Sage Journals: Discover world-class research

Abstract

Background

Obesity is the leading preventable cause of death worldwide. In recent years, artificial intelligence (AI) has shown the potential in the prevention, diagnosis, and management of obesity.

Objective

This study employ bibliometric analysis methods to systematically review applications within this field, identify key research hotspots, and provide novel insights for future research.

Methods

The Web of Science Core Collection database was employed as the source. All relevant publications were searched from the database's inception to June 2025. VOSviewer was used to analyze co-occurrence networks among countries, institutions, authors, and keywords. CiteSpace was utilized for keyword burst analysis and to detect emerging research fields, while the Bibliometrix R package was applied to identify influential papers, institutions, and authors.

Results

A total of 420 publications were included, which were affiliated with 806 institutions in 42 countries. These publications were authored by 2589 individuals and published in 122 journals. The United States had the highest number of publications and citations. Harvard University produced the greatest volume of papers. El Chaar, M. was the most influential author. Obesity Surgery published the most articles, while Journal of Medical Internet Research was cited most frequently. Based on keyword cluster analysis, this study identified three major research themes within the field of obesity management: telemedicine, machine learning applications, and perioperative management.

Conclusion

This study provides a comprehensive bibliometric analysis of AI in obesity management. Future research should focus on conducting multi-center, long-term, randomized controlled trials, obtaining high-quality longitudinal data, and formulating reasonable reimbursement policies and data security regulations to enhance the generalizability, effectiveness, and safety of AI applications. These findings are crucial for addressing the global burden of obesity.

Keywords

Artificial intelligence obesity management bibliometrics CiteSpace VOSviewer

Introduction

Obesity is defined as the abnormal or excessive accumulation of fat tissue, which may pose a significant risk to health.¹ The global prevalence of overweight and obesity had continued to rise between 1990 and 2021. China has the largest number of obesity adults (402 million), followed by India (180 million) and the United States (172 million).² By 2035, global figures for overweight and obesity are projected to reach 1.77 billion and 1.53 billion individuals, respectively.³ Beyond adults, childhood obesity rates are soaring globally. By 2030, 400 million children are projected to be overweight and 254 million obese.⁴ Obesity and overweight have become significant public health issues as they are associated with a variety of diseases and increased mortality.⁵ Obesity is closely related to multiple chronic diseases, significantly increasing the risk of type 2 diabetes, cardiovascular disease, cancer, and other non-communicable diseases.^6,7 Obesity also imposes a substantial economic burden, including health-care expenditures, the cost of ambulatory care, and indirect costs due to decreased productivity.^8,9 To address the increasing burden of obesity-related diseases, it is essential to incorporate the prevention and control of obesity into public health strategies. Traditional obesity management strategies primarily emphasize personal responsibility and dietary control, which may be difficult to sustain over the long term.¹⁰ Obesity is a chronic, relapsing disease driven by multiple genetic, psychological, behavioral, environmental, health status, and socioeconomic status factors.¹¹ Therefore, early diagnosis, risk factor prediction, and personalized obesity management interventions are essential to improve obesity treatment.

Artificial intelligence (AI) refers to computer systems designed to simulate human cognitive functions, capable of performing tasks such as pattern recognition, learning, and decision support.¹² In the field of healthcare, AI has become a transformative technology, demonstrating significant value in real-time physiological monitoring, disease risk prediction, individualized treatment, remote patient monitoring, and robot-assisted surgery.¹³ To address global public health challenges, AI is being widely applied in the prevention, diagnosis, and management of obesity.

As a key branch of AI, machine learning (ML) can develop multimodal prediction models to accurately identify individuals at high risk of obesity and their associated comorbidities.¹⁴ For example, ensemble learning algorithms (such as random forest, gradient boosting machines, and XGBoost) can facilitate the early identification of obesity risk and support timely intervention by analyzing anthropometric data, electronic health records, dietary patterns, and health datasets related to lifestyle behaviors.^15,16 Additionally, ML can identify and verify obesity-related biomarkers by integrating genomics, metabolomics, and microbiome data, thereby enabling clinicians to better understand obesity and develop more precise interventions for its management.¹⁷ Since the outbreak of COVID-19, digital health technologies, including telemedicine, mobile health (mHealth) applications, and wearable sensor technology, have been widely adopted in the field of obesity management.¹⁸ These tools enhance the effectiveness of obesity care by providing remote, real-time feedback and personalized guidance to patients through the monitoring of health data, including physical activity, dietary intake, and sleep quality.¹⁰ Overall, these technological advances underscore the crucial role of AI in disease identification, risk prediction, and personalized treatment planning, offering an opportunity to enhance the management of obesity.

Bibliometric analysis is a quantitative method used to visualize knowledge structures and identify research frontiers in specific fields.¹⁴ Specifically, by quantitatively analyzing the contributions of countries, institutions, authors, and journals, as well as the patterns of keywords and references, bibliometrics is able to identify shifts in research focus and emerging hot areas.¹⁹ Presently, the application of AI in obesity management has yielded substantial research outcomes.^20–22 However, no studies have yet employed bibliometric analysis methods to explore this interdisciplinary field. To address this gap, this study employs bibliometric analysis to systematically synthesize and map the applied research in this domain, identifying core research themes and current frontiers. The findings provide valuable insights for advancing the application of AI in obesity management.

Materials and methods

Data sources and search strategy

We conducted a comprehensive search of the Web of Science Core Collection (WoSCC) database, encompassing all literature from the database's inception to June 2025. The search formula was as follows: TS = (“artificial intelligence” OR “deep learning” OR “machine learning” OR “big data” OR “supervised learning” OR “supervised machine learning” OR “robot*” OR “assistant robot” OR “computer reasoning” OR “computer vision system” OR “telehealth” OR “telemedicine” OR “digital health tools” OR “digital monitoring” OR “health information technology”) AND TS = (“Obesity Management” OR “Obesity Management System” OR “Bariatrics” OR “Weight Reduction Programs” OR “Bariatric Surgery” OR “Weight Loss”).

Inclusion criteria

Involves AI technology.

Related to obesity management research.

Only English articles are included.

Published articles and reviews.

Exclusion criteria

Includes news articles, conference papers, abstracts, case reports, books, and retracted publications.

Articles that only include abstracts and no full-text content.

Publications with incomplete information or those irrelevant to the study topic.

To guarantee that the chosen publications were relevant to the study topic, two researchers independently screened each article. In the event of a disagreement, a third researcher was consulted to resolve the issue. Figure 1 illustrates the detailed screening process.

Figure 1.

The selection of literature and the flow diagram representing the research framework.

Data extraction and analysis

VOSviewer (v1.6.20), CiteSpace (v6.2.R.4), and the Bibliometrix R package in R language were used for bibliometric and visual analysis. VOSviewer is used to construct co-citation networks among countries, institutions, journals, and authors, as well as co-occurrence relationships among keywords. Clustering and visualization of these networks facilitate the rapid identification of research hotspots in the field.²³ Bibliometrix is an R-based bibliometric analysis tool capable of performing comprehensive scientific mapping analyses.²⁴ It was employed to analyze publication trends and identify the most influential papers, institutions, journals, and authors. To more comprehensively reflect scholars’ academic impact, we integrated three complementary citation-based metrics—the h-index, g-index, and m-index—into the analytical workflow implemented using the Bibliometrix R package. The h-index combines publication output with citation impact, thereby reducing the statistical bias caused by either a small number of highly cited papers or a large number of papers with few citations, and thus provides a robust measure of sustained academic influence.²⁵ The g-index extends this method by calculating the total number of citations of the author's most frequently cited papers, thereby better capturing the impact of outstanding contributions.²⁶ The m-index, defined as the h-index divided by the number of years since the author's first publication, standardizes research impact relative to career length and effectively mitigates the cumulative advantage associated with a longer research trajectory.²⁷

CiteSpace is a visualization tool for detecting the evolution of research themes in scientific literature.²⁸ In this study, CiteSpace was used to detect burst keywords, thereby identifying emerging research fronts and tracking the evolution of research themes.

Data preprocessing and standardization

A total of 420 publications from the Web of Science (WoS) core collection were included for analysis. All records were exported from WoS in the “Plain Text” format, and the option “Full Record and Cited References” was selected. Meanwhile, to support Bibliometrix analysis, the same dataset is imported through the convert2df() function of the R package bibliometrix and the following cleaning is carried out: (1) Delete duplicate records; (2) Create a synonym library file in plain text format (.txt). Each line contains a pair of terms separated by tabs: the first term is a substitute variant (for example, “AI”), and the second term is a retained standardized form (for example, “artificial intelligence”). This file is imported into VOSviewer. During the network construction process, VOSviewer automatically merges the specified variants into the corresponding unified terms. (3) Check the completeness of key fields such as DOI, publication year, and keywords.

Bibliometric analysis parameter settings

Co-occurrence networks of countries, institutions, authors, journals, and keywords were generated using VOSviewer (v1.6.20) with full counting. To ensure network clarity and analytical robustness, minimum inclusion thresholds were applied, with countries, institutions, and journals included if appearing in ≥2 publications, authors in ≥3 publications, and keywords in ≥5 publications. Networks were normalized using the “association strength” method, and Total Link Strength was used to assess the relative importance of each node. Temporal and burst analyses were performed in CiteSpace (v6.2.R4). The study period (2003–2025) was divided into annual time slices. Keyword burst detection was conducted using the default algorithm with a sigma threshold of 1.0 and a minimum burst duration of 2 years to identify significant emerging trends. To enhance network clarity, the Pathfinder pruning algorithm was applied to retain only the most salient co-occurrence links.

Results

Publication overview

A total of 1728 articles were retrieved from the WoSCC. After screening, 420 documents were included in the analysis (Figure 1). These documents were published in 122 academic journals, written by 2589 authors from 42 countries, and cited a total of 1,3326 references. The average number of co-authors per paper was 7.19, with an international collaboration rate of 25.27%. Additionally, the average citation count per paper was 22.2 (Figure 2(a)).

Figure 2.

Descriptive statistics and publication trends. (a) Summary of quantitative publication analysis. (b) Annual trend in research output from 2003 to 2025.

From 2003 to 2025, the annual output of papers showed a continuous upward trend. This trend can be divided into three stages: between 2003 and 2010, output was relatively low, with only about two papers published per year. From 2011 to 2017, the number of publications remained stable, averaging 10 per year. Subsequently, from 2018 to 2025, there was a sharp increase in scholarly output, with an average of 51 papers published annually. The decrease in the number of papers published in 2025 may be related to the data retrieval deadline of July 2025, resulting in some papers not yet included in the data retrieval (Figure 2(b)).

Distribution and collaboration networks among countries

These papers come from 42 countries. The United States had the most publications (n = 188, 44.8%), followed by China (n = 33, 8.3%) and Australia (n = 23, 5.5%). The United States had the highest total citation frequency (TC = 4896), followed by Australia (TC = 1097) and China (TC = 355). In addition, the United States had the highest number of internationally co-authored publications (multi-country paper (MCP) = 23) (Table 1). The collaboration network visualization reveals that the United States has close relationships with a variety of countries, including China, Canada, the United Kingdom, Australia, and Italy. These relationships are represented on the map by thicker connection lines (Figure 3).

Figure 3.

Network map of scientific cooperation among countries/regions.

Table 1.

Publication and citation profiles of leading countries.

Country	Articles, n	Articles, %	MCP	TC	TC-rank
USA	188	44.8	23	4896	1
China	35	8.3	8	355	3
Australia	23	5.5	4	1097	2
Italy	15	3.6	5	311	4
France	13	3.1	4	233	7
Spain	12	2.9	5	81	17
Canada	10	2.4	3	130	11
Korea	10	2.4	3	55	19
United Kingdom	10	2.4	0	149	10
Switzerland	8	1.9	5	97	15
Germany	7	1.7	2	298	5
Netherlands	7	1.7	0	167	9
Singapore	7	1.7	2	117	13
Iran	6	1.4	3	29	19
Sweden	6	1.4	3	105	14
Brazil	5	1.2	2	199	8
Greece	5	1.2	3	255	6
Japan	5	1.2	0	78	16
Belgium	4	1.0	1	123	12
Saudi Arabia	4	1.0	3	6	32

Notes: Article: Papers published solely by the corresponding author. Article %: Percentage of total published papers. MCP: Papers involving multiple countries; MCP_Ratio: Percentage of multi-country papers; TC: Total number of times the papers have been cited.

Distribution among institutions and collaboration networks

A total of 806 institutions participated in the publication of relevant articles. As shown in Figure 4(a), Harvard University had the highest number of publications (n = 41), followed by the Pennsylvania State University (n = 30) and the University of Texas system (n = 30). Notably, the top 10 institutions are all from the United States, demonstrating the country's leadership in the field. The institutional co-occurrence network (Figure 4(b)) highlights Harvard Medical School, the University of Illinois, and the University of Oxford as central nodes in the collaboration structure.

Figure 4.

Analysis of institutions. (a) Top 10 institutions ranked by article count. (b) Visualization map depicting collaboration among different institutions.

Authors and co-authors analysis

A total of 2589 authors participated in the study of AI in obesity management. EL CHAAR M ranked first in the total number of publications, h-index, g-index, and M-index (NP = 8, h-index = 7, g-index = 8, and m-index = 0.875) (Table 2). The second was FORMAN EM (NP = 8, h-index = 7, g-index = 8, m-index = 0.875). These publication metrics indicate their significant scholarly influence in the field. Notably, despite EL CHAAR M's high publication volume, their total citations (TC = 147) ranked only 58th. Figure 5(a) shows that all publications by EL CHAAR M were published between 2020 and 2023. Co-citation analysis may be biased towards the early literature with more extended accumulation periods, thereby obscuring recent research results.²⁹ The author co-occurrence network (Figure 5(b)) shows clusters of tight connections among the top 20 most productive authors, indicating active collaboration between this core research group.

Figure 5.

Author output and collaborative networks. (a) Authors’ production over time. (b) Visual map describing collaboration between authors.

Table 2.

Publication and citation profiles of high-impact authors.

Author	h-index	g-index	m-index	NP	NP-rank	TC	TC-rank
El Chaar, M	7	8	1	8	1	147	58
Forman, E.M.	7	8	0.875	8	2	167	40
Stoltzfus, J.	6	7	0.75	7	3	136	61
Goldstein, S.P.	5	5	0.714	5	10	150	57
Batsis, J.A.	4	4	0.444	4	12	108	112
Butryn, M.L.	4	4	0.571	4	13	134	64
Cao, Y.	4	5	0.571	5	7	100	129
Chew, H.S.J.	4	5	0.8	5	8	97	135
Juarascio, A.S.	4	4	0.571	4	15	124	85
Ottosson, J.	4	4	0.571	4	19	99	133
Zhang, F.Q.	4	4	0.571	4	23	124	86
Bergman, M.	3	3	0.429	3	26	61	265
Bindal, V.	3	3	0.25	3	27	56	286
Buchs, N.C.	3	3	0.2	3	28	122	87
Caiazzo, R.	3	3	0.6	3	29	42	444
Claros, L.	3	3	0.375	3	30	60	279
Collins, C.E.	3	3	0.188	3	31	53	301
Currin, M.	3	3	0.429	3	32	61	266
Dang, J.T.	3	5	0.6	5	9	49	325
Dumon, K.R.	3	3	0.3	3	33	37	517

Notes: h-index: An indicator that reflects both the volume of an author's publications and the overall impact of their citations. g-index: An extension of the h-index, this metric emphasizes the importance of highly cited papers by considering the total citations of the most frequently cited papers. The m-index is the h-index divided by the number of years from the author's first published paper, which represents the average annual research output. NP: Number of publications. NP Rank: Ranking based on the number of publications.TC: Total citations. TC Rank: Ranking based on total citations.

Contributions and collaborative networks of journals

A total of 122 journals published articles on the application of AI in obesity management. According to the number of publications (Table 3), the first-ranked Journal was Obesity Surgery (NP = 60), followed by Journal of Medical Internet Research (NP = 23) and JMIR mHealth and uHealth (NP = 21). Among the top 20 most published journals, JAMA Network Open had the highest Impact Factor (IF = 9.7). The most cited journal was Journal of Medical Internet Research (1427 citations), followed by JMIR mHealth and uHealth (1, 113 citations). Obesity Surgery (h-index = 18), JMIR mHealth and uHealth (h-index = 17), and Journal of Medical Internet Research (h-index = 14) ranked the top three.

Table 3.

Bibliometric indicators of high-impact journals.

Journal	h-index	IF 2024	JCR 2024	TP	TP-rank	TC	TC-rank
Obesity Surgery	18	3.1	1	60	1	1019	5
JMIR mHealth and uHealth	17	6.2	1	21	3	1113	2
Journal of Medical Internet Research	14	6	1	23	2	1427	1
Telemedicine and e-Health	12	3	2	17	6	385	6
Surgery for Obesity and Related Diseases	11	3.1	1	18	5	238	10
Journal of Telemedicine and Telecare	8	3.2	2	10	9	263	7
Nutrients	7	5	1	18	4	175	12
Translational Behavioral Medicine	7	3	2	9	10	239	9
Surgical Endoscopy and Other Interventional Techniques	6	2.7	1	15	7	242	8
Asian Journal of Surgery	5	3.8	1	13	8	50	44
Journal of Clinical Medicine	5	3	1	5	19	120	16
Obesity	5	4.7	1	7	12	109	18
Annals of Surgery	4	6.4	1	4	20	84	30
JAMA Network Open	4	9.7	1	4	26	101	22
JMIR Medical Informatics	4	3.8	2	5	18	127	15
Scientific Reports	4	3.9	1	7	13	106	21
BMC Public Health	3	3.6	1	4	21	92	25
BMJ Open	3	2.3	2	4	22	15	62
Clinical Pediatrics	3	0.7	4	3	30	68	35
Contemporary Clinical Trials	3	1.9	3	7	11	39	50

Notes: h-index: An indicator that reflects both the volume of an author's publications and the overall impact of their citations. IF: The average number of citations received in a given year by articles published in the journal during the preceding two years. JCR: Journal Citation Reports ranking, which shows the journal's percentile position in its subject category (Q1: top 25%, Q2: 25%-50%, Q3: 50%-75%, Q4: bottom 25%). NP: Number of publications; NP Ranking: Ranking based on the volume of published papers; TC: Total citations; TC Rank: Ranking based on total citations; mHealth: mobile health.

Analysis of the top 20 most cited papers

From the 420 Papers retrieved, we selected the 20 most cited papers (Table 4). The number of citations of these articles reflects their importance in the field. The most cited article (TC = 369) was published by Brickwood KJ in 2019. The article revealed wearable activity trackers as a tool for physical activity interventions, with the potential to increase participation in physical activity.³⁰ The second most cited article was Ernsting C, published in 2017. The article suggests that there are age and socioeconomic differences in mobile technology use, and researchers should consider the needs of older individuals and patients with low health literacy.³¹ Among the top 20 cited articles, the Journal of Medical Internet Research contributed six articles, while JMIR mHealth and uHealth and Obesity Surgery contributed two articles each.

Table 4.

Top 20 papers with the highest cited counts.

Paper	DOI	TC	TC per year
Brickwood KJ, 2019, JMIR mHealth and uHealth	10.2196/11819	369	52.71
Ernsting C, 2017, Journal of Medical Internet Research	10.2196/jmir.6838	253	28.11
Mumma SJ, 2016, Journal of Medical Internet Research	10.2196/jmir.5927	224	22.40
Noah B, 2018, npj Digital Medicine	10.1038/s41746-017-0002-4	172	21.50
Beleigoli AM, 2019, Journal of Medical Internet Research	10.2196/jmir.9609	166	23.71
Samaan JS, 2023, Obesity Surgery	10.1007/s11695-023-06603-5	164	54.67
Triantafyllidis AK, 2019, Journal of Medical Internet Research	10.2196/12286	155	22.14
Joiner KL, 2017, Preventive Medicine	10.1016/j.ypmed.2017.04.033	154	17.11
DeGregory KW, 2018, Obesity Reviews	10.1111/obr.12667	134	16.75
Wang YF, 2020, JMIR mHealth and uHealthUHEALTH	10.2196/15400	133	22.17
White MA, 2004, Obesity Research	10.1038/oby.2004.132	115	5.23
Lentferink AJ, 2017, Journal of Medical Internet Research	10.2196/jmir.7288	115	12.78
Kelly JT, 2016, American Journal of Clinical Nutrition	10.3945/ajcn.116.136333	111	11.10
Hammersley ML, 2016, Journal of Medical Internet Research	10.2196/jmir.5893	109	10.90
Sorgente A, 2017, Journal of Medical Internet Research	10.2196/jmir.6972	103	11.44
Hubens G, 2008, Surgical Endoscopy and Other	10.1007/s00464-007-9698-6	102	5.67
Naslund JA, 2015, Telemedicine and e-Health	10.1089/tmj.2014.0105	100	9.09
Debédat J, 2018, Diabetes Care	10.2337/dc18-0567	90	11.25
Angrisani L, 2024, Obesity Surgery	10.1007/s11695-024-07118-3	87	43.50
Baum A, 2017, The Lancet Diabetes & Endocrinology	10.1016/S2213-8587(17)30176-6	87	9.67

Notes: TC: Total Citation; ML: machine learning.

Keyword co-occurrence analysis

Keyword analysis can use bibliometric methods to reveal the knowledge structure of an academic field, thereby identifying core themes and emerging research trends.³² From the 1566 extracted keywords, we selected the 140 high-frequency keywords and visualized them as a co-occurrence network (Figure 6). To further explore research themes related to AI in obesity management, we conducted a cluster analysis of high-frequency keywords, identifying four distinct research clusters. The red cluster encompasses keywords such as “telemedicine,” “digital technology,” “obesity,” “physical activity,” “diet,” and “management”, highlighting the central role of mHealth technologies. The blue cluster encompasses terms such as “telehealth” “obesity”, “children”, “rural areas” and “COVID-19”, indicating that telemedicine provides an opportunity for obesity treatment in remote areas during the pandemic. The green cluster includes keywords like “weight loss surgery,” “gastric bypass surgery,” “robotics,” and “perioperative care”, reflecting the application of AI-driven technologies in perioperative care for obesity. Finally, the yellow cluster includes “artificial intelligence,” “machine learning,” “prediction,” “risk,” and “intervention,” emphasizing the role of ML in risk prediction and personalized interventions.

Figure 6.

Keyword co-occurrence network analysis.

Burst keyword analysis

Burst keywords refer to the significant increase in the citation of keywords within a specific time range, which are typically regarded as a key indicator for tracking the evolution of research topics and emerging research fields.³³ Figure 7 shows the 20 keywords with the highest burst intensity between 2003 and 2025. The green line represents the overall time span (2003–2025), while the red bars represent the duration of each keyword's burst activity. As shown in the figure, “learning curve” was the earliest burst keyword (2.66) and had the longest burst duration (2008–2015). “Machine learning” had the highest burst intensity (5.3). In the keyword mapping, “metabolic surgery” (2.37, 2023–2025), “weight management” (2.14, 2021–2025), and “machine learning” (5.3, 2023–2025) have gained increasing prominence in recent years, highlighting the current research trend.

Figure 7.

Top 20 keywords with the strongest citation bursts.

Discussion

Worldwide trends and publishing landscape

This study employed bibliometric methods to conduct a systematic review of research on AI in the management of obesity. A total of 420 papers, authored by 2589 researchers from 806 institutions in 42 countries and regions. Notably, the volume of research has increased sharply since 2018. This may be related to advances in computing power, broader access to health data, and increased research funding in digital health. These findings highlight the significant progress made in the application of AI in obesity management.

In terms of geographical distribution, the United States and China are the main contributors, highlighting their research capabilities in this interdisciplinary field. In addition, as a core hub of international cooperation, the United States maintains close partnerships with the Americas, Europe, and parts of Asia.

At the institutional level, Harvard University produced the most output, followed by Pennsylvania State University and the University of Texas. Collaborative network analysis reveals that Harvard Medical School, the University of Illinois and the University of Oxford, as core nodes, play a pivotal role in driving research collaboration within this field. The leading research institutions have demonstrated distinct research strengths that drive the continuous evolution of this field. Harvard University focuses extensively on the molecular mechanisms of obesity and large-scale epidemiological longitudinal studies to identify long-term health risks. Pennsylvania State University is prominent in investigating behavioral interventions and nutritional sciences, particularly regarding pediatric obesity. Concurrently, the University of Texas has made significant strides in clinical weight management and the metabolic impacts of pharmacotherapy. Together, these diverse research focuses provide a robust clinical and theoretical foundation for application AI and ML in obesity management.

Among influential authors, El Chaar, M. had the highest publication output and citation impact, as reflected by the h-index, g-index, and m-index. The analysis of the author's influence highlights the outstanding contributions of El Chaar, M. and his team. Their research focuses on the standardization of robotic-assisted metabolism and bariatric surgery.^34,35 By leveraging the MBSAQIP database, they conducted a critical assessment of surgical techniques and established the first improved Delphi consensus for robotic bariatric surgery.³⁶ Their work provides important references for the integration of AI-driven predictive models and intraoperative decision support systems and intraoperative decision support systems in optimizing the efficiency of the operating room. Collaborative network analysis revealed a high level of co-authorship among top researchers, forming a tightly connected core group in this field.

Journal analysis showed that Obesity Surgery published the largest number of articles, while the Journal of Medical Internet Research received the highest total citations. JAMA Network Open had the highest impact factor. These journals have played a pivotal role in disseminating AI research within the field of obesity management.

The most cited articles emphasized the role of digital health technologies—such as mobile applications, telemedicine, and wearable devices—in promoting weight management and modifying health behaviors. Notably, seven of the 20 most cited papers were published in the Journal of Medical Internet Research, and two others in its affiliated journals, JMIR mHealth and uHealth. This highlights the leading position and academic influence of the JMIR journal in the field of digital health and internet-based medicine.

Research hotspots

Keyword analysis showed that AI, ML, telemedicine, bariatric surgery, overweight, and digital health were the central nodes of the keyword co-occurrence network. These themes reflect the current research landscape and can be grouped into three interconnected focal points:

Applications of machine learning in prediction and diagnosis

In recent years, AI and ML have been widely applied in obesity prediction, risk factor identification, and supporting clinical decision-making for obese patients.³⁷ ML algorithms—such as k-nearest neighbor, random forest, logistic regression, support vector machine, decision trees, and gradient boosting classifiers—are employed to analyze patient clinical characteristics.³⁸ These techniques have shown excellent performance in processing large-scale genomic datasets and patient histories, enabling significant improvements in diagnostic accuracy and disease prediction.³⁹ For example, Yu-Chi Lee et al. used ML algorithms to predict obesity in participants by integrating omics and dietary information through the field of statistical nutrigenetics.⁴⁰ The results suggest that individuals respond differently to various treatments, depending on their individual genetic and epigenetic backgrounds, providing valuable insights into precision nutrition strategies for the prevention and treatment of obesity. Additionally, ML can overcome the limitations of body mass index (BMI) in identifying obesity and accurately classifying it. Seungjin Jeon designed a ML-based obesity classification framework, which uses 3D body scan data to evaluate obesity indicators, and compared its performance with traditional methods such as BMI and BIA.⁴¹ The results showed that the accuracy of ML models in prediction or classification tasks was significantly higher than that of traditional methods based on BMI and BIA. Furthermore, ML has demonstrated significant advantages in predicting the risk of obesity. Jinsong Du's team developed a visual obesity risk prediction system based on the XGBoost algorithm.⁴² The study demonstrates that the system has high predictive accuracy and good interpretability, enabling users to visually assess their obesity risk and identify priority Interventions areas.

Telemedicine and personalized weight management

In recent years, telemedicine has been increasingly used in obesity management.⁴³ Among them, mHealth—which includes mobile applications, chatbots, social media, and wearable devices—is increasingly transforming the behavioral Interventions model in weight management and the monitoring of diet and physical activity.⁴⁴ These digital interventions have significantly improved the efficiency, accessibility, quality, and cost-effectiveness of obesity management.⁴⁵

mHealth applications can track physical activity, food intake, and physiological parameters in real time and are valuable tools for personalized obesity management.⁴⁶ Narupong Likitwirawong et al. evaluated the role of a novel mobile or tablet application in reducing weight, promoting healthy eating behaviors, and improving quality of life in obese children.⁴⁷ The findings suggest that mobile apps can increase patient engagement in healthy eating.

In addition, AI-driven chatbots provide personalized, interactive, and on-demand health Interventions programs.⁴⁸ Annisa Ristya Rahmanti et al. designed SlimMe, a weight loss chatbot with artificial empathy incentives.⁴⁹ The use of AI chatbots for weight management can achieve interactive and participatory monitoring of calorie intake and energy expenditure.

During the COVID-19 pandemic, mHealth applications are increasingly expanding the accessibility of health-care services.⁵⁰ At the same time, concerns about face-to-face contact have accelerated the use of telemedicine.⁵¹ Latorre-Rodriguez et al. found that in-person visits were equally effective as video visits for weight loss.⁵² Telemedicine has also expanded access to obesity treatment in rural and underserved areas, improving health-care equity.⁵³ Studies have shown that personalized remote counseling is effective in reducing the rate of weight regain in rural community populations in the United States.⁵⁴ Moreover, the pandemic has further promoted the application of telemedicine in the management of childhood obesity. By providing online dietary guidance, exercise planning, and health monitoring, telemedicine can effectively enhance the self-management abilities of children and adolescents, promoting the development of positive health behaviors and lifestyle changes.⁵⁵

Bariatric surgery and clinically assisted decision-making

AI algorithms have been widely used in all stages of perioperative care for obese patients, from preoperative risk assessment and intraoperative management to postoperative complication prediction.⁵⁶ In the preoperative stage, AI helps improve preoperative risk assessment and identify risk factors associated with difficult intubation, such as obesity, obstructive sleep apnea, and hiatal hernia.⁵⁷ For example, Zohou et al. utilized six ML models to predict difficult preoperative tracheal intubation in obese patients, achieving high prediction accuracy.⁵⁸ In intraoperative management, AI may predict the distribution of propofol, estimate the operation time, and automatically identify the steps of laparoscopic sleeve gastrectomy, providing auxiliary support for clinical decision-making.⁵⁹ In addition, AI models play an important role in predicting postoperative complications and assessing long-term weight loss and metabolic improvement. An artificial neural network model developed by Taheri and his team, which integrates patient age, BMI, smoking history, comorbidities, laboratory measures, and imaging data, can accurately predict the occurrence of postoperative complications and thus prevent deterioration in health status.⁶⁰

Future directions and recommendations

AI and ML have shown great potential to predict obesity risk and provide personalized treatment.⁶¹ However, the application of ML models in obesity management still faces multiple challenges, including data quality, model interpretability, and data privacy issues.⁶² To address these issues, future research should prioritize high-quality longitudinal studies involving diverse populations to enhance the generalizability and robustness of AI models. In addition, the design of AI systems must emphasize the transparency and fairness of the algorithms, ensuring that patient privacy is protected. Moreover, current evidence on the long-term effectiveness of AI in improving health outcomes remains limited, which hinders the widespread adoption of ML in clinical practice.⁶³ Therefore, multi-center randomized controlled trials with long-term follow-up should be prioritized to evaluate the long-term effectiveness of AI-driven interventions. Such studies would provide a scientific basis for integrating AI into long-term obesity management strategies.

During the COVID-19 pandemic, telemedicine interventions have proven effective in improving the delivery and cost-effectiveness of obesity management. However, they still face challenges related to inconsistent reimbursement policies, inadequate digital infrastructure, and low levels of patient digital literacy.⁶⁴ Currently, health systems in many countries do not routinely reimbursed for remote obesity management, resulting in high out-of-pocket costs for patients, especially those in low-income and rural areas. To promote health equity and accessibility, governments should establish clear reimbursement guidelines for digital weight management programs. At the same time, the integration of digital health infrastructure into primary health-care systems is essential to support the use of remote monitoring tools and telehealth platforms. Furthermore, as AI tools are increasingly integrated into obesity care, training providers in the use of these tools will be critical to improving clinical outcomes.

Limitations

All data in this study were retrieved from the WoSCC, which may introduce database-related coverage bias. Although WoSCC covers over 12,400 high-impact journals⁶⁵ and provides the standardized metadata necessary for co-occurrence and collaborative network analysis, its inclusion criteria may show a systematic preference for Western publications. Therefore, the geographical distribution of United States institutions in this study may be affected by the database coverage. However, the citation influence of US-based papers and the frequency of international collaborations led by the United States suggest that these findings reflect actual research output and academic influence, rather than being merely artifacts resulting from database selection. In terms of language, the exclusion of non-English articles may introduce additional bias. Although English publications provide high-quality metadata necessary for bibliometric analysis, they may overlook important localized research. We acknowledge these confounding factors and recommend that future research incorporate multilingual sources and integrate complementary databases, such as Scopus and PubMed, to provide a more inclusive global perspective.

Conclusion

This study systematically analyzed the application of AI in obesity management using a bibliometric approach. The results show that the academic output in this field has steadily increased, and the United States maintains a leading position in research output and collaboration. Since the outbreak of the COVID-19 pandemic, there has been a marked increase in academic publications concerning telemedicine, reflecting the research community's growing interest in its potential to enhance the quality of obesity care, service efficiency, and accessibility, particularly in remote and underserved areas. ML can be employed for obesity prediction, risk assessment, and personalized interventions, thereby supporting clinical decision-making. However, the field still faces certain challenges, such as inadequate data quality, lack of algorithmic explainability, ethical and privacy risks, and the digital divide. In summary, this study systematically sorted out the research topics and frontier trends in this field, which provided valuable reference for researchers and policy makers.

Footnotes

ORCID iD

Ziping Hu

Author contributions

ZH and JW contributed to conceptualization, visualization, and writing—original draft. YG and JY contributed to writing—review & editing. MP and HP contributed to data curation and methodology. YX and NL contributed to formal analysis. HL contributed to supervision and project administration. All authors have read and approved the manuscript for publication.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: 2023 Sanming Project of Medicine in Shenzhen. Project name: The whole life cycle integrated Traditional Chinese and Western Medicine care team of academician Xu Guihua from Nanjing University of Chinese Medicine. Project No.: SZZYSM202311016.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

References

Roomy

Hussain

Behbehani

, et al. Therapeutic advances in obesity management: an overview of the therapeutic interventions. Front Endocrinol (Lausanne) 2024; 15: 1364503.

Gakidou

, et al. Global, regional, and national prevalence of adult overweight and obesity, 1990-2021, with forecasts to 2050: a forecasting study for the Global Burden of Disease Study 2021. Lancet (London, England) 2025; 405: 813–838.

Janić

Janež

El-Tanani

, et al. Obesity: recent advances and future perspectives. Biomedicines 2025; 13: 368.

Ling

Chen

Zahry

, et al. Economic burden of childhood overweight and obesity: a systematic review and meta-analysis. Obes Rev 2023; 24: e13535.

Sarma

Sockalingam

Dash

. Obesity as a multisystem disease: trends in obesity rates and obesity-related complications. Diabetes Obes Metab 2021; 23: 3–16.

Yanovski

. Approach to obesity treatment in primary care: a review. JAMA Intern Med 2024; 184: 818–829.

Battineni

Sagaro

Chintalapudi

, et al. Impact of obesity-induced inflammation on cardiovascular diseases (CVD). Int J Mol Sci 2021; 22: 4798.

Nagi

Ahmed

Rezq

MAA

, et al. Economic costs of obesity: a systematic review. Int J Obes 2024; 48: 33–43.

Huang

, et al. Global, regional, and national burden of type 2 diabetes mellitus caused by high BMI from 1990 to 2021, and forecasts to 2045: analysis from the global burden of disease study 2021. Front Public Health 2025; 13: 1515797.

10.

Hinchliffe

Capehorn

Bewick

, et al. The potential role of digital health in obesity care. Adv Ther 2022; 39: 4397–4412.

11.

Chávez-Manzanera

Vera-Zertuche

Kaufer-Horwitz

, et al. Mexican clinical practice guidelines for adult overweight and obesity management. Curr Obes Rep 2024; 13: 643–666.

12.

Petersson

Larsson

Nygren

, et al. Challenges to implementing artificial intelligence in healthcare: a qualitative interview study with healthcare leaders in Sweden. BMC Health Serv Res 2022; 22: 850.

13.

Wei

, et al. Innovation and challenges of artificial intelligence technology in personalized healthcare. Sci Rep 2024; 14: 18994.

14.

Chatterjee

Gerdes

Martinez

. Identification of risk factors associated with obesity and overweight-a machine learning overview. Sensors 2020; 20: 2734.

15.

Colmenarejo

. Machine learning models to predict childhood and adolescent obesity: a review. Nutrients 2020; 12: 2466.

16.

Niakan Kalhori

Najafi

Hasannejadasl

, et al. Artificial intelligence-enabled obesity prediction: a systematic review of cohort data analysis. Int J Med Inf 2025; 196: 105804.

17.

Awari

Kaushik

Kumar

, et al. Obesity biomarkers: exploring factors, ramification, machine learning, and AI-unveiling insights in health research. MedComm 2025; 6: e70169.

18.

Shen

Chen

Yue

, et al. Digital technology-based telemedicine for the COVID-19 pandemic. Front Med (Lausanne) 2021; 8: 646506.

19.

Zeng

Dong

Liu

, et al. The global research of artificial intelligence on inflammatory bowel disease: a bibliometric analysis. Digit Health 2025; 11: 20552076251326217.

20.

Masi

Risi

Biagi

, et al. Application of a machine learning technology in the definition of metabolically healthy and unhealthy status: a retrospective study of 2567 subjects suffering from obesity with or without metabolic syndrome. Nutrients 2022; 14: 373.

21.

Rigla

Martínez-Sarriegui

García-Sáez

, et al. Gestational diabetes management using smart mobile telemedicine. J Diabetes Sci Technol 2018; 12: 260–264.

22.

Kang

Zhou

Niranjan

, et al. Predicting operative time for metabolic and bariatric surgery using machine learning models: a retrospective observational study. Int J Surg 2024; 110: 1968–1974.

23.

van Eck

Waltman

. Software survey: VOSviewer, a computer program for bibliometric mapping. Scientometrics 2010; 84: 523–538.

24.

Elkoca

Güngörmüş

. Bibliometric study on geriatric care models: intellectual structure, prominent themes, and future-directing topics. Geriatr Nurs 2025; 65: 103452.

25.

Szeto

Presley

Maymone

MBC

, et al. Top authors in dermatology by h-index: a bibliometric analysis of 1980-2020. J Am Acad Dermatol 2021; 85: 1573–1579.

26.

Choudhri

Siddiqui

Khan

, et al. Understanding bibliometric parameters and analysis. Radiographics 2015; 35: 736–746.

27.

Mustafa

Rauf

Tanvir Afzal

. MRET: Modified Recursive Elimination Technique for ranking author assessment parameters. PLoS One 2024; 19: e0303105.

28.

Chen

. Citespace II: detecting and visualizing emerging trends and transient patterns in scientific literature. J Am Soc Inf Sci Technol 2006; 57: 359–377.

29.

Hassan

Duarte

. Bibliometric analysis: a few suggestions. Curr Probl Cardiol 2024; 49: 102640.

30.

Brickwood

Watson

O'Brien

, et al. Consumer-based wearable activity trackers increase physical activity participation: systematic review and meta-analysis. JMIR mHealth uHealth 2019; 7: e11819.

31.

Ernsting

Dombrowski

Oedekoven

, et al. Using smartphones and health apps to change and manage health behaviors: a population-based survey. J Med Internet Res 2017; 19: e101.

32.

Wang

Liang

Cao

, et al. Application of artificial intelligence in geriatric care: bibliometric analysis. J Med Internet Res 2023; 25: e46014.

33.

Pang

Peng

Zheng

, et al. Analysis of research hotspots in COVID-19 genomics based on citespace software: bibliometric analysis. Front Cell Infect Microbiol 2022; 12: 1060031.

34.

El Chaar

Michaud

Stoltzfus

, et al. Improving operating room efficiency of robotic-assisted metabolic and bariatric surgery through standardization. Obes Surg 2023; 33: 3411–3421.

35.

Bauerle

Mody

Estep

, et al. Current trends in the utilization of a robotic approach in the field of bariatric surgery. Obes Surg 2023; 33: 482–491.

36.

Faugno-Fusci

Perrone

Michaud

, et al. Outcomes of staple line reinforcement following robotic assisted sleeve gastrectomy based on MBSAQIP database. Obes Surg 2023; 33: 2662–2670.

37.

Zhou

Chen

Liu

. Applications of machine learning models to predict and prevent obesity: a mini-review. Front Nutr 2022; 9: 933130.

38.

Alsabhan

Alfadhly

. Effectiveness of machine learning models in diagnosis of heart disease: a comparative study. Sci Rep 2025; 15: 24568.

39.

Zahra

Al-Taher

Alquhaidan

, et al. The synergy of artificial intelligence and personalized medicine for the enhanced diagnosis, treatment, and prevention of disease. Drug Metabol Personal Ther 2024; 39: 47–58.

40.

Lee

Christensen

Parnell

, et al. Using machine learning to predict obesity based on genome-wide and epigenome-wide gene-gene and gene-diet interactions. Front Genet 2021; 12: 783845.

41.

Jeon

Kim

Yoon

, et al. Machine learning-based obesity classification considering 3D body scanner measurements. Sci Rep 2023; 13: 3299.

42.

Yang

Zeng

, et al. Visualization obesity risk prediction system based on machine learning. Sci Rep 2024; 14: 22424.

43.

Kupila

SKE

Joki

Suojanen

, et al. The effectiveness of eHealth interventions for weight loss and weight loss maintenance in adults with overweight or obesity: a systematic review of systematic reviews. Curr Obes Rep 2023; 12: 371–394.

44.

Pallan

Liu

, et al. The CHIRPY DRAGON intervention in preventing obesity in Chinese primary-school–aged children: a cluster-randomised controlled trial. PLoS Med 2019; 16: e1002971.

45.

Partridge

Knight

Todd

, et al. Addressing disparities: a systematic review of digital health equity for adolescent obesity prevention and management interventions. Obes Rev 2024; 25: e13821.

46.

Singh

Ahmed

Staiano

, et al. A systematic umbrella review and meta-meta-analysis of eHealth and mHealth interventions for improving lifestyle behaviours. NPJ Digit Med 2024; 7: 179.

47.

Likhitweerawong

Boonchooduang

Kittisakmontri

, et al. Effectiveness of mobile application on changing weight, healthy eating habits, and quality of life in children and adolescents with obesity: a randomized controlled trial. BMC Pediatr 2021; 21: 499.

48.

Aggarwal

Tam

, et al. Artificial intelligence-based chatbots for promoting health behavioral changes: systematic review. J Med Internet Res 2023; 25: e40789.

49.

Rahmanti

Yang

Bintoro

, et al. Slimme, a chatbot with artificial empathy for personal weight management: system design and finding. Front Nutr 2022; 9: 870775.

50.

Alsahli

Hor

Lam

. Factors influencing the acceptance and adoption of mobile health apps by physicians during the COVID-19 pandemic: systematic review. JMIR mHealth uHealth 2023; 11: e50419.

51.

Shariq

Siddiqi

Van Spall

, et al. Role of telemedicine in the management of obesity: state-of-the-art review. Obes Rev 2024; 25: e13734.

52.

Tchang

Morrison

Kim

, et al. Weight loss outcomes with telemedicine during COVID-19. Front Endocrinol (Lausanne) 2022; 13: 793290.

53.

Anawade

Sharma

Gahane

. A comprehensive review on exploring the impact of telemedicine on healthcare accessibility. Cureus 2024; 16: e55996.

54.

Perri

Shankar

Daniels

, et al. Effect of telehealth extended care for maintenance of weight loss in rural US communities: a randomized clinical trial. JAMA Network Open 2020; 3: e206764.

55.

Calcaterra

Verduci

Vandoni

, et al. Telehealth: a useful tool for the management of nutrition and exercise programs in pediatric obesity in the COVID-19 era. Nutrients 2021; 13: 3689.

56.

Bellini

Valente

Turetti

, et al. Current applications of artificial intelligence in bariatric surgery. Obes Surg 2022; 32: 2717–2733.

57.

Bae

. Preoperative risk evaluation and perioperative management of patients with obstructive sleep apnea: a narrative review. J Dent Anesth Pain Med 2023; 23: 179–192.

58.

Zhou

Xue

, et al. Constructing a prediction model for difficult intubation of obese patients based on machine learning. J Clin Anesth 2021; 72: 110278.

59.

Azmi

Kunnathodi

Alotaibi

, et al. Harnessing artificial intelligence in obesity research and management: a comprehensive review. Diagnostics (Basel, Switzerland) 2025; 15: 396.

60.

Sheikhtaheri

Orooji

Pazouki

, et al. A clinical decision support system for predicting the early complications of one-anastomosis gastric bypass surgery. Obes Surg 2019; 29: 2276–2286.

61.

Kim

Park

Shin

, et al. The role of artificial intelligence in obesity medicine. Endocrinol Metab Clin N Am 2025; 54: 207–215.

62.

Shen

Xiao

. Applications of artificial intelligence to obesity research: scoping review of methodologies. J Med Internet Res 2022; 24: e40589.

63.

Avenell

Broom

Brown

, et al. Systematic review of the long-term effects and economic consequences of treatments for obesity and implications for health improvement. Health Technol Assess (Winchester, England) 2004; 8: iii–iiv, 1–182.

64.

Omboni

Padwal

Alessa

, et al. The worldwide impact of telemedicine during COVID-19: current evidence and recommendations for the future. Connected Health 2022; 1: 7–35.

65.

Song

Lin

, et al. Application of artificial intelligence in Alzheimer's disease: a bibliometric analysis. Front Neurosci 2025; 19: 1511350.

Application of artificial intelligence in obesity management: Bibliometric analysis

Abstract

Background

Objective

Methods

Results

Conclusion

Keywords

Introduction

Materials and methods

Data sources and search strategy

Inclusion criteria

Exclusion criteria

Data extraction and analysis

Data preprocessing and standardization

Bibliometric analysis parameter settings

Results

Publication overview

Distribution and collaboration networks among countries

Distribution among institutions and collaboration networks

Authors and co-authors analysis

Contributions and collaborative networks of journals

Analysis of the top 20 most cited papers

Keyword co-occurrence analysis

Burst keyword analysis

Discussion

Worldwide trends and publishing landscape

Research hotspots

Applications of machine learning in prediction and diagnosis

Telemedicine and personalized weight management

Bariatric surgery and clinically assisted decision-making

Future directions and recommendations

Limitations

Conclusion

Footnotes

ORCID iD

Author contributions

Funding

Declaration of conflicting interests

References