Machine Learning Applied to the Analysis of Prolonged COVID Symptoms: An Analytical Review

Abstract

The COVID-19 pandemic continues to constitute a public health emergency of international importance, although the state of emergency declaration has indeed been terminated worldwide, many people continue to be infected and present different symptoms associated with the illness. Undoubtedly, solutions based on divergent technologies such as machine learning have made great contributions to the understanding, identification, and treatment of the disease. Due to the sudden appearance of this virus, many works have been carried out by the scientific community to support the detection and treatment processes, which has generated numerous publications, making it difficult to identify the status of current research and future contributions that can continue to be generated around this problem that is still valid among us. To address this problem, this article shows the result of a scientometric analysis, which allows the identification of the various contributions that have been generated from the line of automatic learning for the monitoring and treatment of symptoms associated with this pathology. The methodology for the development of this analysis was carried out through the implementation of two phases: in the first phase, a scientometric analysis was carried out, where the countries, authors, and magazines with the greatest production associated with this subject can be identified, later in the second phase, the contributions based on the use of the Tree of Knowledge metaphor are identified. The main concepts identified in this review are related to symptoms, implemented algorithms, and the impact of applications. These results provide relevant information for researchers in the field in the search for new solutions or the application of existing ones for the treatment of still-existing symptoms of COVID-19.

1. Introduction

COVID-19 continues to impact the global community through persistent symptoms, often referred to as long-COVID or chronic COVID-19, which extend beyond the initial infection, and the degree to which artificial intelligence has provided opportunities to advance understanding, care, and outcomes remains unclear. The symptoms associated with COVID-19 are still prevalent in the global community; therefore, different mechanisms or strategies have been covered to be able to carry out continuous monitoring that guarantees health around the world [1]. Among the strategies that have been developed to monitor symptoms are those that are based on the use of artificial intelligence and machine learning to be able to explore different psychological and clinical symptoms that directly affect the quality of life of patients [2]. The effective management and intervention of the symptoms associated with COVID-19 generates a highly positive impact on public health and can prevent the appearance of outbreaks and improve the work and personal performance of individuals [3]. To achieve patient follow-up, it is imperative to have a high commitment from both the government and service provider institutions in achieving channeling of information and speed of patient care.

Based on consultations made for specialized WOS and Scopus databases, it has been possible to identify that reviews of the literature associated with the implementation of different teams or ways of identifying COVID-19 have been carried out, some of these reviews address technology used in the context of the emergency to identify risk conditions for patients [4]. Other reviews focused on the use of machine learning based on predictive methods for identification, which could be the critical factor in generating alarms associated with the complications of the infection [5]. Analyses were also carried out on the usability of telecare and remote tools for patient care during and after the global emergency process [6].

To ensure the relevance and timeliness of the reviewed research, precise inclusion and exclusion criteria are established for the systematic literature search on COVID-19. The inclusion criteria encompass documents published within the temporal range from 2019 to 2024, ensuring that the information is relevant and contemporary to the evolving pandemic and its associated challenges. Several types of documents are included, such as journal articles, books, book chapters, and conference proceedings, which allow for a broad coverage of research conducted across various disciplines and academic formats. Additionally, a specific set of keywords has been used for the search: “COVID-19”, “COVID”, “SARS-CoV-2”, “coronavirus”, and “2019-ncov”. These keywords are designed to capture the widest range of research related to the virus, encompassing both its scientific designation and popular name variants. As for the exclusion criteria, documents outside the specified date range will be discarded, as the information may not reflect the most recent discoveries or updated health policies. Also, documents that do not contain the established keywords in their text are excluded, thereby ensuring that the studies reviewed are directly related to COVID-19 and its impacts. This inclusion and exclusion strategy helps focus the review on studies that provide relevant knowledge and data about the management, treatment, and evolution of COVID-19, ensuring that the findings are applicable and useful for the global scientific and medical community. That is why this review aims to identify how different applications using artificial intelligence have provided effective solutions for the treatment of the various symptoms associated with COVID-19, which persist in humanity. The researchers therefore answer the following research question: How have disruptive technologies based on artificial intelligence supported the detection and treatment of the various persistent symptoms of COVID-19?

For this, search equations were applied in the WoS (Web of Science) and Scopus databases where all the records and extracted references were analyzed. This review of references from these two databases was unified in a single data set, which was processed using the Tree of Science (ToS) methodology and was analyzed using the package programmed in R Tosr (https://cran.r-project.org/web/packages/tosr/index.html, consulted on 20 December 2023). After this process, a scientometric analysis was carried out that helped to understand the trends of these contributions at a national level. This article is organized as follows: first, the methodology section provides details about the articles and the data analysis resulting from the search process. Then, in the results section, the detail of the second scientometric analysis of the documentary analysis is presented, which is provided through the Tree of Science methodology: roots, trunks, and branches, thus showing the most relevant aspects and exposing the conclusions of the research performed.

2. Materials and Methods

To develop this scientometric analysis focused on solutions based on Machine Learning or AI to identify and intervene in persistent symptoms of COVID-19, a selection and consultation process of articles published in WoS and Scopus databases was carried out [7]. These two databases were used because they agglomerate the highest concentration of scientific publications in the world, the unification process of the results issued by these databases for this analysis has been carried out using the Bibliometrix tool [8] and the Tosr package, managing to make a list of both records and references made between authors. This methodology has allowed us to have a holistic vision of the current state of research on the use of machine learning in the treatment processes of prolonged symptoms of COVID-19, through the analysis of appointments and collaboration through network analysis. The combination of both data sets also allowed us to have a more complete vision of the current contributions in this line of research [9,10], see

Table 1. Search results for parameters in the databases.

Parameter	WoS	Scopus
Range	2019–2024
Date	December 2024
Type of Document	Paper, book, chapter, conference proceedings
Words	COVID-19 OR COVID OR SARS-CoV-2 OR coronavirus OR 2019-ncov
Results	440	694
Total (WoS + Scopus)	1134

.

The SAP algorithm process, used to structure a group of academic documents into a directed graph known as the “Tree of Science”, begins with the creation of an initial graph G, derived from a collection of documents V extracted from the WoS and Scopus databases. In this graph, each directed edge represents a citation relationship between one article and another. This initial graph is then refined through multiple stages including selecting the largest connected component, removing loops and duplicate edges, and discarding vertices based on specifications for in-degree and out-degree, resulting in a new graph G′. In the classification phase, the algorithm identifies and categorizes vertices into three types: roots, leaves, and trunks, based on citation relationships and other criteria such as in-degree and out-degree. Roots are defined by having an out-degree of zero, highlighting their importance by the number of citations they receive. Leaves, on the other hand, have no incoming citations and their value is measured by the influence they have on the roots through direct paths. Trunks are identified as the vertices that serve as key connections between the roots and leaves, playing a crucial role in the dissemination of knowledge within the network.

To conduct this analysis, various functions from the Igraph library are used that facilitate the manipulation and simplification of the graph according to specific parameters. These functions allow for everything from grouping clusters to selecting and simplifying vertices, making the process more efficient. The result is the Tree of Science, a subgraph that incorporates root, leaf, and trunk vertices along with their corresponding links, offering a hierarchical and structured view of the literature, from foundational works to emerging contributions and pivotal studies that link these extremes. This model, proposed by Robledo [11], is an essential tool for visualizing the evolution of a field of study through its citation topology, as illustrated in their document.

This analysis was carried out in two phases, described below. The first was focused on the general description of the research, including relevant information associated with the production by country, magazine, and authors. In the second phase, a theoretical deepening associated with the current state of the research was carried out and the most relevant findings are presented based on the trends of this research topic. Specifically, in the second phase, the metaphor was used. Figure 1 shows the PRISMA diagram that describes how the article selection process was carried out in support of PRISMA 2020 guidelines, making the pre-processing through the implementation of the R code developed by the Core of Science. (https://github.com/coreofscience, accessed on 20 January 2023) that allowed a detailed analysis of the data resulting from the review.

2.1. Scientometric Analysis

The scientometric analysis, also called scientometrics, focuses on carrying out a quantitative analysis of the scientific data, among which the behavior of the annual production in a line of research, the collaborations between different authors, and their academic networks to work together [12]. There are different ways of being able to represent the results of scientometric analysis, among which the following stand out: Analysis by citation [13], Collaboration Networks [14], and Intellectual Structure [15]. The methodology used to describe the results of this review is based on an analysis from the general to the, which will start from the analysis of production and trends by ranges of years, to later identify the countries and journals that are publishing the most in the world—topic and finally to know how the collaboration process between the different authors has been. The analysis is based on a condensed view of the results issued by the WoS and Scopus databases and in this review, there will be a descriptive analysis of the findings found.

2.2. Tree of Science (ToS)

The Tree of Science or ToS is a way of representing the evolution of a research topic through the implementation of a tree metaphor [16,17]. In this order of ideas, the articles that are in the “root” are considered fundamental or seminal of the investigative process, while the “trunk” shows us the evolution of research based precisely on fundamental knowledge, and finally, the “leaves” show us the detail the different applications or lines of research that are derived around the subject [11]. With this technique, it is possible to determine how the different subfields or applications of information have allowed the evolution to the frontiers of current knowledge and that currently has had innumerable applications in different areas and sectors of knowledge, among which the following stand out: Economics [18], Education [19], Marketing [20], Entrepreneurship [21,22], Psychology [23] among others.

3. Results

3.1. Scientific Analysis of Annual Production

Considering that the pandemic occurred in December 2019, Figure 2 shows the evolution of scientific production from 2020 to 2022. As can be seen, there was a significant increase between the publications of 2020 and 2021, in 2021, publications increased by 276.25%, and in WoS by 246.29%. Then for the year 2022 compared to the year 2021, they increased by 25% for the Scopus database and 33% for the WoS database. Currently, because the year has not yet ended, there are 117 publications, 73 in WoS and 107 in Scopus.

Growth Phase (2020): During this year, 103 articles were published that were not repeated in the two databases, of which 88 were found in Scopus and 54 in WoS. This aspect is highly relevant because it allowed us to note the interest of the scientific community in knowing the associated symptoms and possible methods of intervention of a virus that was rapidly spreading to the point of a pandemic throughout the world, this was an aspect that gained high relevance in the community of different scientists from all areas of knowledge. The most cited article is by the author Monaghesh1, where he describes through a review of the literature how telecare has become a decisive factor in the processes of prevention, diagnosis, treatment, and control of COVID-19. Because these works were the first around the world to expose the subject, they are the ones that have many 7318 citations.

Rapid development phase (2021): This was the year in which publications around the world related to this topic became enormously widespread because scientists sought to socialize the different methods that were being experimented with to counteract the virus and the experimental results, as well as ways of monitoring patients with comorbidities (hypertension, diabetes, obesity) who, due to their condition, had to have special monitoring to avoid complications and massive infections. During 2021, a total of 222 articles were published in the Scopus database and 133 in the WoS database, the citations decreased a bit but are still high due to the novelty of the publications, in this case, 5041. The most cited article is by author Hatmal [23], who used machine learning-based methods to predict the severity of vaccination side effects in Jordan.

Stability phase (2022–2023): During the last two years of scientific production, publications have stabilized, 310 publications with 1562 citations were published. This conglomerate of publications during these four years denotes the current validity of the research topic. The most cited article is by the author Choi18, where he shows a sociodemographic analysis of telecare for the elderly during COVID-19 and how the process of appropriation of technology by this population has occurred.

3.2. Network of Countries

Globally, there was significant concern about the impact that COVID-19 had on society, regardless of nationality, creed, or race. Millions of people felt the endemic effects closely, whether on the economy, health, or their emotions. However, the global community has slowly forgotten about this infection, and unfortunately, the associated symptomatology persists worldwide. Nevertheless, the experience gained by researchers from different knowledge areas in the identification and intervention processes of COVID-19, with relevant works that could reach 9183 citations in the top 10 countries in the world, cannot be dismissed. The scientific community demonstrated that it is possible to manage this infection effectively using disruptive technologies.

As shown in

Table 2. Countries.

Country	Production		Citation		Q1	Q2	Q3	Q4
USA	315	33.65%	3551	28.38%	153	47	14	2
India	81	8.65%	477	3.81%	8	3	7	10
Australia	69	7.37%	1325	10.59%	24	18	8	3
China	41	4.38%	1305	10.43%	24	5	1	0
Canada	36	3.85%	382	3.05%	16	4	0	1
United Kingdom	31	3.31%	293	2.34%	15	2	0	3
Italy	29	3.1%	536	4.28%	14	8	3	0
Iran	28	2.99%	834	6.67%	5	6	2	4
Brazil	27	2.88%	125	1%	11	6	3	1
Spain	20	2.14%	355	2.84%	12	3	0	0

, the United States was the country that published the most on the use of disruptive technologies for the treatment of COVID-19 (315), thus conglomerating 33.65% of the scientific production and boasting the highest number of citations (3551), i.e., 23.38%. An analysis of 315 publications identifying specifically those found in journals indexed in scientific databases, 153 of them were published in Q1 journals (49%), 47 in Q2 (15%), 14 in Q3 (4%), and 2 in Q4 (1%). The most cited article by author Aziz with 131 citations aims to identify how telehealth can be carried out for high-risk pregnancies in the context of the COVID-19 pandemic [24]. India contributed 8.65% of the global production with 81 articles. However, this production is ranked sixth in the ranking of most-cited articles with 477 citations, equivalent to 3.81% of global citations. Of these 81 articles, 8 of them are published in Q1 journals (10%), 3 in Q2 journals (4%), 7 in Q3 journals (9%), and 10 in Q4 journals (12%). Regarding the most cited article, the work of Anuradha Khattar [25], with 50 citations, analyzed the effects of COVID-19 on the learning styles, activities, and mental health of young Indian students using a machine learning approach.

Australia has 69 articles, i.e., 7.37% of the overall production, and is the second country with the highest number of article citations after the United States, 1325, which equates to 10.59% of global citations. Of the 69 articles, 24 of them were published in Q1 journals (35%), 18 in Q2 (26%), 8 in Q3 (12%), and 3 in Q4 (4%). The author Xiaoyun Zhou [2] with 584 citations, focused on analyzing the role of telehealth in reducing the mental health burden from COVID-19. China contributed 41 articles, i.e., 4.38% of the scientific production, and ranked as the third country in terms of citation index 1305, i.e., 10.43%, only surpassed by the United States and Australia. Of the 41 published articles, 24 were published in Q1 journals (59%), 5 in Q2 (12%), and 1 in Q3 (2%). The scientific community had high expectations for scientific publications from China because the COVID outbreak originated precisely in this country. The author Xin Guan is the most cited author with 65 citations, and his publication is based on showing the features of a machine learning model based on clinical and inflammatory characteristics for predicting fatal risk in hospitalized COVID-19 patients [26].

Regarding Canada, there are known to be 36 publications contributing 3.85% of the production, ranking it seventh in the number of citations associated with these articles, 382 (3.05%). Of the 36 publications, 16 of them are in Q1 journals (44%), 4 in Q2 journals (11%), and 1 in Q4 journals (3%). The article by author Jia Xue [27] has the most citations, 141 citations, focusing on the analysis of discussions and emotions on Twitter about the COVID-19 pandemic through machine learning implementation, which is a highly relevant topic focused on mental health. In the United Kingdom, 31 articles were published, i.e., 3.31% of the global production, and it ranked ninth in the citation ranking with 293 (2.34%). Of the 31 articles, 15 of them were published in Q1 journals (48%), 2 in Q2 journals (6%), and 3 in Q4 journals (10%). It is worth highlighting the work of author John Torous [28], whose article has 76 citations, and focuses on the opportunity that COVID-19 has provided to transform the processes associated with psychiatric patient care using teleassistance.

Italy contributed 29 publications to this scientometric study, but it is worth noting that the publications had a high number of citations, placing this country fifth in the citation index, 536, equivalent to 4.28%. Just like China, the global academic community was very expectant of Italian publications, as it was one of the countries most affected by this virus worldwide and a reference for many treatments. Of the 29 publications, 14 are in Q1 journals (48%), 8 in Q2 journals (28%), and 3 in Q3 journals (10%). Among the most cited publications is the one by author Luca Flesia [29], whose work focused on predicting perceived stress related to the COVID-19 outbreak through stable psychological traits based on machine learning models. In Iran, 28 works were published, i.e., 2.99%, and even though the number of publications is lower, it is important to highlight that they are the fourth country with the most publications, 834 (6.67%). Of the 28 published articles, 5 articles were published in Q1 journals (18%), 6 in Q2 journals (21%), 2 in Q3 journals (7%), and 4 in Q4 journals (14%). It is evident that the author Monaghesh [1], currently has the highest number of citations, 597, where the role that tele-assisted technologies must have during the pandemic is exposed.

Regarding Brazil, 27 publications were achieved in this theme with 2.88% and it is the last country with the number of citations, 125 (1%). Of the 27 articles, 11 of them were published in Q1 journals (41%), 6 in Q2 journals (22%), 3 in Q3 journals (11%), 1 in Q4 journal (4%). The article by Janine Alessi [30], has the highest number of citations, 29, and establishes the telehealth strategy to mitigate the negative psychological impact of the COVID-19 pandemic on type 2 diabetes. Finally, in Spain, 20 articles were published (2.14%), surpassing Brazil and the United Kingdom in citations with 355 (2.84%). Of the 20 publications, 12 of them were in Q1 journals (60%), and 3 in Q2 journals (15%). The article by Jessica Marian Goodman-Casanova [3] stands out, she focused her research on identifying how telehealth home support can be performed during COVID-19 confinement for community-dwelling older adults with mild cognitive impairment or mild dementia.

In this same sequence, it is also important to identify how collaboration and citation indices among countries have behaved (see Figure 3). Where five communities can be identified, led by the United Kingdom, USA, Saudi Arabia, and Nigeria. The first community (light green) shows a relationship of journal citations from different countries but predominantly from the European continent, led by the United Kingdom. These publications focus on analyzing the different risk factors of the coronavirus and how to make contributions to identify these factors early to avoid mortality risks. Among the most accepted articles in this community is the author Torous [28] about how COVID-based processes have supported the identification of procedures for implementing teleassistance in older adults. The second community (blue), is led by the United States, with a relationship with countries of high technological development such as China, Canada, Japan, and Russia among others, the themes that relate to this community are largely focused on the development of technology for tracking symptoms associated with COVID-19, where the work of Aziz [24] stands out, showing how teleassistance can help in the monitoring of patients with high-risk pregnancies.

The third community (purple) is led by publications made by oriental countries, among which Saudi Arabia, Lebanon, and India stand out, emphasizing the identification of different global efforts so that health systems can respond to the high demand for services given the health emergency of COVID-19, using of course systems based on artificial intelligence, the article by Alballa [5] is located as a reference in this community by showing the usability of artificial intelligence to analyze the risk factors of COVID-19. The fourth community (orange) involves relations of themes and publications from some European countries, Chile, and Nigeria, being led by the latter. The publications focus on identifying the different factors that can be analyzed with machine learning solutions for mental health disorders, author Montoya [31] through his research details how teleassistance-based systems can affect the health of health professionals. Finally, the last community (dark green), led by Australia with the participation of neighboring countries such as New Zealand with geographical proximity, is very specifically focused on analyzing the impact on the mental health of patients during the pandemic. The article published in this community by author Zhou constitutes the most cited article in the entire review process.

3.3. Journal Analysis

In

Table 3. Top Ten Journals.

Journal	WoS	Scopus	Impact Factor	H-Index	Quartile
Journal of Medical Internet Research	12	18	1.99	178	Q1
Telemedicine and E-Health	14	14	1.24	87	Q1
Plos One	8	14	0.89	404	Q1
International Journal of Environmental Research and Public Health	7	9	0.83	167	Q2
Scientific Reports	9	8	0.97	282	Q1
Frontiers in Psychiatry	8	9	1.22	96	Q1
Cureus Journal of Medical Science	7	0	-	-	-
Diagnostics	4	7	0.67	52	Q2
Frontiers in Public Health	7	5	1.13	80	Q1
Journal of Clinical Medicine	4	7	0.94	95	Q1

, the top 10 journals that publish the most on the mentioned topic are described, where it can be noted that the Journal of Medical Internet Research has the highest number of publications and additionally has the highest impact factor. Among the most outstanding works is that of author Isautier [32], which offers a comprehensive view of the perceptions, experiences, and satisfaction levels of users with telemedicine services in Australia during the COVID-19 pandemic. This cross-sectional survey study captures a critical moment when telemedicine experienced massive adoption, driven by the need to maintain medical care continuity while minimizing virus transmission risks. A fundamental contribution of this work is the empirical assessment of the shift towards telemedicine in a health crisis environment. The findings reveal widespread acceptance and positive satisfaction among telemedicine users, highlighting how the unprecedented circumstances of the pandemic have accelerated the digitalization of health services and have modified patients’ expectations and preferences. The second journal, Telemedicine, and E-Health has also included a compendium of various highly relevant articles, among which can be found that of author Andersen [33], which focuses on the use of telemedicine services by people diagnosed with mental health conditions during the COVID-19 pandemic. This study provides a detailed analysis of how the pandemic has affected how patients with mental health needs access and use telemedicine services. One of the key contributions of this work is its focus on a specific population: individuals diagnosed with mental health conditions. By focusing on this group, the authors provide valuable insights into the challenges and opportunities that telemedicine presents for the continuity of care in a critically important sector of the health system. The study investigates aspects such as the adoption rates of telemedicine, the barriers to its use, and patient satisfaction with this type of care during the unprecedented period of the pandemic.

The third journal in terms of publications is PLOS ONE, which has made significant research contributions in this field, such as the work of author Cheon [34], representing a significant advance in the field of epidemiology and pharmacovigilance through the application of language models and unsupervised machine learning to analyze adverse events associated with COVID-19 vaccines. This study focuses on extracting and understanding the characteristics and patterns of adverse events reported following vaccination against COVID-19, using free-text data obtained from public databases and platforms for reporting side effects. The main contribution of this work lies in its innovative approach to analyzing vaccine adverse events, which traditionally has been conducted through manual or semi-automated methods requiring a considerable number of resources and time. By employing advanced language models and unsupervised machine learning techniques, the authors can efficiently process and analyze large sets of free-text data, identifying significant patterns and trends in adverse event reports without the need for prior manual labeling. The fourth journal is the International Journal of Environmental Research and Public Health, with notable contributions such as those by author Werkmeister [35], who provides a detailed and thoughtful analysis of the experiences of mental health clinicians in Aotearoa New Zealand during the COVID-19 lockdown, especially regarding the implementation and use of telehealth. This qualitative study explores the perceptions, challenges, and opportunities that mental health professionals faced when adapting their practices to a telehealth environment during the lockdown. Through in-depth interviews with clinicians working in various mental health contexts, the study offers valuable insight into how the pandemic has accelerated the adoption of telehealth, highlighting both the benefits and obstacles experienced by mental health service providers. A key contribution of the article is the identification of effective practices and adaptive strategies developed by clinicians to overcome the limitations imposed by the lockdown and maintain the continuity and quality of care for their patients. These include the need for flexibility in therapeutic approaches, the importance of establishing a connection and effective communication through virtual means, and the recognition of the digital divide as a significant barrier for some patients.

The fifth journal in the number of publications is Scientific Reports, with significant contributions such as that by An [36] as a pioneering study that employs advanced machine learning techniques to predict the mortality of patients diagnosed with COVID-19 in South Korea. This study represents a significant effort to address one of the biggest challenges during the pandemic: quickly identifying patients at high risk of mortality to optimize clinical decisions and resource allocation. The study uses a machine learning approach to analyze data from a broad cohort of patients diagnosed with COVID-19 in South Korea, including a variety of demographic, clinical, and laboratory variables. The main goal is to develop and validate a predictive model that can estimate the probability of mortality in COVID-19 patients, based on information available at the time of diagnosis or shortly thereafter. Frontiers in Psychiatry is ranked sixth, where the article by author Lynch [37] stands out, addressing the challenge of providing comprehensive recovery services through telehealth for people living with complex psychosis in New York City during the COVID-19 crisis. This study is a crucial exploration of how mental health interventions can be adapted and applied in a telehealth format to meet the needs of a vulnerable population during a global pandemic. The research focuses on three main components: the design of a comprehensive recovery service adapted to telehealth, the implementation of this service in a densely populated urban context during an unprecedented health crisis, and the evaluation of its acceptability among users, including both patients and mental health providers.

Cureus Journal of Medical Science ranks seventh, highlighting the work of author Abraham [38] which addresses a critical aspect of medical education during the COVID-19 pandemic: the integration of third-year medical students into telemedicine practice during their internal medicine clinical rotations. This study provides valuable insight into how medical education institutions can adapt to the challenges imposed by the pandemic, ensuring both the continuity of medical education and the safety of students and patients. One of the most significant contributions of the article is the detailed description of the design and implementation of a telemedicine program specifically tailored for medical students in their internal medicine rotations. This program not only allowed students to continue their clinical education safely during the peak of the pandemic but also offered them the opportunity to develop essential skills in delivering care through telemedicine platforms, a competence increasingly important in the current health landscape. In eighth place is the journal Diagnostics, with contributions from author Fatima [39] presenting an innovative application of deep learning technology for COVID-19 detection within vehicles. This study addresses the critical need to implement preventive measures and rapid virus detection in confined spaces, such as vehicles, where the risk of transmission can be significantly high due to the proximity of occupants and limited air circulation. The core of the article focuses on the development and validation of a detection system based on an advanced machine learning approach, named Deep Extreme Learning Machine (DELM), which combines the strengths of deep learning with the computational efficiency of extreme learning machines. This approach allows for the rapid identification of potential COVID-19 cases through the analysis of biometric and environmental parameters within the vehicle, using non-invasive sensors. One of the most significant contributions of this work is the proposal of a model that can be integrated into modern vehicle systems to continuously monitor vital signs and other health indicators of the occupants, such as body temperature, respiratory rate, and the concentration of certain compounds in the air that may indicate the presence of the virus. By employing machine learning techniques to analyze these data in real time, the proposed system can alert occupants to the possible presence of the virus, facilitating early isolation measures and preventing transmission.

In ninth place is the journal Frontiers in Public Health, with significant contributions such as that by author Chapman [40], which explores the development and implementation of a new brain health index (BrainHealth Index) that was improved through training delivered by telemedicine during the COVID-19 pandemic. This innovative study addresses the urgent need to maintain and enhance brain health at a time when pandemic restrictions have limited access to in-person health interventions. A fundamental contribution of this work is the creation of a brain health index, which seeks to comprehensively quantify brain well-being across various dimensions, including cognition, emotional resilience, and social well-being. The novelty of this index lies in its holistic approach to evaluating brain health, moving beyond traditional measures that predominantly focus on specific cognitive or neurological aspects. Lastly, there is the Journal of Clinical Medicine, highlighting the work of author Callejon [41] which investigates the utility of loss of smell and taste as predictors of COVID-19 infection, using advanced machine learning techniques for this purpose. This study represents a significant advancement in the early detection of COVID-19, providing an effective tool to improve diagnostic and containment strategies for the virus. One of the main contributions of the study is the application of machine learning algorithms to analyze patient data and determine the relationship between the loss of smell and taste and COVID-19 infection. The authors use a broad database of patients, including both those with a confirmed diagnosis of COVID-19 and individuals with symptoms but without a confirmed diagnosis of the disease, to train and validate predictive models that can identify infection based on these sensory symptoms.

Through network analysis, it can be identified that based on searches performed in the two databases, Scopus and WoS (Figure 4), it is possible to determine that each node corresponds directly to a specific journal, and the edges refer to the relationship existing in the references that are generated between the journals. By analyzing the three communities that emerge from this analysis, it can be identified that the largest community (purple) is focused on the use of data mining techniques to identify and treat different COVID risk factors. One of the most outstanding articles in this network focuses on the use of machine learning to assess the risk of mortality from COVID-19 [42]. The second community (orange) focuses on different variables analyzed during the pandemic and how telemedicine was accepted by the global community. Among the most notable articles is a literature review that analyzes how telehealth provided support for patient care amid lockdown due to COVID-19 infection, which was published in the journal Computers in Biology and Medicine. While it is true that it is not among the top 10 journals with the most published articles, it leads to the inter-citation relationship between journals [4]. The third community (green) identifies how data analytics-based techniques have allowed the understanding of different infectious processes worldwide and how some lessons learned have been used to address issues related to COVID. Among the most referenced works of this community, the work of Keshvardoost can be highlighted, analyzing the role of telehealth in the management of COVID-19 based on lessons learned from previous outbreaks of SARS, MERS, and Ebola. This was published in the journal Telemedicine and e-Health, which is second in terms of the number of published works [43].

3.4. Analysis by Author

In

Table 4. Top Ten Authors.

No	Researcher	Total Articles	Scopus-Index	Affiliation
1	Chen H. [44]	6	35	Shantou University, Shantou, China
2	Ghosh A. [45]	6	1	Christ University, Bengaluru, Bengaluru, India
3	Liu Y. [46]	6	2	School Of Medicine, Xiamen, China
4	Looi J. [47]	6	24	The Australian National University, Canberra, Australia
5	Wang Y. [48]	6	5	Universidad De La Academia China De Ciencias, Beijing, China
6	Chen C. [49]	5	11	College Of Medicine, Taipei, Taiwan
7	Moni M.A. [50]	5	40	The University Of Queensland, Brisbane, Australia
8	Reay R. [51]	5	12	The Australian National University, Canberra, Australia
9	Alessi J. [52]	4	6	Universidade Federal Do Rio Grande Do Sul, Porto Alegre, Brazil
10	Amaral B. [53]	4	3	Pontifícia Universidade Católica Do Rio Grande Do Sul, Porto Alegre, Brazil

, we can observe the list of top authors concerning the development of IoT-based technologies for the treatment of persistent COVID-19 symptoms, where there is a predominance of authors from China and Australia who have contributed to different developments of support technology for monitoring and managing COVID-19. The most cited author is Chen H. [44], from Shantou University, Shantou, China, who is a co-author of a wide number of works related to this topic. In his most cited article, he focuses on the development and validation of a machine learning model to predict the daily incidence of COVID-19 in 215 countries and territories in real-time, to foresee the spread of COVID-19 worldwide using historical disease data, as well as other relevant factors, to make predictions about the number of daily COVID-19 cases expected in different geographical locations. In addition to the model development, the article also includes information on how its accuracy was validated using real COVID-19 incidence data from various countries and territories. The second author is Ghosh A. [45] from Christ University, Bengaluru, India, who addresses the issue of predicting depression in the context of the COVID-19 pandemic. In this work, the authors explore how the pandemic and its impacts can have significant effects on people’s mental health, particularly in the manifestation or exacerbation of depression. The core of the article focuses on the application of machine learning techniques to develop a model capable of identifying signs of depression in data related to the pandemic.

The third author is Liu Y. [46] from the School of Medicine, Xiamen, China, who focuses on developing a machine learning-based risk prediction model for Post-Traumatic Stress Disorder (PTSD) in the context of the COVID-19 pandemic. In this study, the authors investigate how the pandemic and its aftermath can contribute to the development of PTSD in individuals. The core of the article focuses on the application of machine learning techniques to create a model that can identify risk factors and patterns associated with the onset of PTSD during the pandemic. This involves collecting and analyzing data related to COVID-19 and people’s mental health. The main goal is to provide a predictive tool that helps identify individuals at risk of developing PTSD in this specific context, which is valuable for medical care and early intervention. The fourth author is Looi J. [47] from The Australian National University, Canberra, Australia, which focuses on reviewing and summarizing the available evidence on the provision of mental health services through telemedicine during the COVID-19 pandemic. In the context of the pandemic, when mobility restrictions and social distancing limited access to traditional mental health services, telemedicine became an essential tool for providing care to people in need of emotional and psychological support. The article analyzes the evidence accumulated up to that point on the effectiveness and safety of mental health services through telemedicine and offers recommendations and clinical guidelines based on this evidence.

The fifth author is Wang Y. [48] from the University of the Chinese Academy of Sciences, Beijing, China, focuses on investigating the subjective well-being of Chinese users of Sina Weibo, a social media platform, during the period of residential lockdown imposed due to the COVID-19 pandemic. This study uses machine learning techniques to analyze data collected from posts and conversations on Sina Weibo to understand how the emotional well-being of users is affected in the context of movement restrictions and social distancing. The article examines the emotions, concerns, and topics discussed by users in their online interactions during this period and seeks to identify patterns and trends in their moods and thoughts. The sixth author is Chen C. [49] from the College of Medicine, Taipei, Taiwan, and focuses on analyzing and predicting disease outcomes in patients affected by COVID-19 and pneumonia. The authors employ a combination of statistical analysis and machine learning techniques to examine clinical and epidemiological data to anticipate the medical outcomes of these patients. Essentially, the study focuses on the application of advanced tools to improve understanding and forecasting capability in the management of these diseases.

The seventh and eighth authors are from Australia. Moni M.A. [50] from the University of Queensland, Brisbane, Australia, focuses on developing a machine-learning model aimed at identifying early-stage symptoms in patients infected by the SARS-CoV-2 virus, responsible for COVID-19. The authors propose using advanced artificial intelligence techniques and data analysis to detect initial signs of the disease in patients, which is crucial for early detection, adequate treatment, and prevention of the disease’s spread. Reay R. [51] from the Australian National University, Canberra, Australia, addresses a significant change in the provision of mental health services in Australia during the COVID-19 pandemic. The study focuses on implementing telemedicine services, specifically in the field of mental health, through the “Better Access” program. The authors analyze how this transition to telemedicine impacted the delivery of mental health services by allied health professionals, evaluating both the benefits and associated challenges.

Finally, the last two authors are from Brazil. Alessi J. [52] from the Universidade Federal do Rio Grande do Sul, Porto Alegre, Brazil, and Amaral B from the Pontifícia Universidade Católica do Rio Grande do Sul, Porto Alegre, Brazil, who have worked together on various publications. Among the most notable works is a publication that focuses on evaluating the effect of a telemedicine intervention on the metabolic profile of patients with diabetes mellitus during the COVID-19 pandemic. This study is based on a randomized clinical trial in which researchers implemented a telemedicine approach to manage patients with diabetes. The article analyzes how the use of telemedicine influenced diabetes control and the evolution of the metabolic parameters of the participants, compared with those who received traditional medical care. This is especially relevant in the context of the pandemic, where telemedicine became an important alternative for monitoring chronic diseases such as diabetes, due to restrictions on in-person care.

When analyzing the different interactions between the leading authors on the topic, we can observe the consolidation of four communities. The first, in purple, shows a strong relationship between the production of authors from Asian origins such as China Taiwan, and Australia, whose contributions are focused on crucial strategies for the prevention of infections and the disruption of the pathogen transfer chain in the context of elective surgeries. These studies are particularly relevant in the field of arthroplasty and other elective orthopedic surgeries, where the prevention of postoperative infections is fundamental to the success of the procedure and the well-being of the patient. One of the main contributions of the work is its detailed analysis of the pathways through which pathogens can be transferred and introduced into the surgical site, highlighting the importance of a series of preventive measures to break this chain of transmission. The authors discuss in-depth infection control practices before, during, and after surgery, including the optimization of the surgical environment, the proper handling of instruments and materials, and the use of sterile techniques to minimize the risk of infection [30]. The second community, in green, showcases the contributions of authors from India who have worked on a crucial aspect of medical care during the COVID-19 pandemic: the management of diabetes mellitus through teleconsultation. This study provides practical and evidence-based guidelines developed by an expert group from the Indian Council of Medical Research (ICMR) to facilitate effective and safe care for diabetes patients in the context of the pandemic and in similar situations that limit in-person medical consultations. A fundamental contribution of this work is the recognition of telemedicine as an essential tool for diabetes management in times of health crisis. The authors highlight how teleconsultation can overcome physical barriers, ensure the continuity of care, and minimize the risk of virus exposure for both patients and health professionals [53], see Figure 5.

The third community, in orange, shows the contributions of Brazilian authors who explore an innovative application of machine learning technology for the detection of positive and recovered COVID-19 patients through voice analysis. This study represents a significant advance in the field of telemedicine and remote diagnosis, offering a non-invasive and accessible methodology for disease monitoring. One of the main contributions of the work is the development of a machine learning model that can differentiate between the voices of individuals infected by SARS-CoV-2 and those of non-infected individuals, as well as identify those who have recovered from the disease. To achieve this, the authors analyzed a wide range of voice characteristics collected from participants both positive and recovered from COVID-19, along with a control group of non-infected individuals [54]. The fourth community focuses on the interaction of Australian authors examining how the COVID-19 pandemic has influenced the provision of outpatient psychiatric services in private practice in Australia. This analysis, based on the study by Looi, J. C., et al., specifically centers on the use of new telepsychiatry services enabled by the Medicare Benefits Schedule (MBS) compared to face-to-face psychiatric services during the third quarter of 2020. The study provides a detailed analysis of the change in the provision of psychiatric services in response to the pandemic, highlighting a significant increase in the adoption of telepsychiatry. This adoption was facilitated by the introduction of new telemedicine items in the MBS, a measure designed to support the continuity of psychiatric care while minimizing virus transmission risks. The authors analyze trends in the use of outpatient psychiatric services, comparing telepsychiatry sessions with face-to-face consultations, and evaluate the impact of these modalities on psychiatric care during a period of unprecedented public health restrictions [55].

4. Tree of Science

4.1. Root

In the current landscape, marked by the COVID-19 pandemic, the understanding and management of the virus’s prolonged effects, especially in the realm of mental health, have emerged as a critical area of research. The studies conducted by Spitzer [56] have established valuable tools for the assessment of common mental health disorders, such as anxiety and depression, through the GAD-7 and the PHQ-9, respectively. These tools have proven to be essential not only for clinical diagnosis but also for applied research, facilitating the standardized and validated evaluation of symptoms prevalent among patients affected by COVID-19. The integration of machine learning techniques with the responses obtained from these scales opens new avenues for identifying complex patterns in the manifestation of psychological symptoms associated with COVID-19. This methodological approach allows not only for predicting the evolution of anxiety and depression disorders in individuals affected by the virus but also for adapting therapeutic interventions more accurately and personalized, based on the detailed analysis of large data sets.

On the other hand, the work of Brooks [57] offers a critical view of the psychological impact of quarantine measures, highlighting how isolation and pandemic-related stress can significantly contribute to the global burden of mental disorders. The research proposes evidence-based strategies to reduce the negative impact of quarantine on mental health, an aspect particularly relevant for addressing the prolonged symptoms of COVID-19. The synthesis of these fundamental findings with the advanced analysis provided by machine learning represents a significant step forward in our ability to address the emerging mental health challenges of the pandemic. This holistic approach not only enhances our understanding of the etiology and progression of mental disorders in the context of COVID-19 but also informs the development of public policies and clinical strategies to mitigate the adverse effects of the pandemic on the mental health of the population. The intersection of validated psychometry, machine learning, and applied research in mental health highlights a promising direction for future research, pointing toward more effective and empathetic management of mental disorders in a world still navigating the consequences of COVID-19.

The research conducted by Kroenke [58], focused on the validation of the PHQ-9 as a brief measure for assessing depression severity, represents a fundamental milestone in the field of psychometry and mental health. This study, published in the Journal of General Internal Medicine, has provided health professionals with a quick, reliable, and highly valid tool for depression detection, a disorder that has seen an increase in prevalence in the context of the COVID-19 pandemic. In the current context, marked by the global health crisis of COVID-19, the study by Kroenke and his collaborators gains even greater relevance. The psychological challenges imposed by the pandemic—including social isolation, economic uncertainty, and fear of the disease—have exacerbated mental health conditions in many people, making tools like the PHQ-9 indispensable for effectively identifying and managing emerging or worsened cases of depression, see Figure 6.

4.2. Trunk

In the vast and complex ecosystem of human knowledge, where the metaphor of the science tree helps us visualize the interconnection between different areas of knowledge, the studies by Prout [59], Roma [60], and Eder [61] stand out as robust trunks in the branch that deals with the psychological and behavioral effects of the COVID-19 pandemic. These works, grounded in the application of machine learning and deep data analysis, constitute solid foundations that support the growth of new research aimed at facing one of the most significant challenges of our era: the global health crisis caused by COVID-19.

The study by Prout [59], by identifying predictors of psychological distress during the pandemic through machine learning, not only sheds light on the complex patterns underlying states of anxiety and depression in this unique context but also offers fertile ground for the development of psychological interventions and public policies focused on the protection and promotion of mental health. This work provides critical insights into how various factors, from pre-existing conditions to current circumstances of isolation and stress, can deeply affect individuals’ mental well-being, underlining the need for personalized and evidence-based support strategies.

On their part, Roma et al. [60] explore how to improve adherence to protective health measures during the outbreak, using machine learning algorithms to analyze the effectiveness of various communication strategies and their impact on public behavior. This article broadens our understanding of the social and psychological dynamics that influence the acceptance and practice of preventive behaviors, highlighting the importance of empathetic and informative public health communication approaches. The ability to accurately predict which messages will resonate best with different population segments is crucial for designing more effective public health campaigns, especially at times when citizen cooperation is essential to control the virus’s spread.

The work of Eder et al. [61], focusing on predicting fear and health perception during the pandemic using machine learning, opens new perspectives on the emotional impact of the health crisis on the global population. By deciphering how fear of the virus and risk perception influence health behavior, this study contributes to the knowledge base necessary to design psychological interventions that not only address the stress and anxiety generated by the pandemic but also promote healthy and realistic practices among the population. Together, these articles form a robust trunk from which various research lines emerge, ranging from the development of diagnostic tools to the implementation of intervention strategies based on a detailed understanding of human behavior in times of crisis. The integration of their findings with the emerging field of machine learning not only facilitates the identification of complex patterns in large data sets but also promotes the creation of innovative and effective solutions to mitigate the adverse effects of the pandemic on mental health and social well-being. This body of work underscores the importance of a multidisciplinary approach to addressing health crises, combining insights from psychology, public health, informatics, and beyond, to navigate and eventually overcome the challenges presented by COVID-19 and future pandemics. Authors should discuss the results and how they can be interpreted from the perspective of previous studies and the working hypotheses. The findings and their implications should be discussed in the broadest context possible. Future research directions may also be highlighted.

4.3. Branch 1: Predictive Models and Clinical Prognostics in Pandemics and Their Post-Pandemic Effect

This branch reflects a growing and significant trend in knowledge generation focused on the understanding and application of advanced technologies such as machine learning to predict and manage the consequences of pandemics. The selected studies reveal a shift towards the valuation and leveraging of massive data and predictive analysis in the clinical field, a trend expected to have a lasting impact in the post-pandemic period. Among the most prominent authors is Békés et al. [62] examining the acceptance of telepsychotherapy by psychotherapists during the COVID-19 pandemic. The use of machine learning approaches to understand attitudes towards telepsychotherapy can be fundamental in predicting and improving the long-term delivery of mental health services. This study anticipates an evolution in clinical practice that is likely to endure, consolidating telepsychotherapy as a standard component of psychotherapeutic services.

Lee et al. [63], on the other hand, apply machine learning techniques to analyze national health insurance data and study the relationship between blood transfusions, mortality, and the hospitalization period in COVID-19 patients. This research underlines how artificial intelligence can optimize clinical decision-making and resource management, with implications that extend beyond the management of an immediate health crisis to the improvement of health systems in the future. Finally, the study by Pathak [64] uses neural networks to analyze emotional reactions to World Health Organization posts during the pandemic. This innovative approach offers a new paradigm for monitoring and understanding public mental health and the population’s emotional responses, providing crucial information for the design of effective public health interventions and communications in the recovery phase and preparation for future pandemics.

These studies are indicative of an emerging paradigm in which the combination of large databases with predictive algorithms and machine learning techniques is transforming healthcare. This approach not only improves the accuracy and efficiency of diagnosis and treatment during a pandemic but also lays the groundwork for more proactive and data-based health management in the future, emphasizing the relevance of predictive and personalized medicine. As we move forward in the post-pandemic era, the integration of these technologies into clinical practice is called to be a pillar in the optimization of global health and disease management.

4.4. Branch 2: Mental Health and Psychosocial Responses to the Pandemic and Their Post-Pandemic Impact

The branch reflects a progressive exploration at the intersection of mental health, social behavior, and advanced data analysis. The provided studies demonstrate a multidisciplinary approach that integrates psychology, public health, and data science to understand and address the long-term effects of the COVID-19 pandemic on the psychological well-being of populations. The article by Chen [65] uses machine learning to investigate arterial pulse variability and its correlations with the side effects of the Pfizer-BioNTech COVID-19 vaccine (BNT162b2). Although the focus is biomedical, the findings have implications for mental health, contributing to a better understanding of post-vaccination physiological responses that can influence the perception of well-being and vaccine-related anxiety.

Bakkeli [66] focuses on predicting the perception of the risk of exposure to COVID-19 using machine learning techniques. This study contributes to the knowledge of how people assess their vulnerability to the disease, which is intrinsically linked to psychosocial responses such as anxiety and preventive behavior. Risk perception is a key factor that can continue to affect the mental health of the population and their behavior long after the pandemic. Finally, Rezapour et al. [67] analyze changes in alcohol consumption habits among healthcare workers in the United States during the pandemic, also using machine learning. This study is particularly relevant to post-pandemic mental health, as alcohol consumption can be both a response to stress and a risk factor for the development of psychological disorders. Identifying changes in consumption habits is crucial for implementing support strategies aimed at the healthcare worker population, who have been on the front line during the health crisis.

These studies collectively anticipate a post-pandemic era where machine learning technology will serve as a vital tool for monitoring and responding to emerging challenges in mental health. The knowledge generated by these works highlights the importance of early interventions and the adaptation of mental health services to address the far-reaching aftermath of the pandemic, preparing societies for a comprehensive recovery and promoting resilience in the face of possible future crises. The trend in research reflects a recognition of the complexity of factors affecting mental health and a commitment to the development of proactive and data-based approaches for health and social well-being.

4.5. Branch 3: Innovations in Telehealth and Psychiatric Care during and after the Pandemic

This line encompasses a body of research focused on the use of machine learning and artificial intelligence to enhance the effectiveness of telehealth and psychiatric care at a critical moment for global health. Highlighted studies, such as that by Jamshidi et al. [68], illustrate a significant advancement in the use of machine learning algorithms to predict symptoms and calculate the mortality risk from COVID-19. This approach has direct implications for telehealth, as it allows for more effective remote monitoring and triage, improving resource allocation and rapid response in overloaded clinical settings. Moreover, such predictive models are essential for informing real-time healthcare decisions and optimizing the workload of health personnel.

On the other hand, Enevoldsen et al. [69,70] have explored how monitoring pandemic-related psychopathology through machine learning tools can offer a deeper and more up-to-date understanding of the mental health status of populations. This work reinforces the role of psychiatric telehealth, not only for therapeutic intervention but also for continuous monitoring and mental health prevention. In the future, such tools could be crucial for identifying outbreaks of psychopathology in the general population or vulnerable subgroups. Additionally, the research by Santos et al. [71,72] demonstrates how classification models can help prioritize COVID-19 testing in Brazil. This work highlights the ability of telehealth to improve public health systems by implementing machine-assisted triage strategies, allowing for more agile and adaptive pandemic management. In the long term, this approach could be instrumental in improving response capabilities to future outbreaks and in the efficient management of health resources.

These research efforts not only provide immediate solutions to the challenges posed by the pandemic but also lay the groundwork for a new paradigm in mental health care and psychiatric attention. The integration of telehealth with advanced data analysis promises to transform the healthcare landscape, making care more accessible, proactive, and personalized, which will contribute to the resilience and sustainability of health systems in the post-pandemic world.

5. Discussions

In the discussion section of this article, we delve deeply into how machine learning (ML) and artificial intelligence (AI) are revolutionizing the understanding and management of prolonged COVID-19 symptoms, a critical aspect of the pandemic that continues to affect a significant segment of the recovered population. The studies reviewed in our systematic exploration of the literature demonstrate that ML and AI offer innovative and effective methods for identifying complex patterns and predictors of prolonged symptoms that traditional methods may not capture. This predictive capability is fundamental for customizing treatments and significantly improving early intervention protocols. The use of ML to analyze large patient datasets has revealed specific patterns that can predict the onset of post-COVID-19 symptoms. This information is vital for developing personalized healthcare strategies, allowing healthcare professionals to tailor their therapeutic approaches to individual needs, thus improving long-term health outcomes and the quality of life of patients.

Another crucial point is the optimization of health resources. In times of crisis, such as during a pandemic, health systems can quickly become overwhelmed. AI and ML-based tools not only improve efficiency in identifying and treating patients but also help distribute resources more effectively, ensuring that patients at greater risk receive priority care. Identifying risk factors and demographic characteristics associated with prolonged COVID-19 symptoms can guide more effective public policy formulation. These data enable policymakers to design scientifically grounded, targeted preventive interventions, which can be key to mitigating the impact of future pandemic waves or similar emergencies.

Interdisciplinary collaboration is essential for advancing this field. Interaction between physicians, epidemiologists, data experts, and policymakers is not only crucial for developing effective treatment and prevention strategies but also for ensuring that technological advances are ethically sound and socially beneficial. This collaborative approach will maximize the impact of our findings on public health and clinical care. Despite these advances, research in this field faces limitations, including the need for larger and more diverse datasets to train ML models, and the importance of conducting longitudinal studies to better understand the evolution of COVID-19 symptoms over time. Moreover, future research must address these limitations through the expansion of available databases and the inclusion of multiple perspectives and disciplines. This review has shown that, while there is still much to be performed, the future of personalized, evidence-based medicine in the post-pandemic era is promising thanks to advances in ML and AI.

6. Conclusions

In this study, we have explored the use of machine learning (ML) and artificial intelligence (AI) to analyze the prolonged symptoms of COVID-19, identifying important insights that enhance our understanding of the condition. Our analysis of patient data from those affected by COVID-19 has demonstrated that ML models can detect patterns and predictors of prolonged symptoms effectively, providing a valuable tool for health professionals in the early detection and personalized management of the disease. These insights could potentially inform future strategies in clinical practice and public health policy, although more research is needed to confirm these applications. Our findings suggest that the application of ML to the study of prolonged COVID-19 symptoms significantly enhances our ability to grasp the complexity of the disease’s long-term effects, which could lead to more efficient management of healthcare resources and improved patient care. Furthermore, the identification of risk factors and demographic characteristics associated with prolonged symptoms can aid in the development of targeted prevention strategies.

It is important to integrate advanced data analysis technologies into the healthcare system, not only to address the challenges of the current pandemic but also to enhance our preparedness for future health emergencies. Machine learning and artificial intelligence are crucial for evolving towards more resilient and adaptive health systems. However, we recognize certain limitations in our study. There is a need for more extensive and diverse datasets to train and validate our models effectively. Longitudinal studies are also critical to understanding the evolution of prolonged COVID-19 symptoms over time fully.

Moreover, interdisciplinary collaboration among physicians, epidemiologists, data scientists, and policymakers will be essential to fully realize the potential impacts of our findings on public health and clinical practice. This research provides a strong indication of the potential of machine learning to transform our approach to managing prolonged COVID-19 symptoms, marking a significant advance toward personalized and evidence-based medicine in the post-pandemic era. As we continue to address the challenges posed by this disease, it is crucial to further explore and leverage technological innovations to enhance health outcomes globally.