Songbo Hu, Abigail Oppong, Ebele Mogo, Charlotte Collins, Giulia Occhini, Anna Barford, Anna Korhonen
{"title":"Natural Language Processing Technologies for Public Health in Africa: Scoping Review.","authors":"Songbo Hu, Abigail Oppong, Ebele Mogo, Charlotte Collins, Giulia Occhini, Anna Barford, Anna Korhonen","doi":"10.2196/68720","DOIUrl":"https://doi.org/10.2196/68720","url":null,"abstract":"<p><strong>Background: </strong>Natural language processing (NLP) has the potential to promote public health. However, applying these technologies in African health systems faces challenges, including limited digital and computational resources to support the continent's diverse languages and needs.</p><p><strong>Objective: </strong>This scoping review maps the evidence on NLP technologies for public health in Africa, addressing the following research questions: (1) What public health needs are being addressed by NLP technologies in Africa, and what unmet needs remain? (2) What factors influence the availability of public health NLP technologies across African countries and languages? (3) What stages of deployment have these technologies reached, and to what extent have they been integrated into health systems? (4) What measurable impact has these technologies had on public health outcomes, where such data are available? (5) What recommendations have been proposed to enhance the quality, cost, and accessibility of health-related NLP technologies in Africa?</p><p><strong>Methods: </strong>This scoping review includes academic studies published between January 1, 2013, and October 3, 2024. A systematic search was conducted across databases, including MEDLINE via PubMed, ACL Anthology, Scopus, IEEE Xplore, and ACM Digital Library, supplemented by gray literature searches. Data were extracted and the NLP technology functions were mapped to the World Health Organization's list of essential public health functions and the United Nations' sustainable development goals (SDGs). The extracted data were analyzed to identify trends, gaps, and areas for future research. This scoping review follows the PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses Extension for Scoping Reviews) reporting guidelines, and its protocol is publicly available.</p><p><strong>Results: </strong>Of 2186 citations screened, 54 studies were included. While existing NLP technologies support a subset of essential public health functions and SDGs, language coverage remains uneven, with limited support for widely spoken African languages, such as Kiswahili, Yoruba, Igbo, and Zulu, and no support for most of Africa's >2000 languages. Most technologies are in prototyping phases, with only one fully deployed chatbot addressing vaccine hesitancy. Evidence of measurable impact is limited, with 15% (8/54) studies attempting health-related evaluations and 4% (2/54) demonstrating positive public health outcomes, including improved participants' mood and increased vaccine intentions. Recommendations include expanding language coverage, targeting local health needs, enhancing trust, integrating solutions into health systems, and adopting participatory design approaches. The gray literature reveals industry- and nongovernmental organizations-led projects focused on deployable NLP applications. However, these projects tend to support only a few major languages and specif","PeriodicalId":16337,"journal":{"name":"Journal of Medical Internet Research","volume":"27 ","pages":"e68720"},"PeriodicalIF":5.8,"publicationDate":"2025-03-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143575733","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Enhancing Doctor-Patient Shared Decision-Making: Design of a Novel Collaborative Decision Description Language.","authors":"XiaoRui Guo, Liang Xiao, Xinyu Liu, Jianxia Chen, Zefang Tong, Ziji Liu","doi":"10.2196/55341","DOIUrl":"https://doi.org/10.2196/55341","url":null,"abstract":"<p><strong>Background: </strong>Effective shared decision-making between patients and physicians is crucial for enhancing health care quality and reducing medical errors. The literature shows that the absence of effective methods to facilitate shared decision-making can result in poor patient engagement and unfavorable decision outcomes.</p><p><strong>Objective: </strong>In this paper, we propose a Collaborative Decision Description Language (CoDeL) to model shared decision-making between patients and physicians, offering a theoretical foundation for studying various shared decision scenarios.</p><p><strong>Methods: </strong>CoDeL is based on an extension of the interaction protocol language of Lightweight Social Calculus. The language utilizes speech acts to represent the attitudes of shared decision-makers toward decision propositions, as well as their semantic relationships within dialogues. It supports interactive argumentation among decision makers by embedding clinical evidence into each segment of decision protocols. Furthermore, CoDeL enables personalized decision-making, allowing for the demonstration of characteristics such as persistence, critical thinking, and openness.</p><p><strong>Results: </strong>The feasibility of the approach is demonstrated through a case study of shared decision-making in the disease domain of atrial fibrillation. Our experimental results show that integrating the proposed language with GPT can further enhance its capabilities in interactive decision-making, improving interpretability.</p><p><strong>Conclusions: </strong>The proposed novel CoDeL can enhance doctor-patient shared decision-making in a rational, personalized, and interpretable manner.</p>","PeriodicalId":16337,"journal":{"name":"Journal of Medical Internet Research","volume":"27 ","pages":"e55341"},"PeriodicalIF":5.8,"publicationDate":"2025-03-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143575745","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Use of Artificial Intelligence, Internet of Things, and Edge Intelligence in Long-Term Care for Older People: Comprehensive Analysis Through Bibliometric, Google Trends, and Content Analysis.","authors":"Shuo-Chen Chien, Chia-Ming Yen, Yu-Hung Chang, Ying-Erh Chen, Chia-Chun Liu, Yu-Ping Hsiao, Ping-Yen Yang, Hong-Ming Lin, Tsung-En Yang, Xing-Hua Lu, I-Chien Wu, Chih-Cheng Hsu, Hung-Yi Chiou, Ren-Hua Chung","doi":"10.2196/56692","DOIUrl":"https://doi.org/10.2196/56692","url":null,"abstract":"<p><strong>Background: </strong>The global aging population poses critical challenges for long-term care (LTC), including workforce shortages, escalating health care costs, and increasing demand for high-quality care. Integrating artificial intelligence (AI), the Internet of Things (IoT), and edge intelligence (EI) offers transformative potential to enhance care quality, improve safety, and streamline operations. However, existing research lacks a comprehensive analysis that synthesizes academic trends, public interest, and deeper insights regarding these technologies.</p><p><strong>Objective: </strong>This study aims to provide a holistic overview of AI, IoT, and EI applications in LTC for older adults through a comprehensive bibliometric analysis, public interest insights from Google Trends, and content analysis of the top-cited research papers.</p><p><strong>Methods: </strong>Bibliometric analysis was conducted using data from Web of Science, PubMed, and Scopus to identify key themes and trends in the field, while Google Trends was used to assess public interest. A content analysis of the top 1% of most-cited papers provided deeper insights into practical applications.</p><p><strong>Results: </strong>A total of 6378 papers published between 2014 and 2023 were analyzed. The bibliometric analysis revealed that the United States, China, and Canada are leading contributors, with strong thematic overlaps in areas such as dementia care, machine learning, and wearable health monitoring technologies. High correlations were found between academic and public interest, in key topics such as \"long-term care\" (τ=0.89, P<.001) and \"caregiver\" (τ=0.72, P=.004). The content analysis demonstrated that social robots, particularly PARO, significantly improved mood and reduced agitation in patients with dementia. However, limitations, including small sample sizes, short study durations, and a narrow focus on dementia care, were noted.</p><p><strong>Conclusions: </strong>AI, IoT, and EI collectively form a powerful ecosystem in LTC settings, addressing different aspects of care for older adults. Our study suggests that increased international collaboration and the integration of emerging themes such as \"rehabilitation,\" \"stroke,\" and \"mHealth\" are necessary to meet the evolving care needs of this population. Additionally, incorporating high-interest keywords such as \"machine learning,\" \"smart home,\" and \"caregiver\" can enhance discoverability and relevance for both academic and public audiences. Future research should focus on expanding sample sizes, conducting long-term multicenter trials, and exploring broader health conditions beyond dementia, such as frailty and depression.</p>","PeriodicalId":16337,"journal":{"name":"Journal of Medical Internet Research","volume":"27 ","pages":"e56692"},"PeriodicalIF":5.8,"publicationDate":"2025-03-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143575784","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Jayne Barclay, Clair Sullivan, Michael Beckmann, Graeme Mattison, Rebecca Runciman, Elizabeth Martin
{"title":"Use of Remote Assessment Tools to Substitute Routine Outpatient Care: Scoping Review.","authors":"Jayne Barclay, Clair Sullivan, Michael Beckmann, Graeme Mattison, Rebecca Runciman, Elizabeth Martin","doi":"10.2196/65938","DOIUrl":"https://doi.org/10.2196/65938","url":null,"abstract":"<p><strong>Background: </strong>The increasing global demand for health care, driven by demographic shifts, the rise of personalized medicine, and technological innovations necessitate novel approaches to health care delivery. Digital remote assessment tools have emerged as a promising solution, enabling hybrid care models that combine traditional and remote patient management. These tools support the quadruple aim of health care by enhancing the monitoring and evaluation of patient-reported data, thereby improving patient care, boosting operational efficiency, reducing costs, and improving the experience of patients and clinicians. This review seeks to understand how remote assessment tools are used for routine consultation substitution in adult tertiary care centers.</p><p><strong>Objective: </strong>This scoping review aims to evaluate the implementation and health outcomes of digital remote assessment tools used for routine consultation substitutions in adult tertiary care centers. The objectives include assessing the extent of use, types, and effectiveness of these tools in substituting conventional outpatient care.</p><p><strong>Methods: </strong>A comprehensive scoping review was conducted, adhering to the PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews) checklist. The review focused on studies that used internet-dependent remote assessment technologies for patient data transfer in tertiary care settings. A detailed search strategy was used across multiple databases, with studies selected based on predefined inclusion and exclusion criteria. Data extraction and analysis were performed by independent reviewers, with a focus on the functionalities of the tools and their alignment with the Quadruple Aim of Healthcare.</p><p><strong>Results: </strong>The review included 12 studies, highlighting a growing interest in remote assessment technologies across diverse clinical settings. The interventions varied in length, from 4 weeks to 12 months, and demonstrated a range of functionalities, including symptom monitoring and postsurgical follow-ups. The use of these tools was associated with improved clinical outcomes, such as timely intervention for clinical deterioration and enhanced clinical protocol adherence. Additionally, a small number of studies identified potential cost savings in terms of reduced unplanned health care contacts and optimized clinical resource use. Patient and clinician experiences were generally positive, with high adherence to remote assessments and an appreciation for the personalized and timely care facilitated by these technologies. Barriers included high initial setup costs for digital technologies, leading to an inflated cost per patient in small sample studies.</p><p><strong>Conclusions: </strong>Digital remote assessment tools offer significant potential to enhance health care delivery by improving health outcomes, reducing costs, and enriching patient and ","PeriodicalId":16337,"journal":{"name":"Journal of Medical Internet Research","volume":"27 ","pages":"e65938"},"PeriodicalIF":5.8,"publicationDate":"2025-03-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143575832","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Ashlyn Beecroft, Olivia Vaikla, Nora Engel, Thomas Duchaine, Chen Liang, Nitika Pant Pai
{"title":"Evidence on Digital HIV Self-Testing From Accuracy to Impact: Updated Systematic Review.","authors":"Ashlyn Beecroft, Olivia Vaikla, Nora Engel, Thomas Duchaine, Chen Liang, Nitika Pant Pai","doi":"10.2196/63110","DOIUrl":"https://doi.org/10.2196/63110","url":null,"abstract":"<p><strong>Background: </strong>HIV self-testing has gained momentum following the approval of self-testing methods and novel technological advancements. Digital HIV self-testing involves completing an oral or blood-based HIV self-test with support from a digital innovation.</p><p><strong>Objective: </strong>We conducted a systematic review on the existing data analyzing digital HIV self-testing accuracy while updating research on digital HIV self-test acceptability, preference, feasibility, and impact.</p><p><strong>Methods: </strong>We searched Embase and PubMed for records on HIV self-testing with digital support. Included studies significantly incorporated a form of digital innovation throughout the HIV self-test process and reported quantitative data. For accuracy measures, the search spanned January 1, 2013, to October 15, 2024; for patient-centered and impact outcomes, we updated existing literature (June 16, 2021, to October 15, 2024) reported in a previous systematic review. Studies' quality was assessed using the QUADAS 2 Tool, Newcastle-Ottawa Scale, and Cochrane Risk of Bias Tool 2.</p><p><strong>Results: </strong>Fifty-five studies (samples ranging 120-21,035, median 1267 participants) were summarized from 19 middle- to high-income countries. Seven studies reported on the accuracy of HIV self-testing with innovations from >5000 participants. Diagnostic performance metrics, including point estimates of specificity, sensitivity, positive predictive value, and negative predictive value were measured (n=3), and ranged from: 96.8% to 99.9%, 92.9% to 100.0%, 76.5% to 99.2%, and 99.2% to 100.0%, respectively. The percentage of invalid test results for oral and blood-based self-tests ranged from 0.2% to 12.7% (n=4). Fifty-one studies reported data on metrics beyond accuracy, including acceptability, preference, feasibility, and impact outcomes from >30,000 participants. Majority (38/51, 74.5%) were observational studies, while 25.5% (13/51) reported data from randomized controlled trials. Acceptability and preference outcomes varied from 64.5% to 99.0% (14/51) and 4.6% to 99.3% (8/51), respectively. Feasibility outcomes included test uptake (30.9% to 98.2%; 28/51), response rate (26.0% to 94.8%; 7/51), and visits to web-based providers (43.0% to 70.7%; n=4). Impact outcomes assessed new infections (0.0% to 25.8%; 31/51), first-time testers (2.0% to 53.0%; 26/51), result return proportions (22.1% to 100.0%; 24/51), linkage to care as both connections to confirmatory testing and counseling (53.0% to 100.0%; 16/51), and referrals for treatment initiation (44.4% to 98.1%; 8/51). The quality of studies varied, though they generally demonstrated low risk of bias.</p><p><strong>Conclusions: </strong>Digital innovations improved the accuracy of HIV self-test results, and were well-accepted and preferred by participants. Operationally, they were found to be feasible and reported impacting the HIV self-testing process. These findings are in favor of t","PeriodicalId":16337,"journal":{"name":"Journal of Medical Internet Research","volume":"27 ","pages":"e63110"},"PeriodicalIF":5.8,"publicationDate":"2025-03-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143575746","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Assessing Short-Video Dependence for e-Mental Health: Development and Validation Study of the Short-Video Dependence Scale.","authors":"AnHang Jiang, Shuang Li, HuaBin Wang, HaoSen Ni, HongAn Chen, JunHong Dai, XueFeng Xu, Mei Li, Guang-Heng Dong","doi":"10.2196/66341","DOIUrl":"https://doi.org/10.2196/66341","url":null,"abstract":"<p><strong>Background: </strong>Short-video dependence (SVD) has become a significant mental health issue around the world. The lack of scientific tools to assess SVD hampers further advancement in this area.</p><p><strong>Objective: </strong>This study aims to develop and validate a scientific tool to measure SVD levels, ensuring a scientifically determined cutoff point.</p><p><strong>Methods: </strong>We initially interviewed 115 highly engaged short-video users aged 15 to 63 years. Based on the summary of the interview and references to the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) criteria for behavioral addictions, we proposed the first version of the short-video dependence scale (SVDS). We then screened the items through item analysis (second version) and extracted common factors using exploratory factor analysis (third version) and confirmatory factor analysis (final version). Convergent validity was tested with other scales (Chinese Internet Addiction Scale [CIAS] and DSM-5). Finally, we tested the validity of the final version in 16,038 subjects and set the diagnostic cutoff point through latent profile analysis and receiver operating characteristic curve analysis.</p><p><strong>Results: </strong>The final version of the SVDS contained 20 items and 4 dimensions, which showed strong structural validity (Kaiser-Meyer-Olkin value=0.94) and internal consistency (Cronbach α=.93), and good convergent validity (r<sub>CIAS</sub>=0.61 and r<sub>DSM-5</sub>=0.68), sensitivity (0.77, 0.83, 0.87, and 0.62 for each of the 4 dimensions), and specificity (0.75, 0.87, 0.80, and 0.79 for each of the 4 dimensions). Additionally, an SVDS score of 58 was determined as the best cutoff score, and latent profile analysis identified a 5-class model for SVD.</p><p><strong>Conclusions: </strong>We developed a tool to measure SVD levels and established a threshold to differentiate dependent users from highly engaged nondependent users. The findings provide opportunities for further research on the impacts of short-video use.</p>","PeriodicalId":16337,"journal":{"name":"Journal of Medical Internet Research","volume":"27 ","pages":"e66341"},"PeriodicalIF":5.8,"publicationDate":"2025-03-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143575619","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Avinash Koka, Loric Stuby, Emmanuel Carrera, Ahmed Gabr, Margaret O'Connor, Nathalie Missilier Peruzzo, Olivier Waeterloot, Friedrich Medlin, Fabien Rigolet, Thomas Schmutz, Patrik Michel, Thibaut Desmettre, Mélanie Suppan, Laurent Suppan
{"title":"Asynchronous Distance Learning Performance and Knowledge Retention of the National Institutes of Health Stroke Scale Among Health Care Professionals Using Video or e-Learning: Web-based Randomized Controlled Trial.","authors":"Avinash Koka, Loric Stuby, Emmanuel Carrera, Ahmed Gabr, Margaret O'Connor, Nathalie Missilier Peruzzo, Olivier Waeterloot, Friedrich Medlin, Fabien Rigolet, Thomas Schmutz, Patrik Michel, Thibaut Desmettre, Mélanie Suppan, Laurent Suppan","doi":"10.2196/63136","DOIUrl":"https://doi.org/10.2196/63136","url":null,"abstract":"<p><strong>Background: </strong>Stroke treatment has significantly improved over the last decades, but the complexity of stroke cases requires specialized care through dedicated teams with specific knowledge and training. The National Institutes of Health Stroke Scale (NIHSS), widely used to assess neurological deficits and make treatment decisions, is reliable but requires specific training and certification. The traditional didactic training method, based on a video, may not adequately address certain NIHSS intricacies nor engage health care professionals (HCPs) in continuous learning, leading to suboptimal proficiency. In the context of time-constrained clinical settings, highly interactive e-learning could be a promising alternative for NIHSS knowledge acquisition and retention.</p><p><strong>Objective: </strong>This study aimed to assess the efficacy of a highly interactive e-learning module compared with a traditional didactic video in improving NIHSS knowledge among previously trained HCPs. Furthermore, its impact on knowledge retention was also assessed.</p><p><strong>Methods: </strong>A prospective, multicentric, triple-blind, and web-based randomized controlled trial was conducted in 3 Swiss university hospitals, involving HCPs previously trained in NIHSS. Invitations were sent through email, and participants were randomized to either the e-learning or traditional didactic video group through a fully automated process upon self-registration on the website. A 50-question quiz was administered before and after exposure to the training method, and scores were compared to assess knowledge acquisition. The quiz was repeated after 1 month to evaluate retention. Subjective assessments of learning methods that is, user satisfaction, probability of recommendation, perceived difficulty, and perception of duration, were also collected through a Likert-scale questionnaire. A sample size of 72 participants were deemed necessary to have an 80% chance of detecting a difference of 2 points in the postcourse quiz between groups at the 5% significance level.</p><p><strong>Results: </strong>Invitations to participate were sent through email to an estimated 325 HCPs. 174 HCPs enrolled in the study, of which 97 completed the study course. Both learning methods significantly improved NIHSS knowledge, with an improvement of 3.2 (range 2.0-4.3) points in the e-learning group and of 2.1 (1.2-3.1) points in the video group. However, the e-learning group performed better, with higher scores in knowledge acquisition (median score 39.0, IQR 36.0-41.0 vs 37, IQR 34.0-39.0; P=.03) and in knowledge retention (mean score 38.2, 95% CI 36.7-39.7 vs 35.8, 95% CI 34.8-36.8; P=.007). Participants in the e-learning group were more likely to recommend the learning method (77% vs 49%, P=.02), while no significant difference was found for satisfaction (P=.17), perceived duration (P=.17), and difficulty (P=.32).</p><p><strong>Conclusions: </strong>A highly interactive e-learning","PeriodicalId":16337,"journal":{"name":"Journal of Medical Internet Research","volume":"27 ","pages":"e63136"},"PeriodicalIF":5.8,"publicationDate":"2025-03-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143575660","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Large Language Models' Accuracy in Emulating Human Experts' Evaluation of Public Sentiments about Heated Tobacco Products on Social Media: Evaluation Study.","authors":"Kwanho Kim, Soojong Kim","doi":"10.2196/63631","DOIUrl":"https://doi.org/10.2196/63631","url":null,"abstract":"<p><strong>Background: </strong>Sentiment analysis of alternative tobacco products discussed on social media is crucial in tobacco control research. Large language models (LLMs) are artificial intelligence models that were trained on extensive text data to emulate the linguistic patterns of humans. LLMs may hold the potential to streamline the time-consuming and labor-intensive process of human sentiment analysis.</p><p><strong>Objective: </strong>This study aimed to examine the accuracy of LLMs in replicating human sentiment evaluation of social media messages relevant to heated tobacco products (HTPs).</p><p><strong>Methods: </strong>GPT-3.5 and GPT-4 Turbo (OpenAI) were used to classify 500 Facebook (Meta Platforms) and 500 Twitter (subsequently rebranded X) messages. Each set consisted of 200 human-labeled anti-HTPs, 200 pro-HTPs, and 100 neutral messages. The models evaluated each message up to 20 times to generate multiple response instances reporting its classification decisions. The majority of the labels from these responses were assigned as a model's decision for the message. The models' classification decisions were then compared with those of human evaluators.</p><p><strong>Results: </strong>GPT-3.5 accurately replicated human sentiment evaluation in 61.2% of Facebook messages and 57% of Twitter messages. GPT-4 Turbo demonstrated higher accuracies overall, with 81.7% for Facebook messages and 77% for Twitter messages. GPT-4 Turbo's accuracy with 3 response instances reached 99% of the accuracy achieved with 20 response instances. GPT-4 Turbo's accuracy was higher for human-labeled anti- and pro-HTP messages compared with neutral messages. Most of the GPT-3.5 misclassifications occurred when anti- or pro-HTP messages were incorrectly classified as neutral or irrelevant by the model, whereas GPT-4 Turbo showed improvements across all sentiment categories and reduced misclassifications, especially in incorrectly categorized messages as irrelevant.</p><p><strong>Conclusions: </strong>LLMs can be used to analyze sentiment in social media messages about HTPs. Results from GPT-4 Turbo suggest that accuracy can reach approximately 80% compared with the results of human experts, even with a small number of labeling decisions generated by the model. A potential risk of using LLMs is the misrepresentation of the overall sentiment due to the differences in accuracy across sentiment categories. Although this issue could be reduced with the newer language model, future efforts should explore the mechanisms underlying the discrepancies and how to address them systematically.</p>","PeriodicalId":16337,"journal":{"name":"Journal of Medical Internet Research","volume":"27 ","pages":"e63631"},"PeriodicalIF":5.8,"publicationDate":"2025-03-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143575596","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"The Impact of Stay-At-Home Mandates on Uncertainty and Sentiments: Quasi-Experimental Study.","authors":"Carolina Biliotti, Nicolò Fraccaroli, Michelangelo Puliga, Falco J Bargagli-Stoffi, Massimo Riccaboni","doi":"10.2196/64667","DOIUrl":"https://doi.org/10.2196/64667","url":null,"abstract":"<p><strong>Background: </strong>As the spread of the SARS-CoV-2 virus coincided with lockdown measures, it is challenging to distinguish public reactions to lockdowns from responses to COVID-19 itself. Beyond the direct impact on health, lockdowns may have worsened public sentiment toward politics and the economy or even heightened dissatisfaction with health care, imposing a significant cost on both the public and policy makers.</p><p><strong>Objective: </strong>This study aims to analyze the causal effect of COVID-19 lockdown policies on various dimensions of sentiment and uncertainty, using the Italian lockdown of February 2020 as a quasi-experiment. At the time of implementation, communities inside and just outside the lockdown area were equally exposed to COVID-19, enabling a quasi-random distribution of the lockdown. Additionally, both areas had similar socioeconomic and demographic characteristics before the lockdown, suggesting that the delineation of the strict lockdown zone approximates a randomized experiment. This approach allows us to isolate the causal effects of the lockdown on public emotions, distinguishing the impact of the policy itself from changes driven by the virus's spread.</p><p><strong>Methods: </strong>We used Twitter data (N=24,261), natural language models, and a difference-in-differences approach to compare changes in sentiment and uncertainty inside (n=1567) and outside (n=22,694) the lockdown areas before and after the lockdown began. By fine-tuning the AlBERTo (Italian BERT optimized) pretrained model, we analyzed emotions expressed in tweets from 1124 unique users. Additionally, we applied dictionary-based methods to categorize tweets into 4 dimensions-economy, health, politics, and lockdown policy-to assess the corresponding emotional reactions. This approach enabled us to measure the direct impact of local policies on public sentiment using geo-referenced social media and can be easily adapted for other policy impact analyses.</p><p><strong>Results: </strong>Our analysis shows that the lockdown had no significant effect on economic uncertainty (b=0.005, SE 0.007, t125=0.70; P=.48) or negative economic sentiment (b=-0.011, SE 0.0089, t125=-1.32; P=.19). However, it increased uncertainty about health (b=0.036, SE 0.0065, t125=5.55; P<.001) and lockdown policy (b=0.026, SE 0.006, t125=4.47; P<.001), as well as negative sentiment toward politics (b=0.025, SE 0.011, t125=2.33; P=.02), indicating that lockdowns have broad externalities beyond health. Our key findings are confirmed through a series of robustness checks.</p><p><strong>Conclusions: </strong>Our findings reveal that lockdowns have broad externalities extending beyond health. By heightening health concerns and negative political sentiment, policy makers have struggled to secure explicit public support for government measures, which may discourage future leaders from implementing timely stay-at-home policies. These results highlight the need for authoritie","PeriodicalId":16337,"journal":{"name":"Journal of Medical Internet Research","volume":"27 ","pages":"e64667"},"PeriodicalIF":5.8,"publicationDate":"2025-03-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143575813","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Chen-Chi Duan, Chen Zhang, Hua-Lin Xu, Jing Tao, Jia-Le Yu, Dan Zhang, Shan Wu, Xiu Zeng, Wan-Ting Zeng, Zhi-Yin Zhang, Cindy-Lee Dennis, Han Liu, Jia-Ying Wu, Ben Willem J Mol, He-Feng Huang, Yan-Ting Wu
{"title":"Internet-Based Cognitive Behavioral Therapy for Preventing Postpartum Depressive Symptoms Among Pregnant Individuals With Depression: Multicenter Randomized Controlled Trial in China.","authors":"Chen-Chi Duan, Chen Zhang, Hua-Lin Xu, Jing Tao, Jia-Le Yu, Dan Zhang, Shan Wu, Xiu Zeng, Wan-Ting Zeng, Zhi-Yin Zhang, Cindy-Lee Dennis, Han Liu, Jia-Ying Wu, Ben Willem J Mol, He-Feng Huang, Yan-Ting Wu","doi":"10.2196/67386","DOIUrl":"https://doi.org/10.2196/67386","url":null,"abstract":"<p><strong>Background: </strong>Women are particularly vulnerable to depression during pregnancy, which is one of the strongest risk factors for developing postpartum depression (PPD). Addressing antenatal depressive symptoms in these women is crucial for preventing PPD. However, little is known about the effectiveness of internet-based cognitive behavioral therapy (ICBT) in preventing PPD in this high-risk group.</p><p><strong>Objective: </strong>This study aims to evaluate the short- and long-term effects of ICBT in preventing PPD among women with antenatal depressive symptoms.</p><p><strong>Methods: </strong>Participants were screened for antenatal depressive symptoms using the Edinburgh Postnatal Depression Scale (EPDS) and randomly allocated (1:1) to either the ICBT group (receiving weekly online modules starting antenatally and continuing into early postpartum) or the control group (observed without treatment). Follow-up assessments were conducted up to 12 months postpartum, and data were analyzed using generalized estimating equations. The primary outcome was the prevalence of depressive symptoms at 6 weeks postpartum. A subgroup analysis based on the severity of antenatal depressive symptoms was also performed. The secondary outcomes included the long-term effects of ICBT on maternal depression, as well as its impact on anxiety, sleep quality, social support, parenting stress, co-parenting relationships, and infant development.</p><p><strong>Results: </strong>Between August 2020 and September 2021, 300 pregnant individuals were recruited from 5 centers across China. No significant differences were observed in depressive symptoms at 6 weeks postpartum (P=.18) or at any longer-term follow-up time points (P=.18). However, a post hoc subgroup analysis showed that participants with antenatal EPDS scores of 10-12 in the ICBT group had a lower risk of developing depression during the first year postpartum (odds ratio 0.534, 95% CI 0.313-0.912; P=.02), but this was not observed for participants with more severe depression. Additionally, this subgroup demonstrated higher levels of co-parenting relationships (P=.02).</p><p><strong>Conclusions: </strong>Among individuals with antenatal depression, ICBT did not prevent the development of PPD. However, ICBT may be a preferable option for those with mild to moderate antenatal depressive symptoms. Future research is needed to explore modifications to ICBT to address more severe depressive symptoms.</p><p><strong>Trial registration: </strong>Chinese Clinical Trial Registry ChiCTR2000033433; https://www.chictr.org.cn/showproj.html?proj=54482.</p><p><strong>International registered report identifier (irrid): </strong>RR2-10.1186/s13063-022-06728-5.</p>","PeriodicalId":16337,"journal":{"name":"Journal of Medical Internet Research","volume":"27 ","pages":"e67386"},"PeriodicalIF":5.8,"publicationDate":"2025-03-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143575569","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}