitative experience. This age group implies that the majority of the respondents are likely to be mid-career employees with considerable working experience but who continue to learn and update themselves, given the dynamic nature of digital twin technology. Regarding the gender distribution, 62.8% of the respondents were male while 37.2% were female, which is quite common in the fields of technology and urban planning.

At present, the largest group of respondents occupies professional positions related to planning (45.8%), followed by digital infrastructures (28.3%), and data scientists (25.9%). From this distribution, it indicates that the survey received input from a good mix of professionals who actively engage in the use of digital twin technologies in urban planning. A majority of the respondents, 59.1%, had between 6 and 15 years of experience as city planners; this means that the sample had high and adequate levels of experience in the urban planning and digital twin systems which can be used to inform the viability of using advanced technologies in predictive city planning.

4.2. Simulation Fidelity and Predictive Urban Planning Effectiveness

This study assessed the impact of high-fidelity and low-fidelity simulations on predictive planning for urban areas. These are depicted in Figure 1, with participants required to indicate the extent to which they agree or disagree with statements about the accuracy and efficiency of the simulation models used in urban planning.

Figure 1 shows that the respondents have a positive perspective concerning the utilization of high-fidelity simulation models within the sphere of urban planning. This study showed that 95% of the respondents were either rather likely or highly likely to participate in future surveys, demonstrating future participation willingness and a reinforcement of the survey results. Similarly, 72% of the respondents acknowledged that the simulation models are accurate in their general forecasting of urban development needs and probabilities, thus stressing the reliability of the models.

Regarding the extent to which simulation fidelity is considered as having a direct influence on the efficiency of urban planning, 18 individuals agreed or strongly agreed with this notion, which shows the significant role that model realism plays in the overall performance of planning activities. In addition, 75.7% of the respondents agreed that there is a direct relationship between the quality of the simulation models developed and the level of accuracy of the plans made, further reiterating the belief that great simulation models yield better planning decisions. Collectively, the findings imply that the respondents believe that high-fidelity simulation is critical in increasing predictive urban planning. This underlines the significance of providing urban development strategies that employ accurate and reliable simulations for the planning and design of better cities, with the majority of respondents expressing their trust in the use of digital twins for urban planning.

4.3. Real-Time Data Integration and Predictive Urban Planning Effectiveness

The second research question in the study sought to assess the impact of real-time data integration on the effectiveness of predictive urban planning.

Table 3. Results on real-time data integration and predictive urban planning effectiveness.

Statement	%	SD	D	N	A	SA
Real-time data integration significantly improves planning forecasts.	%	4.1	5.3	10.5	42.1	38.0
The integration of live data makes urban planning more accurate.	%	3.7	6.5	14.1	41.6	34.1
Real-time data inputs contribute to better decision-making in urban planning.	%	2.9	7.4	12.7	43.6	33.4
Integration of real-time data makes it easier to predict future urban trends.	%	4.3	8.2	15.6	39.4	32.5
The use of real-time data increases the efficiency of urban planning.	%	3.1	6.2	13.3	41.0	36.4
Real-time data plays a critical role in urban forecasting and planning.	%	3.9	7.0	12.2	41.7	35.2
The integration of live data has improved predictive urban planning.	%	2.8	5.6	14.0	42.5	35.1

Key: SD = strongly disagree, D = disagree, N = neutral, A = agree, SA = strongly agree. Source: Authors’ elaboration.

below shows the responses from the participants regarding their perception of the importance of live data in improving the forecast of urban dynamics.

Table 3 above shows that real-time data integration is crucial to enhance the efficiency of predictive urban planning. The necessity and importance of real-time data integration can be highlighted in particular with reference to the respondents’ perception of assistant real-time data integration, where 80.1% of the respondents agreed or strongly agreed with the statement that it makes a significant contribution to the planning forecasts. Likewise, 75.7% of the participants stated that adopting a live data integration method enhances the credibility of urban planning information, indicating its usefulness. In addition, 77% of the respondents pointed to the idea that the real-time data assists in decision making; this supports the notion that real-time data helps the planners to make more sound decisions.

This is apparent based on the 71.9% of respondents who opined that incorporating real-time data helps in envisaging the future trends in urbanization. Moreover, 77.4% of the participants opined that the employment of real-time data enhances the efficiency of urban planning, thus enhancing the emergence. Concerning real-time data in the context of urban forecasts and planning, 76.9% of the respondents agreed with this statement. In essence, 77.6% of the participants noted that live data integration has had a positive effect on predictive urban planning. Targeted surveys reveal that real-time data integration is received as a helpful tool with an acknowledged contribution to the accuracy, efficiency, and efficacy of urban planning, which gives confidence to smart city adoption.

4.4. AI-Based Analytics and Predictive Urban Planning Effectiveness

The third research question was to establish the impact of AI-based analytics on the performance of predictive urban planning. These findings are presented in Figure 2.

According to Figure 2, the perceptions towards the use of AI-based analytics for improving the predictability of urban planning are generally positive. The highest percentage of yes answers was received concerning the items that stated that AI analytics helped to enhance the projection precision of urban planning and refine the decision-making procedures by 75% to 80%. Also, 78.7% of the participants shared the opinion that with AI-based analytics, the predictive estimations of urban planning are more accurate, whereas 79.3% also approved of the concept of applying ML to digital twins to optimize urban planning. In addition, 78.3% said that AI analytics are crucial to identifying better future urban environments, and 77.5% said that using AI in digital twins helped to enhance the quality of urban planning decisions.

Indeed, such responses indicate that AI is viewed as useful in enhancing the credibility of urban planning through analysis. Specifically, 75.8% of the respondents considered that the use of AI-based predictive models has increased planning precision while 78.8% also asserted that AI analytics has increased the speed of urban planning and its responsiveness. This supports the argument that AI averts the time and effort needed in the planning process. Similarly, the research results indicate that AI-based analytics are considered ‘must-haves’ for enhancing the efficiency, efficacy, and velocity of the urban planning decision-making process and stress the relevance of AI solutions in contemporary smart city planning.

4.5. System Maturity and Predictive Urban Planning Effectiveness

This study also assessed the link between system maturity and the effectiveness of predictive urban planning. The results are presented in

Table 4. Results on system maturity and predictive urban planning effectiveness.

Statement	%	SD	D	N	A	SA
The maturity of digital twin systems has improved urban planning outcomes.	%	5.1	7.2	12.9	39.3	35.5
Long-term use of digital twin systems increases predictive planning accuracy.	%	4.0	8.3	13.5	41.6	32.6
The experience of using digital twins over time enhances their effectiveness.	%	4.3	7.5	13.1	40.1	35.0
System maturity has a positive effect on urban planning forecasts.	%	5.2	6.4	14.0	40.9	33.5
The longer the use of digital twins, the better the planning outcomes.	%	4.6	8.1	13.3	39.6	34.4
Mature digital twin systems provide more accurate urban predictions.	%	3.9	7.8	12.2	42.1	34.0
System maturity improves the overall effectiveness of urban planning.	%	4.5	7.0	12.4	40.0	36.1

Key: SD = strongly disagree, D = disagree, N = neutral, A = agree, SA = strongly agree. Source: Authors’ elaboration.

.

In Table 4, it is observed that 57% of the respondents agreed or strongly agreed with the statement that the maturity of digital twin systems enhances urban planning results. The current response indicates a certain understanding that the more advanced the digital twin systems are, the more useful they are in urban planning, providing more insights and predictions. Considering that more than half of the respondents shared this view indicates the perceived worth of a mature system on the process of urban decision-making. It is an indication of the general recognition that these systems are improving and that they will provide more dependable data, resulting in a subsequent increase in the quality and efficacy of urban planning decisions.

Moreover, 74.8% of the respondents reported that the maturity of the digital twin systems impacts positively on urban planning, with 35.5% completely agreeing. This affirms the unanimity that more established systems translate into better implementation of the urban planning procedures, and it implies that exposure to such technologies aids urban planners in refining their planning methods and forecasts. It may be concluded that the more the digital twin systems are utilized, the better they are at simulating the actual urban environments in a metropolis, and the more accurate the projections become on the development of the cities and their management. The active use of digital twin systems over a long period, based on the answers of 74.2% of the respondents, is also considered a method for increasing the accuracy of planning. These systems are able to develop knowledge and retain information over time, and as they become more experienced they can make more accurate predictions. The fact that 74.3% of the respondents feel that mature systems offer more predictability further substantiates the premise that systems evolve to become more accurate at predicting what is likely to happen in the city, along with population growth, traffic flows, and thus resource consumption. In addition, 74.4% of the interviewees contended that system maturity makes increasing predictions more precise, and 74.0% observed that the prolonged application of digital twins made the outcomes of planning improve. This gives all the more reason to bolster the assertion that the ongoing implementation and enhancement of digital twin systems are no longer just to optimize the quality of urban planning but also to elevate the precision of the urban forecasts. As time goes by, the system can be fed with the previous data to improve its algorithms and provide more accurate results.

The combination of these outcomes demonstrates that the maturity of the digital twin system is an essential factor in enhancing the efficacy of predictive urban planning. Advanced systems can process complicated datasets and combine multiple variables influencing urban spaces, giving decision-makers a more detailed and accurate understanding. The persistent utilization of digital twins over the years enables them to improve and change over time to thus better manage dynamic urban issues.

4.6. Diagnostic Tests

4.6.1. Normality Test

Multiple regression assumes the normality of the residuals, such that the residuals become normally distributed. The Shapiro–Wilk test was applied in this research to test the normal distribution hypothesis. The p-value obtained was 0.145 (

Table 5. Normality test (Shapiro–Wilk).

Test	Statistic	p-Value	Conclusion
Shapiro–Wilk Test	0.987	0.145	Residuals are normally distributed (p > 0.05)

Source: Primary data (2025).

), which is greater than the significance level of 0.05 (p > 0.05). We thus do not reject the null hypothesis that the residuals are normally distributed. This signifies that the residuals comply with the normality condition, which is fundamental for the validity of the regression analysis.

4.6.2. Multicollinearity Test

To check for multicollinearity, the Variance Inflation Factor (VIF) was calculated for each independent variable (

Table 6. VIF multicollinearity.

Predictor Variable	VIF	Conclusion
Simulation Fidelity	1.83	No multicollinearity
Real-Time Data Integration	2.21	No multicollinearity
AI-Based Analytics	1.67	No multicollinearity
System Maturity	4.21	No multicollinearity

Source: Primary data (2025).

). All the predictors of VIF were very low compared to the five threshold, meaning that no significant multicollinearity exists among the independent variables. This will affirm that the predictor variables do not overlap in their contributions to the model.

4.6.3. Test of Homoscedasticity

Homoscedasticity determines whether the variance in the residuals is homogeneous at every level of the independent variables. The Breusch–Pagan test was run to test heteroscedasticity (

Table 7. Breusch-Pagan Test of Homoscedasticity.

Test for Heteroscedasticity by Breusch and Pagan	Ho: Constant Variance
chi2 (1)	0.4732
Prob > chi2	0.4732

Source: Primary data (2025).

). The p-value was 0.4732, which is greater than 0.05 (p > 0.05). Thus, we cannot disapprove of the null hypothesis of similarity in the error variance, meaning that the residuals are homoscedastic and the variance remains constant among the observations.

4.7. Regression Test

4.7.1. Fitness of Model

Table 8. Model fitness.

Model	R	R Square	Adjusted R Square	Std. Error of the Estimate
	0.829 *	0.688	0.674	0.276

* Predictors: (Constant), simulation fidelity, real-time data integration, AI-based analytics, system maturity.

displays an overview of the fitness of the model used in the regression analysis and assesses the extent of predictability of the dependent variable (predictive urban planning effectiveness) using the five independent variables: simulation fidelity, real-time data integration, AI-based analytics, and system maturity. This shows that it is possible to develop a well-fitted model that will capture the relevance of the digital twin technology features on the related city planning experience.

Furthermore, the obtained value of R = 0.829 suggests that the increased values of the independent variables are associated with the increased value of the dependent variable, thus showing that these predictors are impacting the predictive urban planning effectiveness. The obtained value of R² = 0.688 indicates that 68.8% of the total variability in the measurement of the effectiveness of the use of predictive urban planning can be explained by the five independent variables. Therefore, this result validates the approach used in developing the model and supports our choice of the predictors that have the potential to capture the nature of variation in urban planning.

4.7.2. Regression of Coefficients

Table 9. Regression results.

Predictive Variables	Standardized Coefficients (Beta)	p-Value
Simulation Fidelity	0.248	0.003
Real-Time Data Integration	0.315	0.001
AI-Based Analytics	0.382	0.000
System Maturity	0.289	0.004

Dependent Variable: Predictive Urban Planning Effectiveness.

shows the regression coefficients for each independent variable in the study. Here are the findings of the regression analysis test:

Practically, these results imply that enhancing the AI analytics capabilities yields the greatest incremental improvement in predictive urban planning effectiveness, followed by real-time data integration, simulation fidelity, and system maturity. These findings underscore the importance of leveraging AI and real-time data for more accurate and responsive urban planning processes.

Out of all the regression factors, AI-based analytics (0.382, p = 0.000) had the biggest impact on predictive urban planning in this particular regression analysis. This remarkable discovery suggests that AI analytics play a crucial role in improving the accuracy and the swiftness of urban planning decision making. The large value of the coefficient (0.382) indicates the power of AI to analyze large data, uncover concealed patterns, and create predictive models that enable real-time decision making. With the increasing complexity of city spaces, the ability of AI to integrate and analyze large volumes of data means that planners can make more accurate predictions, thus enhancing the effectiveness of the planning process. The statistical significance (p = 0.000) also supports the centrality of AI to predictive urban planning, indicating that it makes a deep impression in this respect.

The integration of data in real time (Β = 0.315, p = 0.001) is also an important factor in urban planning efficiency. Real-time data enables urban planners to change their approaches to suit the most up-to-date available information. This direct impact on the planning models emphasizes the significance of the availability of current data in order to make informed decisions. The significance (p = 0.001) suggests that real-time data plays a critical role in enhancing the accuracy and responsiveness of urban planning systems, indicating that we can accept hypothesis 2 (H2).

The coefficients of simulation fidelity include beta = 0.248 and p = 0.003. The obtained coefficients indicate that there is a strong positive correlation between the simulation fidelity and the effectiveness of predictive urban planning. Since the calculated p-value is less than the predetermined level of 0.05, we reject hypothesis 0 (H0) and accept hypothesis 1 (H1), which confirms that there is a positive relationship between the simulation fidelity and effective urban planning.

The beta result obtained for the system maturity is 0.289 and the p-value is 0.004. This means that a higher level of digital twin maturity is associated with enhanced effectiveness in urban plans, implying the acceptance of hypothesis 4 (H4). All the hypotheses are supported at a significant level by the regression test, displaying the contributions of all elements of digital twin technology in overcoming the deficits of predictive urban planning in smart cities.

5. Discussion

The purpose of this research was to investigate the effect of AI-based digital twin technology on predictive urban planning in Europe with an emphasis on selected smart cities in Greece. This study provides valuable information on how these technologies apply to improve the effectiveness of planning in urban settings by evaluating indices including the simulation fidelity, the integration of real-time data processing and AI-informed analytics, and the system maturity [31,48,67].

Simulation realism is a critical determinant of the integrity of simulation-based models in urban planning. They accurately mimic real-life conditions, which makes it possible to predict future tendencies in urban processes like traffic movement, energy usage, and climate [9,31]. Regarding the specific benefits, this study also shows part B, where the majority of the respondents mentioned high-fidelity simulation as contributing to accurate urban planning. This is particularly complemented by the fact that the digital twin-based simulations of urban conditions as seen by the users are sensitive to the reliability of the underlying sources of information, and as the simulation fidelity increases, so does the effectiveness of the planning interventions for city management [7,8].

However, reaching high levels of simulation fidelity is not without its challenges. This is in line with the assertions of Sheraz et al. [39] and Zong and Guan [30] that while higher fidelity models are necessary due to the increasing complexity of urban systems, their simulation consumes a large amount of processing power and requires sophisticated algorithms. In line with these concerns, the respondents stated that as the systems become increasingly dynamic and integrated into the urban environments, the necessary simulation fidelity results in increased computational cost. The ability to achieve a balance between a precise simulation and a reasonable calculation time remains a crucial point in the effective application of digital twin technology in urban planning.

The use of real-time data is another influential factor that improves the flexibility of the models used in urban planning. Real-time data helps in decision making by allowing the planner to revise their plans in response to the current situation in the city, like traffic patterns, energy consumption, and weather, among others [13,25]. This research also supports previous theories indicating that the integration of real-time data in favor of urban planning is perceived to enhance accuracy and efficiency. The respondents noted that live data inputs not only make the forecasts more accurate but also offer the most current trends in urban areas [7,42].

Nevertheless, the approach of real-time integration also has certain drawbacks, the most critical of which are the handling of large amounts of data and quality and compatibility issues [19,23]. Maintaining the data integrity from several sources is also problematic due to the existing legacy structures and data interfaces that slow down the processing and utilization of real-time information, as this study shows. Similarly, data quality becomes a challenge whereby inadequate data quality impacts the performance of the planning models [4,43].

Machine learning is critical in improving the prognostication of digital twin models. AI also helps in giant dataset analysis for pattern exploration and is much more helpful for urban planning [30,31]. This research supports the use of AI in urban planning, since more than half of the participants agreed that it can assist in enhancing the accuracy of the forecast and decision making. According to Herath et al. [20] and Rahman et al. [34], the use of AI in analytics enhances the effectiveness of planning, making it more flexible to changes in the city environment. Nevertheless, as numerous authors assert [13,23], one major issue with AI analytics is the lack of transparency.

Claims that AI systems, particularly deep learning, are black boxes stems from the fact that it is not very clear how AI arrives at certain decisions. This can reduce confidence in the AI-driven advice as planners may not be able to comprehend the algorithms behind the forecasts, which slows the integration of AI technology into the urban policies [55,62]. To mitigate this problem, further advancements must be made in improving the explicability of the AI solutions and guaranteeing that the outcomes are clear to the planners and policymakers.

The safety and stability of the digital twin systems are critical in recognizing their effectiveness and ability to become an integral part of urban planning methodology [19,23]. Closely related to system maturity, the respondents noted that when organizations work with more mature digital twin systems, they will see better results in planning for cities. This was found to be the case, as many systems were found to become more reliable at predicting urban conditions as the system was used over time, benefiting from the experience of the planners. The maturity of the systems is a critical component of the credibility and success of digital twin technologies. The wider experience becomes the digital twin system, and the more likely the system is to predict the conditions in the city and offer valuable solutions to city planning. Our study found that mature systems yielded better results, as planners became more familiar with their capabilities over time [3,9]. Nonetheless, the system maturity involves considerable investment in infrastructure, training, as well as long-term upkeep. Many cities, especially those in developing regions, face challenges in maintaining and scaling their digital twin systems, which can affect the sustainability of these technologies [49,55]. Interoperability issues, particularly when integrating digital twins into existing urban planning processes, were also highlighted as key challenges [6,31]. This discovery highlights the need to create long-term programs to maintain, improve, and support digital twin systems so that cities can generate ongoing insights into digital twins [75].

Still, the establishment of system maturity is a long-term and intricate process that calls for the commitment of substantial amounts of capital and personnel. It has also been revealed that many cities, especially the developing ones, experience difficulties in the maintenance, let alone scaling, of their digital twin systems, which can cause a problem with their sustainability [49,55]. Indeed, some studies have reported on the effects of system immaturity, primarily in regard to the integration of DTs into the urban planning processes, where interoperability problems emerge as a key challenge [6,31]. Therefore, one of the implications of this study is the importance of developing long-term initiatives to support, sustain, and upgrade the digital twin systems for cities.

Although this study provides clear evidence regarding the possible benefits of digital twin technologies, a number of challenges were observed. A major limitation to the effective utilization of such systems are the computational loads and data flow issues. The cities will face certain data quality, astronomical, and integration concerns as they upgrade to more advanced digital twins. The findings of this study highlight the importance of developing robust data management systems and establishing partnerships between data owners to facilitate seamless data integration [31,43]. In addition to this, the interpretation of the AI models and AI transparency play a crucial role in building confidence in AI applications. Addressing these concerns will be essential for incorporating AI-driven tools into urban governance and policymaking [23,62]. On the policy side, urban planners and policymakers need to also reflect on the ethical considerations of applying AI in urban planning, specifically the issues of data privacy and governance. The usage of AI-driven technologies creates some significant questions concerning who owns the data and how it is utilized. Making the process of AI decision making transparent will be central to building public trust and subsequently guarantee that the technologies will be used in a responsible and accountable way.

The findings of this study have several important implications for urban planners and policymakers: First, the simulation of fidelity, real-time integration of data, use of artificial intelligence analytical tools, and system maturity, all highly correlate with the success of urban planning outcomes, meaning that cities must adopt high-quality digital twin systems that comprise all these aspects. Such an approach suggests that the flow of information that accumulates in big cities should be matched with the predictive and real-time capabilities of AI technologies, so that policymakers are better able to support the solutions for urban issues [3,10].

Likewise, the study reveals that, for digital twin systems, issues like data integration and computational requirements should also be considered important. Cities must address the challenge of data fragmentation by embracing sound data management systems and building partnerships between the various data owners to incorporate real-time data into planning [31,43]. However, decision-making during AI-based analytics should be made more transparent to increase the trust in AI and incorporate AI-based tools into urban governance [23,62]. However, there are several limitations of this study that need to be noted. First, due to the use of cross-sectional data it would be difficult to draw causal conclusions regarding the association between the parameters of DT and the results of urban planning.

Longitudinal analysis would offer better evidence on the sustained effects of digital twin systems on urban planning. Second, since the study is based on the Greek context, the results may not pertain to all European cities, as there could be differences in the requirements of the urban environment, digital twin deployment, and region-specific policies. Further studies should be conducted, not only with a focus on multiple countries and cites but on the variation within countries in order to apply the results on a broader scale [7,19].

While the benefits of digital twin technology in predictive urban planning are evident, cities face significant challenges regarding their implementation. These challenges include infrastructural incompatibility, particularly in older cities with outdated systems, high computational demands when running complex simulations, and limitations in real-time data quality and integration. Overcoming these challenges will require substantial investment in infrastructure, the development of efficient data management strategies, and ensuring interoperability between different urban systems.

6. Conclusions

This study examined the role of the digital twin based on artificial intelligence to support and improve predictive urban planning in the context of smart cities, with an emphasis on Greek cities. By analyzing the key components of the digital twin system, such as the simulation centrality, integrated data feeds, AI analytics, and system sophistication, this study offers strategic insights into the potential of these technologies for advancing urban planning practices. The study has demonstrated that increasing the simulation of fidelity, the use of real-time data, and the incorporation of artificial intelligence improve the planning process in cities.

Firstly, the use of simulation fidelity enhances the accuracy of predicting the evolution of urban systems and structures. The use of real-time information enhances the ability of urban planners to make decisions that are informed by real-time information. Furthermore, AI-based analytics are useful in enhancing the rate and accuracy of prediction and decision making, which are essential in effective urban management. Another notable finding was the need to focus on the system maturity, as the usage and incremental improvement of the DT systems over a long-term period may translate into better planning outcomes and the ability to address the dynamic urban environment. This study also reveals some of the limitations of implementing digital twin systems in urban planning, in particular the requirement for significant computational power to achieve high levels of system realism, and the obstacles to handling time-sensitive information.

However, the lack of transparency was recognized as a major hurdle in developing trust and accepting the models. These issues show that there is a need for continuous investment in technology, staff development, and data management strategies to enhance the efficiency of the digital twins in cities. Considering the effectiveness of various features of the digital twin on urban planning, it is crucial that cities incorporate the following: adoption of advanced AI technologies, real-time data acquisitions, and system maturity. It is also crucial to focus on the improvement of the data-sharing standards and the regulation of AI transparency to make its usage more widespread. Altogether, this research extends the knowledge regarding smart city solutions and can be useful for those city managers and planners interested in adopting digital twin technologies to enhance the various aspects of urban environment functioning.