Navigating the Landscape of Telemedicine Research: A Topic Modeling Approach for the Present and Future

Abstract

Introduction:

Telemedicine, which is the provision of remote clinical services via telecommunication technology, has undergone an upsurge since the COVID-19 pandemic. To capture this paradigm, this study surveyed telemedicine literature, including postpandemic publications, to identify dominant research themes and temporal trends and suggest directions for future research.

Methods:

A corpus of 56,445 telemedicine studies is sourced from PubMed. Latent Dirichlet allocation (LDA) topic modeling performed using the Konstanz Information Miner platform. The textual data for topic modeling were processed by following standard procedures for natural language processing. Moreover, the term frequency–inverse document frequency approach was used to capture the importance of words within the corpus. We assessed perplexity, coherence, and the elbow method to determine the optimal number of topics for modeling.

Results:

The findings confirm the surge in telemedicine research after 2020, signifying its prominence. LDA topic modeling reveals seven distinct research themes, with the most prominent topic being “patient satisfaction” (21.38%) followed by “perspectives and challenges” (17.95%), and “smartphone apps” (14.32%). Furthermore, the results demonstrate a noticeable shift in topics from screening to therapeutic applications of telemedicine.

Conclusions:

This study serves as a guide for a broad range of telemedicine research topics. This synthesis of themes reflects the commitment of scholars to address the changing dynamics and health care needs, such as the COVID-19 pandemic, aging in place, smartphone usage, and technological advancement. The analysis also reveals flexible research responses to policy and contextual shifts, highlighting the collective drive to broaden the application of telemedicine in community health care.

Introduction

Telemedicine is an accessible health care service that uses information and communication technologies to facilitate health-related services between patients and health care providers despite the physical distance.¹ The global spreading of the COVID-19 pandemic and technological advances have increased the use of telemedicine. Before the pandemic, telemedicine had already gained traction as a promising approach for improving the management of various chronic conditions and has been implemented as a health care service in several countries. However, the emergence of COVID-19 accelerated the integration of telemedicine into health care systems.² Advancements in data science and information and communication technology are facilitating the proliferation of telemedicine. These technological advances promote the seamless exchange of health information between patients and health care providers, support informed decision-making, and encourage the continued expansion of telemedicine, overcoming geographical limitations.^3,4 Thus, telemedicine has become a rapidly growing research topic.

As telemedicine research continues to expand, there is a growing demand for evidence supporting its impact on individual and community health outcomes. To fulfill this demand, systematic reviews and meta-analyses (SRMA) have been used to synthesize results from existing studies and discover empirical evidence. These comprehensive analyses have covered topics from the use of mobile health technologies in developing countries, the impact of telemedicine on the management of diseases such as cancers, chronic conditions, mortality, health care for the elderly, and home monitoring after surgery.^5

–10 However, it is vital to recognize the inherent limitations of these approaches. The time-consuming and resource-intensive nature of the SRMA, which stems from manual study selection and appraisal processes, poses a significant challenge.¹¹

Moreover, SRMA methodologies require strict inclusion and exclusion criteria, constraining their ability to illuminate broader research trends and topics.¹¹ Given the challenge of condensing a large volume of literature into a systematic review and interpreting the findings coherently, there is a growing demand for alternative methods that can extract knowledge in a more automated and unbiased manner, which can be achieved via text-mining techniques.

Topic modeling, a prominent technique in text-mining, has emerged as a powerful solution to address these challenges.¹² Topic modeling is widely used for text data analysis and is well-suited for exploring the vast landscape of telemedicine articles.¹³ Studies utilizing topic modeling have analyzed various data source related to telemedicine, including Twitter,¹⁴ online discussion forums,¹⁵ and mobile applications' reviews.¹⁶ Baird et al.¹⁴ identified the themes of consumer perceptions of telemedicine from 10,689 tweet text data. Similarly, Nitiema¹⁵ examined health care professionals' opinions regarding telemedicine by analyzing 914 comments from an online discussion forum. Shan et al.¹⁶ investigated public trust in artificial intelligence-based mental health applications by analyzing 3,931 reviews. However, these studies had relatively small sample sizes,^15,16 or were limited to a specific area of telemedicine research.^14
–16

Moreover, to the best of our knowledge, no studies have captured the trends and revealed the research topics on telemedicine scientific literature, including the postpandemic era. Considering the substantial increase in the number of articles related to telemedicine in PubMed,¹⁷ which consistently published 4,000–5,000 articles in 2021 and 2022, this increase in publication volume reflects the burgeoning interest in telemedicine and non-face-to-face care after the COVID-19 pandemic. Therefore, to comprehensively understand the overall trends and changes in telemedicine, the dataset must contain the most recent research, including articles published after the COVID-19 pandemic. In addition, by analyzing the recently published articles, the results can provide information in terms of telemedicine application in the community, where “Aging in Place” is a crucial concept in the modern aging society.¹⁸

On the contrary, a lack of methodological rigor of the previous topic modeling studies exists. Determining the number of topics without proper validation may lead to overfitting or underfitting of the topic modeling, which in turn distorts the underlying themes. Studies in which scholars determined the number of topics subjectively, relied on a single metric, such as coverage or coherence, without reverification, or constructed topic models solely based on frequently occurring words encountered methodological limitations related to robustness and objectivity in identifying topics.^19,20

In fact, relying completely on frequently occurring words for topic modeling can lead to bias toward high-frequency terms, disregarding the semantic context, and potentially neglecting important, yet, less frequently used words that contribute to the richness of the topic. Therefore, a more systematic and empirical approach is required to increase the reliability and validity of topic modeling and ensure a more accurate reflection of the complex themes within telemedicine research.

The aim of this study is to comprehensively investigate extant telemedicine research, including publications in the postpandemic period. Latent Dirichlet allocation (LDA) topic modeling, which utilizes diverse scientific approaches to determine the optimal number of topics, was used. Thus, this study attempts to uncover the underlying themes and evolution of telemedicine research, providing valuable insights for the expanding telemedicine literature. In addition, it offers a unique perspective on the trends and directions of telemedicine research in the context of continually evolving health care.

Methods

LDA topic modeling was used in this study. Data collection, preprocessing, and analysis for the modeling were performed on the Konstanz Information Miner (KNIME) Analytics Platform ver. 4.7.1. The KNIME Analytics Platform is an open-source tool that does not impose limitations on the service type or data quantity (http://www.knime.com/). KNIME was chosen for its capacity to deal with big data, ease of using unstructured data analysis, and preset workflows for machine-learning algorithms, all of which fully meet the requirements of this study.²¹ Python 3.10.12 was used as a supplementary tool to choose the optimal number of topics.

DATA

The search focused on articles related to telemedicine, using the Medical Subject Headings (MeSH) term “Telemedicine” in the query. To ensure the relevance of the retrieved articles, additional search terms such as “eHealth,” “mHealth,” “Telehealth,” and “Telecare” were included^22,23 (Supplementary Appendix SA1). Advanced search was performed on the PubMed website under specific language criteria (English only). The query was then transferred to the “Document Grabber” node in KNIME to retrieve literature from PubMed. Upon excluding studies without abstracts (n = 2) and removing duplicates (n = 5,294), the final sample comprised 56,445 studies.

PREPROCESSING

We followed the standard procedures of natural language processing to preprocess the textual data for topic modeling, which include lemmatization, removal of punctuation, and numerical characters, filtering out words with less than three characters, elimination of common English stop words, and conversion of all words to lowercase. To capture the importance of words within a corpus, we used the term frequency–inverse document frequency (TF–IDF) approach.²⁴ TF–IDF measures the relative significance of a term in a document by considering its frequency within the document and rarity across the entire corpus. Upon calculating TF–IDF scores for each word, we thoroughly reviewed the list and removed less meaningful words (e.g., study, telemedicine, patient, and care) with low TF–IDF values by using a threshold of 0.054. This process enabled us to focus on words with a higher discriminative power to effectively capture the unique characteristics of each document.

TOPIC MODELING

LDA topic modeling is a probabilistic algorithm that operates in an unsupervised manner to identify K topics in a large dataset.¹² The topics are represented by the N most relevant keywords, and K and N are specified by the user as input parameters.²⁵ That is, the algorithm models the documents in the dataset as a mixture proportion (β) of hidden topics, where each topic is characterized by a Dirichlet distribution (α) over a fixed vocabulary. The K topics are considered “latent” as they are not directly observed and must be inferred based on their constituent N terms. Scholars have utilized topic modeling to identify concealed themes in a specific area of interest and recommended future research topics.^14
–16 In this study, we used the parallel LDA topic modeling technique.^26,27

CHOICE OF OPTIMAL K FOR LDA TOPIC MODELING

Determining the optimal number of topics (K) is crucial in LDA topic modeling because it directly affects the quality and interpretability of the resulting latent semantic structures. Insufficient topics result in broad and ambiguous themes, while an excessive number of topics leads to overlapping and redundancy. Therefore, a rigorous method is required to extract subjective topics. To determine the optimal number of topics for topic modeling, we used several evaluation metrics and techniques.

First, we utilized perplexity as a document coverage measure of the accuracy of the model in predicting unobserved data. It was measured via a two-step process involving a wide range of topics (2–80), and subsequently, a smaller range of topics (3–13). This phase enabled the determination of the K value of the lowest perplexity, which indicated that the topic model improved the coverage of documents by more precisely matching the words in the withheld documents to the expected distribution of words in the documents.²⁸

Following that, we assessed the coherence of the generated topics using the Python Gensim library.²⁹ Coherence measures the semantic coherence and interpretability of topics by evaluating the relevance and similarity between words in each topic. Higher coherence scores indicate more coherent and interpretable topics. We used the elbow method to further validate our choice of optimal number of topics. Based on the premise that the optimal number of topics is similar to the number of clusters in k-means clustering,³⁰ k-means clustering was performed on the corpus using a range of K. The within-cluster sum of squared error (SSE), which quantifies the aggregate distance between each document and its cluster center, was computed for each K. These SSE values were then visualized, which enabled the identification of the optimal number of clusters characterized by a pronounced decrease in SSE and a distinct angle in the graphical plot.

By integrating these evaluation methods, we could systematically identify the optimal number of topics that provided the best balance between model performance and topic interpretability and capture the underlying themes and patterns present in the analyzed telemedicine research articles. Figure 1 illustrates the sequence of operations executed in KNIME to accomplish topic modeling.

Fig. 1.

Topic modeling process.

Results

Based on the findings related to the 56,445 studies analyzed, the number of publications has progressively increasing since 1975, with a notable upsurge observed from 2011. Moreover, ∼4,000 studies were published in 2019. Furthermore, since 2020, a significant upward trajectory can be observed with both 2021 and 2022 exhibiting a more than twofold increase in the number of publications compared to 2019 (Fig. 2).

Fig. 2.

Number of yearly publications.

The perplexity, coherence, and elbow methods were used to determine the optimal number of topics for topic modeling. Figure 3 illustrates the results of the evaluation techniques. For the perplexity metric, the lowest value was obtained when K = 5. However, considering the coherence score, this configuration ranked the third lowest, suggesting that it may not be the most appropriate choice for topic modeling. Alternatively, when K = 7, we obtained a perplexity score of −6.987, which ranked second lowest, along with a coherence score of 0.469, the highest among all tested configurations.

Fig. 3.

Evaluation of optimal number of topics using perplexity and coherence (A), and the elbow method (B).

These results indicate that the selection of seven topics yielded more coherent and interpretable results, making it the preferred choice for the topic modeling analysis. In addition, during the elbow method analysis, a sharp decrease in the SSE was observed as the number of clusters transitioned from five to seven. This significant decrease reinforced our decision to choose seven topics for topic modeling.

Table 1 depicts the word clouds and top 15 keywords representing each topic identified via LDA topic modeling. The themes were identified through rigorous discussion and review by a panel of two researchers and a telemedicine expert. Among the seven identified themes, the theme of “Patient satisfaction with telemedicine services” (Topic 5) accounted for the largest proportion of studies with 21.38%. It was closely followed by “Perspectives and challenges in the community” (Topic 2), which accounted for 17.95% of the articles (Supplementary Appendix SA2).

Table 1.

Topics and Keywords Obtained from Latent Dirichlet Allocation Topic Modeling

Topic 1: Remote monitoring systems for patients with heart failure	Topic 2: Perspectives and challenges in the community

Topic 3: Smartphone apps for health promotion and disease prevention in adults	Topic 4: Telerehabilitation for community-dwelling patients and caregivers

Topic 5: Patient satisfaction with telemedicine services	Topic 6: Feasibility and effectiveness of self-management for adults with diabetes

Topic 7: Telemedicine for diagnosis and screening

Based on the provided keywords and their weights, we derived the theme for Topic 1 as “Remote monitoring systems for patients with heart failure.” The prominence of keywords such as “system,” “heart,” “failure,” “management,” “telemonitoring,” and “sensor” indicates a focus on employing remote monitoring systems for the management of heart failure in studies belonging to Topic 1. “Home” indicates emphasis on care and support within the patient's residential setting. The keywords “design,” “data,” “application,” “development,” “evaluation” and “analysis” further underscore the development of systems, collection and analysis of patient data, and evaluation of outcomes pertaining to remote monitoring.

The second theme, titled “Perspectives and challenges in the community,” emerged from the analysis. The prominence of keywords such as “challenge,” “research,” “perspective,” “implementation,” “e-health,” “service,” and “barrier” indicates a strong focus on investigating the implementation of telemedicine services and understanding the associated challenges and barriers. Furthermore, the presence of keywords such as “country,” “community,” “education,” and “child” suggests the wide adoption of telemedicine as a national strategy within communities. Moreover, the inclusion of keywords such as “model,” “development,” and “lesson” indicates the development of various models and the accumulation of valuable insights for telemedicine implementation in the community.

The third theme identified via topic modeling was titled “Smartphone apps for health promotion and disease prevention in adults.” The prominent keywords “app,” “adult,” “e-health,” “hiv,” “behavior,” “weight,” “prevention,” “change,” and “smartphone” suggest research focus on the utilization of smartphone applications to promote health behaviors and prevent diseases among adults. The keywords “development,” “analysis,” and “randomized” indicate a significant emphasis on software application development and analyzing their effects. The presence of “literacy” and “user” as keywords indicates the importance of exploring health literacy and the experiences of smartphone application users in the context of telemedicine. The fourth theme was titled “Telerehabilitation for community-dwelling patients and caregivers.”

The keywords identified in Topic 4, including “stroke,” “caregiver,” “surgery,” “home,” “adult,” “injury,” and “dementia,” collectively stress on community-dwelling patients with stroke, dementia, and postoperative status as well as their caregivers. These keywords reflected specific patient populations of interest and highlighted the relevance of telemedicine in addressing rehabilitation needs. Furthermore, the presence of keywords such as “rehabilitation,” “feasibility,” “telerehabilitation,” “pilot,” “evaluation,” “exercise,” “training,” and “therapy” demonstrates that investigations within this theme aim to provide rehabilitation services such as exercise, training, and therapeutic interventions through telemedicine, while also assessing the feasibility and effectiveness of such approaches.

The fifth theme extracted from topic modeling was titled “Patient satisfaction with telemedicine services.” Prominent keywords such as “service,” “experience,” “survey,” “emergency,” “consultation,” “access,” and “satisfaction” emphasize on the assessment of patient satisfaction with their experience of telemedicine services. The presence of keywords such as “clinic,” “hospital,” “telephone,” “center,” “medical,” and “department” suggests that studies within this theme examined patient satisfaction with telemedicine experiences (e.g., telephone consultations) in a variety of clinical settings.

The sixth theme was titled “Feasibility and effectiveness of self-management for adults with diabetes.” The keywords “diabetes,” “management,” “therapy,” “type,” “disorder,” and “adult” indicate the relevance of this topic to the management of chronic diseases, particularly diabetes. The presence of keywords such as “self-management,” “effectiveness,” “adherence,” “feasibility,” “randomized,” “pilot,” and “meta-analysis” indicates research focus on exploring the effectiveness of self-management interventions using telemedicine and assessing their feasibility. Moreover, keywords such as “pain” and “depression” indicate that studies within this theme examined the impact of comorbid symptoms commonly associated with diabetes.

The seventh theme was titled “Telemedicine for diagnosis and screening.” This theme focused on assessing the precision of telemedicine in the context of diagnosis and screening. The prominence of keywords such as “system,” “diagnosis,” “screening,” “retinopathy,” “application,” “teleradiology,” “teledermatology,” and “detection” emphasizes utilizing telemedicine for diagnosis and screening, with a spotlight on ophthalmology, radiology, and dermatology. In addition, the inclusion of keywords such as “image,” “evaluation,” “accuracy,” and “quality” highlights the importance of image quality in ensuring accurate diagnosis and screenings via telemedicine.

To gain valuable insights into the relative interest and research activity associated with the seven topics over time, we analyzed the numbers (Fig. 4) and percentages (Fig. 5) of publications by year for those seven topics. Figure 4 depicts a line graph with the number of publications per year for each topic. The interpretation of the graph is organized in the descending order of number of publications since 2020. Topic 5, “Patient satisfaction,” had a consistently high volume of publications throughout the analysis period, indicating sustained interest and research efforts. Publications markedly surged around 2018, which continued until 2021. Among the seven topics, Topic 5 exhibited a downtrend in publications from 2021 to 2022. Conversely, Topic 2, “Perspectives and challenges,” consistently ranked among the top three for the number of publications annually and has held the second position since 2020.

Fig. 4.

Research trends by year per topic: Publication numbers.

Fig. 5.

Research trends by year per topic: Publication percentage.

Initially, Topic 3, “Smartphone apps,” had the least number of publications until 2012. However, since 2013, the number of publications based on this topic increased tremendously, which resulted in Topic 3 occupying the top position from 2016 to 2019. Although Topics 3 and 4, “Telerehabilitation,” have reversed their rankings since 2012, Topics 3, 4, and 6, “Feasibility and effectiveness,” exhibited similar trends characterized by a gradual increase in publications, with a notable spike in the early 2020s. Since 2019, Topic 7, “Telemedicine for diagnosis and screening,” consistently exhibited the smallest publication volume, followed by Topic 1, “Remote monitoring systems.” These topics demonstrated similar publication trends over time. Topic 1 garnered minimal interest between 2019 and 2020, as indicated by the marginal change in the number of publications during this period. Topic 7 was the most dominant topic in the mid-to the late 1990s.

Figure 5 presents the research trends of the seven telemedicine topics expressed as a percentage of articles published each year. The literature collection spans from 1975, whereas the analysis focuses on data from 1993, when all the topics emerged. Among the identified themes, Topic 3 “Smartphone apps” and Topic 7 “Telemedicine for diagnosis and screening” demonstrate the most significant changes in their share of total research. Topic 3 “Smartphone apps” experienced a substantial surge in publications between 2012 and 2013, reaching ∼22% of all research in 2019 before experiencing a subsequent decline. In contrast, Topic 7 “Telemedicine for diagnosis and screening” showed active research interest until around 2003, after which its share of publications gradually decreased.

Topic 5 “Patient satisfaction” consistently garnered significant interest throughout the analysis period, accounting for ∼10–30% of the total publication volume. Notably, the share of the total number of studies sharply increased from 2019, accounting for >25% of the studies from 2020 onward. Topic 1 “Remote monitoring systems” displayed a notable proportion of research from 2000 to 2014, with a gradual decline in the subsequent years. Topic 6 “Feasibility and effectiveness” followed a trend similar to that of Topic 3, with a steady increase in studies until a decline in 2019. Meanwhile, research on Topic 2 “Perspectives and challenges” and Topic 4 “Telerehabilitation” remained stable, with a few intermittent spikes over the analysis period.

Discussion

The analysis of publication trends during the literature survey period revealed a remarkable surge in the total number of publications since 2020. This increase indicates the emergence of telemedicine, which can be attributed to the growing demand for telemedicine services during the COVID-19 pandemic. In response to the pandemic, governments, including those of the United States and South Korea, implemented policies to expand telemedicine services to prevent the spread of the virus among communities.³¹ For example, in South Korea, as of February 24, 2020, teleconsultations and teleprescriptions by physicians via telephone were temporarily authorized.³²

In the United States, telemedicine policies have changed since the U.S. Department of Health and Human Services declared a public health emergency on January 31, 2020. These included expanding telemedicine services for Medicare beneficiaries, reducing or eliminating cost sharing, and removing geographic restrictions.³³ Government initiatives and guidelines for infectious disease management have led to a global increase in the adoption of telemedicine, which in turn has escalated research endeavors across various telemedicine topics.

Moreover, the rapidly growing population of older adults and the shift in focus from facility to community-based care have contributed to the expansion of telemedicine. Increasing health care costs for the aging population have spurred a proactive search for viable approaches to effectively maintain, manage, and improve health status, which is considered a promising technology-driven solution.¹⁸ However, despite the growing support for “Aging in place” in telemedicine, evidence confirming the effectiveness of telemedicine in supporting “Aging in place” is not strong. A systematic review by Ollevier et al.³⁴ proposed circumstantial evidence confirming the effectiveness of telemedicine in facilitating “Aging in place.” Several studies have explored the perceptions, barriers, and constraints related to the proliferation of telemedicine, which frequently focuses on small, specific populations.^35,36 This highlights the need for more robust research to understand the impact and effectiveness of telemedicine in supporting aging populations.

We analyzed the annual variations in the number of published articles under each topic and their proportion in the total literature collected over the study period. Topics concerning patient satisfaction, perspectives, and challenges in telemedicine have consistently garnered significant research interest. In this study, these two topics (“Satisfaction” and “Perspectives and challenges”) accounted for 39.33% of the corpus. Studies in these areas have examined the perceptions of both health care providers and recipients regarding the integration of novel science and technology into traditional medical practices.^37,38 Furthermore, they proposed more efficient approaches to improve the impact of telemedicine.^37,38

Desilva and Vaidya³⁷ recommended the frequent delivery of interventions over an extended period to maximize the difference in effectiveness between the treatment and control groups. Moreover, they proposed the integration of diverse technological tools, such as phone and videoconferencing into the research process.³⁷ Aashima et al.³⁸ emphasized the importance of guaranteeing telemedicine accessibility to individuals from different socioeconomic backgrounds. Furthermore, they stressed the need for scholars to address the barriers associated with physical examinations and technical problems that continuously hinder the broader adoption of telemedicine.³⁸ Research on these topics can be expanded by considering a diverse range of telehealth users, including patients, caregivers, and various health care providers, each with different backgrounds in terms of health status and geographic location. Furthermore, given the developing field of telemedicine with growing subcategories and emerging technologies, ongoing research in this area is warranted.

Although research interest in several topics has remained consistent, we observed a decline in studies related to remote monitoring systems, diagnosis, and screening. This decrease can be attributed to challenges faced by researchers in these domains, including the requirement for specialized research teams and high research costs.^39,40 Moreover, the widespread adoption of video technology by the public has contributed significantly to the rapid growth of telemedicine applications. This rise in video-based telemedicine solutions has overshadowed the initial focus on remote patient monitoring (RPM) studies, where interventions were predominantly based on collecting data, such as weight, blood glucose, and blood pressure, remotely to detect changes in the patients' health status and provide necessary interventions. Evidence-based studies have shown that the integration of video consultation and remote monitoring technologies improves health care outcomes and patient satisfaction.⁴¹

In addition, we observed a surge in the interest of implementing telemedicine in rehabilitation medicine and psychiatry, where telerehabilitation and telepsychiatry are becoming instrumental to enhancing the physical and mental wellbeing of community residents. As rehabilitation patients with stroke, postsurgery patients, or psychiatric patients frequently require continuous sessions and support, telemedicine has gained attraction. Telemedicine offers the potential to deliver professional sessions remotely, thereby facilitating access to care, regardless of the patient's location, such as rural or underserved areas.^42
–44 In fact, systematic reviews of the literature on telerehabilitation interventions for stroke and heart failure survivors have revealed substantial improvements in both the physical and mental health of participants within the intervention groups.^45,46

Developments in information and communication technologies, such as the internet of things (IoT), have helped realize other avenues in real-time remote monitoring research.⁴⁷ These findings demonstrate the transformative potential of video- or IoT-based telemedicine technologies to complement traditional RPM approaches. These integration technologies not only enhance real-time patient–provider interactions but also offer a more comprehensive and personalized approach to health care delivery. Similar trends in telemedicine were highlighted by Virginia Anikwe et al.⁴⁸ The application of telemonitoring using mobile and wearable sensors has demonstrated its potential across diverse domains, including sleep, falls, anxiety, Parkinson's disease, and dementia, which have led to their prevention, intervention, diagnosis, and treatment strategies.⁴⁸ However, this study emphasizes a system architecture methodology that encompasses monitoring, data collection, data transmission, data integration, feature extraction, data preprocessing, and data analysis as crucial precursors to enable efficient diagnostic, preventive, interventional, and treatment approaches.⁴⁸

The exponential growth in smartphone usage and the rapid advancements in related technologies have had a profound impact on the health care landscape, giving rise to an expanding health care market utilizing these technologies. This phenomenon is reflected in the substantial focus on mHealth, which particularly concerns the development, effectiveness, and potential applications of smartphone applications in diagnosis, education, treatment, data collection, health monitoring, and communication with health professionals.^49,50

Telemedicine significantly grew during the COVID-19 pandemic; however, sustaining its effectiveness in the postpandemic era is crucial. In response, subsequent to the declaration of the endemic, the U.S. Department of Health and Human Services, announced prospective policies in May 2023 to maintain and expand telemedicine use.⁵¹ The extension of Medicare telemedicine access, regardless of location, was extended to December 31, 2024. In addition, the Centers for Medicare & Medicaid Services (CMS) offered a telemedicine toolkit and state-specific supplements to support telemedicine regulations. Addressing privacy concerns, during the COVID-19 pandemic, the CMS temporarily suspended penalties for noncompliance with the Health Insurance Portability and Accountability Act (HIPAA).

However, starting May 12, 2023, HIPAA compliance was re-enforced, accompanied by a 90-day transitional period to facilitate provider preparation, thus ensuring the continuity of telemedicine utilization. Federal funding for the Health Professions Licensure Portability Grant Program has been tripled, enabling health care providers to practice across state boundaries. Moreover, in 2021, $3.2 billion was allocated for broadband access and internet-enabled devices, particularly for the underprivileged population. Furthermore, an additional $65 billion has been allocated to broaden broadband accessibility, thereby reducing technological barriers.

Based on the insights from previous studies, we suggest nurturing a skilled workforce, empowering consumers in health care decisions, revising financing for broader implementation, standardizing telemedicine usage, revising relevant legislation (such as licensure), enhancing digital literacy, strengthening digital ecosystems for seamless integration, integrating telemedicine into routine health care, and developing robust telemedicine delivery models.^48,52
–54 Moreover, a novel integrated telemedicine paradigm tailored to specific fields should be developed and subsequently implemented in a community setting. It requires the integration of various investigations, including telemonitoring, telediagnosis, tele-education, teleassistance, and teletherapy, which should result in a comprehensive telemedicine blueprint. In addition, an extensive examination of multiple aspects for all stakeholders, including health outcomes, satisfaction, and cost-effectiveness, is required within this complex framework.

Limitations

This study has several limitations that should be considered when interpreting the results. First, the dataset did not encompass all articles published in 2023 at the time of the survey, which may have restricted the accuracy of representation of the telemedicine publication trends between 2022 and 2023. Second, LDA topic modeling as a probabilistic model treats words as independent tokens without considering their context or semantic meaning. Despite our efforts to identify keywords using TF–IDF, unrelated words may have been grouped under the same topic.

Third, while the approach involved the use of the LDA topic modeling algorithm to carefully categorize the collated telemedicine literature into seven topics that were cross-referenced via a secondary validation process, some potential subtopics associated with each main topic may not have been explicitly addressed. Although the primary objective of this study was to provide a comprehensive overview of the general trends in telemedicine research, further investigation is required in subsequent research efforts to comprehensively explore subtopics within each major theme at a granular level. Finally, this study has an inherent language bias—only in English—in the data collection process, which may have diluted the comprehensiveness of the findings. To mitigate this limitation, future research should extend the scope by incorporating telemedicine literature written in non-English languages, possibly utilizing artificial intelligence.

Conclusions

This study offers valuable insights into the diverse landscape of telemedicine by highlighting clear telemedicine research trends from screening to therapy, and from RPM to video and mobile phone usage. Moreover, we observed the dynamic responses of research topics to policy and context shifts, indicating a broad interest in expanding telemedicine applications in the community. Our findings underscore the importance of exploring telemedicine to drive transformative changes in health care and enhance patient outcomes. This presents an opportunity to address health disparities in underprivileged populations, mitigate the centralization of health care in an aging society, and expand access to quality services.

Footnotes

Authors' Contributions

A.K.: Conceptualization, Methodology, Software, Validation, Formal analysis, Investigation, Visualization, Writing—Original draft, and Writing—Review and Editing. S.C.: Software, Writing—Original draft, and Writing—Review and Editing. K.W.: Conceptualization, Methodology, Software, Investigation, Writing—Review and Editing, and Supervision.

Disclosure Statement

No competing financial interests exist.

Funding Information

No funding was received for this article.

Supplementary Material

Supplementary Appendix SA1

Supplementary Appendix SA2

References

World Health Organization. WHO-ITU Global Standard for Accessibility of Telehealth Services. World Health Organization and International Telecommunication Union: Geneva; 2022.

Al Meslamani

, Aldulaymi

, El Sharu

, et al. The patterns and determinants of telemedicine use during the COVID-19 crisis: A nationwide study. J Am Pharm Assoc, 2022; 62(6):1778–1785; doi: 10.1016/j.japh.2022.05.020

Awotunde

, Adeniyi

, Ogundokun

, et al. MIoT-based Big Data analytics architecture, opportunities and challenges for enhanced telemedicine systems. In: Enhanced Telemedicine and e-Health: Advanced IoT Enabled Soft Computing Framework. (Marques G, Bhoi AK, de la Torre Díez I, et al. eds.) Springer International Publishing: Cham; 2021; pp. 199–220.

Gao

, Fan

, Chen

, et al. Telemedicine is becoming an increasingly popular way to resolve the unequal distribution of healthcare resources: Evidence from China. Front Public Health, 2022; 10; doi: 10.3389/fpubh.2022.916303

Alghamdi

, Gashgari

, Househ

. A systematic review of mobile health technology use in developing countries. Enabling Health Informatics Appl, 2015; 213:223–226.

Choi

, Cho

, Woo

. mHealth approaches in managing skin cancer: Systematic review of evidence-based research using integrative mapping. JMIR mHealth uHealth, 2018; 6(8):e8554; doi: 10.2196/mhealth.8554

Lewis

, Ray

, Liaw

S-T

. Recent worldwide developments in eHealth and mHealth to more effectively manage cancer and other chronic diseases—A systematic review. Yearb Med Inform, 2016; 25(01):93–108; doi: 10.15265/IY-2016-020

Kirakalaprathapan

, Oremus

. Efficacy of telehealth in integrated chronic disease management for older, multimorbid adults with heart failure: A systematic review. Int J Med Inform, 2022; 162:104756; doi: 10.1016/j.ijmedinf.2022.104756

Snoswell

, Stringer

, Taylor

, et al. An overview of the effect of telehealth on mortality: A systematic review of meta-analyses. J Telemed Telecare, 2023; 29(9):659–668; doi: 10.1177/1357633X211023700

10.

Dawes

, Lin

, Varghese

, et al. Mobile health technology for remote home monitoring after surgery: A meta-analysis. Br J Surg, 2021; 108(11):1304–1314; doi: 10.1093/bjs/znab323

11.

Higgins

JPT

, Thomas

, Chandler

, et al. (eds). Cochrane Handbook for Systematic Reviews of Interventions. Second ed. John Wiley & Sons: Chichester, UK; 2019.

12.

Blei

DM.

Probabilistic topic models. Commun ACM, 2012; 55(4):77–84; doi: 10.1145/2133806.2133826

13.

Kherwa

, Bansal

. Topic modeling: A comprehensive review. EAI Endorsed Trans Scalable Inf Syst, 2020; 7(24):1–16; doi: 10.4108/eai.13-7-2018.159623

14.

Baird

, Xia

, Cheng

. Consumer perceptions of telehealth for mental health or substance abuse: A Twitter-based topic modeling analysis. JAMIA Open, 2022; 5(2):ooac028; doi: 10.1093/jamiaopen/ooac028/6574721

15.

Nitiema

Telehealth before and during the COVID-19 pandemic: Analysis of health care workers' opinions. J Med Internet Res, 2022; 24(2):e29519; doi: 10.2196/29519

16.

Shan

, Ji

, Xie

, et al. Public trust in artificial intelligence applications in mental health care: Topic modeling analysis. JMIR Hum Factors, 2022; 9(4):e38799; doi: 10.2196/38799

17.

National Center for Biotechnology Information. Telemedicine articles in PubMed. Available from: https://pubmed.ncbi.nlm.nih.gov./?term=telemedicine%5BmeSH+Terms%5D&ac=no&sort=relevance [Last accessed: June 5, 2023 ].

18.

Doarn

, Yellowlees

, Jeffries

, et al. Societal drivers in the applications of telehealth. Telemed J E Health, 2008; 998–1002; doi: 10.1089/tmj.2008.0111

19.

Dastani

, Atarodi

. Health information technology during the COVID-19 epidemic: A review via text mining. Online J Public Health Inform, 2022; 14(1):e3; doi: 10.5210/ojphi.v14i1.11090

20.

Park

, Park

. Identifying the knowledge structure and trends of nursing informatics: A text network analysis. Comput Inform Nurs, 2023; 41(1):8–17; doi: 10.1097/CIN.0000000000000919

21.

Ihrmark

, Tyrkkö

. Learning text analytics without coding? An introduction to KNIME. Golub K, Colavizza G, Kamal AM, et al. eds. Educ Inf, 2023; 39(2):121–137; doi: 10.3233/EFI-230027.

22.

Preaux

, Casadesús

, Bernardo

. A conceptual model to evaluate service quality of direct-to-consumer telemedicine consultation from patient perspective. Telemed J E Health, 2023; 29(2):156–171; doi: 10.1089/tmj.2022.0089

23.

Federal Communications Commission. Telehealth, Telemedicine, and Telecare: What's What? Available from: https://www.fcc.gov/general/telehealth-telemedicine-and-telecare-whats-what [Last accessed: April 22, 2023].

24.

Manning

, Raghavan

, Schütze

An Introduction to Information Retrieval. Cambridge University Press: Cambridge, England; 2009.

25.

Chauhan

, Shah

. Topic modeling using latent Dirichlet allocation: A survey. ACM Comput Surv, 2021; 54(7):1–35; doi: 10.1145/3462478

26.

Singhal

, Baxi

, Mago

. Synergy between public and private health care organizations during COVID-19 on Twitter: Sentiment and engagement analysis using forecasting models. JMIR Med Inform, 2022; 10(8):e37829; doi: 10.2196/37829

27.

Hoffman

, Bach

, Blei

Online learning for Latent Dirichlet Allocation. In: Proceedings of the 23rd International Conference on Neural Information Processing Systems. Curran Associates Inc.: Red Hook, New York, USA; 2010; pp. 856–864.

28.

Hofmann

Probabilistic latent semantic indexing. In: Proceedings of the 22nd annual international ACM SIGIR Conference on Research and Development in Information Retrieval, SIGIR 1999. ACM: New York, USA; 1999; pp. 50–57.

29.

Röder

, Both

, Hinneburg

Exploring the space of topic coherence measures. In: Proceedings of the Eighth ACM International Conference on Web Search and Data mining. ACM: Shanghai, China; 2015; pp. 399–408.

30.

Steinskog

, Therkelsen

, Gambäck

Twitter topic modeling by tweet aggregation. In: Proceedings of the 21st Nordic Conference on Computational Linguistics. Linköping University Electronic Press; Gothenburg, Sweden; 2017; pp. 77–86.

31.

Centers for Medicare and Medicaid Services. Medicare telemedicine health care provider fact sheet. 2020. Available from: https://www.cms.gov/newsroom/fact-sheets/medicare-telemedicine-health-care-provider-fact-sheet [Last accessed: August 12, 2023 ].

32.

Kim

, Kim

, Lee

, et al. COVID-19 case surge and telemedicine utilization in a Tertiary Hospital in Korea. Telemed J E Health, 2022; 28(5):666–674; doi: 10.1089/tmj.2021.0157

33.

Schofield

Regulatory and legislative issues on telehealth. Nutr Clin Pract, 2021; 36(4):729–738; doi: 10.1002/ncp.10740

34.

Ollevier

, Aguiar

, Palomino

, et al. How can technology support ageing in place in healthy older adults? A systematic review. Public Health Rev, 2020; 41(1): 26; doi: 10.1186/s40985-020-00143-4

35.

Bhatia

, Gilliam

, Aliberti

, et al. Older adults' perspectives on primary care telemedicine during the COVID-19 pandemic. J Am Geriatr Soc, 2022; 70(12):3480–3492; doi: 10.1111/jgs.18035

36.

Mao

, Tam

, Xu

, et al. Barriers to telemedicine video visits for older adults in independent living facilities: Mixed methods cross-sectional needs assessment. JMIR Aging, 2022; 5(2):e34326; doi: 10.2196/34326

37.

DeSilva

, Vaidya

. The application of telemedicine to pediatric obesity: Lessons from the past decade. Telemed J E Health, 2021; 27(2):159–166; doi: 10.1089/tmj.2019.0314

38.

Aashima, Nanda

, Sharma

. A review of patient satisfaction and experience with telemedicine: A virtual solution during and beyond COVID-19 pandemic. Telemed J E Health, 2021; 27(12):1325–1331; doi: 10.1089/tmj.2020.0570

39.

Dinesen

, Nonnecke

, Lindeman

, et al. Personalized telehealth in the future: A Global Research Agenda. J Med Internet Res, 2016; 18(3):e53; doi: 10.2196/jmir.5257

40.

Lustig

TA.

Board on Health Care Services, Institute of Medicine. Chapter 4. Challenges in Telehealth. In: The Role of Telehealth in an Evolving Health Care Environment: Workshop Summary. National Academies Press: Washington, DC, USA; 2012; pp. 17–30.

41.

Atilgan

, Onuk

, Köksal Coşkun

, et al. Remote patient monitoring after cardiac surgery: The utility of a novel telemedicine system. J Card Surg, 2021; 36(11):4226–4234; doi: 10.1111/jocs.15962

42.

Laver

, Lange

, George

, et al. Virtual reality for stroke rehabilitation. Cochrane Database Syst Rev, 2017;(11):CD008349; doi: 10.1002/14651858.CD008349.pub4

43.

James

HE.

Pediatric neurosurgery telemedicine clinics: A model to provide care to geographically underserved areas of the United States and its territories. J Neurosurg Pediatr, 2016; 18(6):753–757; doi: 10.3171/2016.6.PEDS16202

44.

Hall

, Danna

, Hoeft

, et al. Patient and clinician perspectives on two telemedicine approaches for treating patients with mental health disorders in underserved areas. J Am Board Fam Med, 2022; 35(3):465–474; doi: 10.3122/jabfm.2022.03.210377

45.

Sarfo

, Ulasavets

, Opare-Sem

, et al. Tele-rehabilitation after stroke: An updated systematic review of the literature. J Stroke Cerebrovasc Dis, 2018; 27(9):2306–2318; doi: 10.1016/j.jstrokecerebrovasdis.2018.05.013

46.

Cavalheiro

, Silva Cardoso

, Rocha

, et al. Effectiveness of tele-rehabilitation programs in heart failure: A systematic review and meta-analysis. Health Serv Insights, 2021; 14:11786329211021668; doi: 10.1177/11786329211021

47.

Ahmadi

, Arji

, Shahmoradi

, et al. The application of internet of things in healthcare: A systematic literature review and classification. Univ Access Inf Soc, 2019; 18:837–869; doi: 10.1007/s10209-018-0618-4

48.

Virginia Anikwe

, Friday Nweke

, Chukwu Ikegwu

, et al. Mobile and wearable sensors for data-driven health monitoring system: State-of-the-art and future prospect. Expert Syst Appl, 2022; 202:117362; doi: 10.1016/j.eswa.2022.117362

49.

Wood

, Thomas

, Budd

, et al. Taking connected mobile-health diagnostics of infectious diseases to the field. Nature, 2019; 566(7745):467–474; doi: 10.1038/s41586-019-0956-2

50.

Marshall

, Lewis

, Whittaker

mHealth Technologies in Developing Countries: A Feasibility Assessment and a Proposed Framework. University of Queensland: Herston, Australia; 2013; pp. 1–47.

51.

U.S. Department of Health and Human Services. HHS Fact Sheet: Telehealth Flexibilities and Resources and the COVID-19 Public Health Emergency. Available from: https://www.hhs.gov/about/news/2023/05/10/hhs-fact-sheet-telehealth-flexibilities-resources-covid-19-public-health-emergency.html [Last accessed: August 14, 2023 ].

52.

Thomas

, Haydon

, Mehrotra

, et al. Building on the momentum: Sustaining telehealth beyond COVID-19. J Telemed Telecare, 2022; 28(4):301–308; doi: 10.1177/1357633X20960638

53.

Ware

, Shah

, Ross

, et al. Challenges of telemonitoring programs for complex chronic conditions: Randomized controlled trial with an embedded qualitative study. J Med Internet Res, 2022; 24(1):e31754; doi: 10.2196/31754

54.

de Faria

, Kring

, Keable

, et al. Tele-psychiatry for college students: Challenges, opportunities, and lessons learned from the pandemic. J Med Regul, 2023; 109(2):21–28; doi: 10.30770/2572-1852-109.2.21

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.01 MB

0.06 MB

0.00 MB