Journal of Medical Internet Research最新文献

{"title":"Safety and User Experience of a Generative Artificial Intelligence Digital Mental Health Intervention: Exploratory Randomized Controlled Trial.","authors":"Timothy R Campellone, Megan Flom, Robert M Montgomery, Lauren Bullard, Maddison C Pirner, Aaron Pavez, Michelle Morales, Devin Harper, Catherine Oddy, Tom O'Connor, Jade Daniels, Stephanie Eaneff, Valerie L Forman-Hoffman, Casey Sackett, Alison Darcy","doi":"10.2196/67365","DOIUrl":"10.2196/67365","url":null,"abstract":"<p><strong>Background: </strong>General awareness and exposure to generative artificial intelligence (AI) have increased recently. This transformative technology has the potential to create a more dynamic and engaging user experience in digital mental health interventions (DMHIs). However, if not appropriately used and controlled, it can introduce risks to users that may result in harm and erode trust. At the time of conducting this trial, there had not been a rigorous evaluation of an approach to safely implementing generative AI in a DMHI.</p><p><strong>Objective: </strong>This study aims to explore the user relationship, experience, safety, and technical guardrails of a DMHI using generative AI compared with a rules-based intervention.</p><p><strong>Methods: </strong>We conducted a 2-week exploratory randomized controlled trial (RCT) with 160 adult participants randomized to receive a generative AI (n=81) or rules-based (n=79) version of a conversation-based DMHI. Self-report measures of the user relationship (client satisfaction, working alliance bond, and accuracy of empathic listening and reflection) and experience (engagement metrics, adverse events, and technical guardrail success) were collected. Descriptions and validation of technical guardrails for handling user inputs (eg, detecting potentially concerning language and off-topic responses) and model outputs (eg, not providing medical advice and not providing a diagnosis) are provided, along with examples to illustrate how they worked. Safety monitoring was conducted throughout the trial for adverse events, and the success of technical guardrails created for the generative arm was assessed post trial.</p><p><strong>Results: </strong>In general, the majority of measures of user relationship and experience appeared to be similar in both the generative and rules-based arms. The generative arm appeared to be more accurate at detecting and responding to user statements with empathy (98% accuracy vs 69%). There were no serious or device-related adverse events, and technical guardrails were shown to be 100% successful in posttrial review of generated statements. A majority of participants in both groups reported an increase in positive sentiment (62% and 66%) about AI at the end of the trial.</p><p><strong>Conclusions: </strong>This trial provides initial evidence that, with the right guardrails and process, generative AI can be successfully used in a digital mental health intervention (DMHI) while maintaining the user experience and relationship. It also provides an initial blueprint for approaches to technical and conversational guardrails that can be replicated to build a safe DMHI.</p><p><strong>Trial registration: </strong>ClinicalTrials.gov NCT05948670; https://clinicaltrials.gov/study/NCT05948670.</p>","PeriodicalId":16337,"journal":{"name":"Journal of Medical Internet Research","volume":"27 ","pages":"e67365"},"PeriodicalIF":5.8,"publicationDate":"2025-05-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144127818","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":"Using Digital Phenotyping to Discriminate Unipolar Depression and Bipolar Disorder: Systematic Review.","authors":"Rongrong Zhong, XiaoHui Wu, Jun Chen, Yiru Fang","doi":"10.2196/72229","DOIUrl":"https://doi.org/10.2196/72229","url":null,"abstract":"&lt;p&gt;&lt;strong&gt;Background: &lt;/strong&gt;Differentiating bipolar disorder (BD) from unipolar depression (UD) is essential, as these conditions differ greatly in their progression and treatment approaches. Digital phenotyping, which involves using data from smartphones or other digital devices to assess mental health, has emerged as a promising tool for distinguishing between these two disorders.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Objective: &lt;/strong&gt;This systematic review aimed to achieve two goals: (1) to summarize the existing literature on the use of digital phenotyping to directly distinguish between UD and BD and (2) to review studies that use digital phenotyping to classify UD, BD, and healthy control (HC) individuals. Furthermore, the review sought to identify gaps in the current research and propose directions for future studies.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Methods: &lt;/strong&gt;We systematically searched the Scopus, IEEE Xplore, PubMed, Embase, Web of Science, and PsycINFO databases up to March 20, 2025. Studies were included if they used portable or wearable digital tools to directly distinguish between UD and BD, or to classify UD, BD, and HC. Original studies published in English, including both journal and conference papers, were included, while reviews, narrative reviews, systematic reviews, and meta-analyses were excluded. Articles were excluded if the diagnosis was not made through a professional medical evaluation or if they relied on electronic health records or clinical data. For each included study, the following information was extracted: demographic characteristics, diagnostic criteria or psychiatric assessments, details of the technological tools and data types, duration of data collection, data preprocessing methods, selected variables or features, machine learning algorithms or statistical tests, validation, and main findings.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Results: &lt;/strong&gt;We included 21 studies, of which 11 (52%) focused on directly distinguishing between UD and BD, while 10 (48%) classified UD, BD, and HC. The studies were categorized into 4 groups based on the type of digital tool used: 6 (29%) used smartphone apps, 3 (14%) used wearable devices, 11 (52%) analyzed audiovisual recordings, and 1 (5%) used multimodal technologies. Features such as activity levels from smartphone apps or wearable devices emerged as potential markers for directly distinguishing UD and BD. Patients with BD generally exhibited lower activity levels than those with UD. They also tended to show higher activity in the morning and lower in the evening, while patients with UD showed the opposite pattern. Moreover, speech modalities or the integration of multiple modalities achieved better classification performance across UD, BD, and HC groups, although the specific contributing features remained unclear.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Conclusions: &lt;/strong&gt;Digital phenotyping shows potential in distinguishing BD from UD, but challenges like data privacy, security concerns, and equitable access must be addressed. Further rese","PeriodicalId":16337,"journal":{"name":"Journal of Medical Internet Research","volume":"27 ","pages":"e72229"},"PeriodicalIF":5.8,"publicationDate":"2025-05-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144132441","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":"Early Predictive Accuracy of Machine Learning for Hemorrhagic Transformation in Acute Ischemic Stroke: Systematic Review and Meta-Analysis.","authors":"Benqiao Wang, Bohao Jiang, Dan Liu, Ruixia Zhu","doi":"10.2196/71654","DOIUrl":"https://doi.org/10.2196/71654","url":null,"abstract":"<p><strong>Background: </strong>Hemorrhagic transformation (HT) is commonly detected in acute ischemic stroke (AIS) and often leads to poor outcomes. Currently, there is no ideal tool for early prediction of HT risk. Recently, machine learning has gained traction in stroke management, prompting the exploration of predictive models for HT. However, systematic evidence on these models is lacking.</p><p><strong>Objective: </strong>In this study, we assessed the predictive capability of machine learning models for HT risk in AIS, aiming to inform the development of HT prediction tools.</p><p><strong>Methods: </strong>We conducted a thorough search of medical databases, such as Web of Science, Embase, Cochrane, and PubMed up until March 2025. The risk of bias was determined through the Prediction Model Risk of Bias Assessment Tool (PROBAST). Subgroup analysis was performed based on treatment backgrounds, diagnostic criteria, and types of HT.</p><p><strong>Results: </strong>A total of 83 eligible articles were included, containing 106 models and 88,197 patients with AIS with 9323 HT cases. There were 104 validation sets with a total c-index of 0.832 (95% CI 0.814-0.849), sensitivity of 0.82 (95% CI 0.79-0.84), and specificity of 0.78 (95% CI 0.74-0.81). Subgroup analysis indicated that the combined model achieved superior prediction accuracy. Moreover, we also analyzed the predictive performance of 6 mature models.</p><p><strong>Conclusions: </strong>Currently, although several prediction methods for HT have been developed, their predictive values are not satisfactory. Fortunately, our findings suggest that machine learning methods, particularly those combining clinical features and radiomics, hold promise for improving predictive accuracy. Our meta-analysis may provide evidence-based guidance for the subsequent development of more efficient clinical predictive models for HT.</p><p><strong>Trial registration: </strong>PROSPERO CRD42024498997; https://www.crd.york.ac.uk/PROSPERO/view/CRD42024498997.</p>","PeriodicalId":16337,"journal":{"name":"Journal of Medical Internet Research","volume":"27 ","pages":"e71654"},"PeriodicalIF":5.8,"publicationDate":"2025-05-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144132432","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":"Leveraging Social Media Data to Understand the Impact of COVID-19 on Residents' Dietary Behaviors: Observational Study.","authors":"Chuqin Li, Alexis Jordan, Yaorong Ge, Albert Park","doi":"10.2196/51638","DOIUrl":"https://doi.org/10.2196/51638","url":null,"abstract":"&lt;p&gt;&lt;strong&gt;Background: &lt;/strong&gt;The COVID-19 pandemic has inflicted global devastation, infecting over 750 million and causing 6 million deaths. In an effort to control the spread of the virus, governments around the world implemented a variety of measures, including stay-at-home orders, school closures, and mask mandates. These measures had a substantial impact on dietary behavior, with individuals discussing more home-cooked meals and snacking on social media.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Objective: &lt;/strong&gt;The study explores pandemic-induced dietary behavior changes using Twitter images and text, particularly in relation to obesity, to inform interventions and understand societal influences on eating habits. Additionally, the study investigates the impact of COVID-19 on emotions and eating patterns.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Methods: &lt;/strong&gt;In this study, we collected approximately 200,000 tweets related to food between May and July in 2019, 2020, and 2021. We used transfer learning and a pretrained ResNet-101 neural network to classify images into 4 health categories: definitely healthy, healthy, unhealthy, and definitely unhealthy. We then used the state obesity rates from the Behavioral Risk Factor Surveillance System (BRFSS) to assess the correlation between state obesity rates and dietary images on Twitter. The study further investigates the effects of COVID-19 on emotional changes and their relation to eating patterns via sentiment analysis. Furthermore, we illustrated how the popularity of meal terms and health categories changed over time, considering varying time zones by incorporating geolocation data.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Results: &lt;/strong&gt;A significant correlation was observed between state obesity rates and the percentages of definitely healthy (r=-0.360, P=.01) and definitely unhealthy (r=0.306, P=.03) food images in 2019. However, no trend was observed in 2020 and 2021, despite higher obesity rates. A significant (P&lt;.001) increase in the percentage of healthy food consumption was observed during (39.99% in 2020) and after the shutdown (39.32% in 2021), as compared with the preshutdown period (37.69% in 2019). Sentiment analysis from 2019, 2020, and 2021 revealed a more positive sentiment associated with dietary posts from 2019. This was the case regardless of the healthiness of the food mentioned in the tweet. Last, we found a shift in consumption time and an increase in snack consumption during and after the pandemic. People ate breakfast later (ie, from 7 AM to 8 AM in 2019 to 8 AM to 9 AM in 2020 and 2021) and dinner earlier (ie, from 6 PM to 7 PM in 2019, to 5 PM to 6 PM in 2020). Snacking frequency also increased. Taken together, dietary behavior shifted toward healthier choices at the population level during and after the COVID-19 shutdown, with potential for long-term health consequences.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Conclusions: &lt;/strong&gt;We were able to observe people's eating habits using social media data to investigate the effects of COVID-19 on dietary behaviors","PeriodicalId":16337,"journal":{"name":"Journal of Medical Internet Research","volume":"27 ","pages":"e51638"},"PeriodicalIF":5.8,"publicationDate":"2025-05-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144132436","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":"Effectiveness and Methodologies of Virtual Reality Dental Simulators for Veneer Tooth Preparation Training: Randomized Controlled Trial.","authors":"Yaning Li, Hongqiang Ye, Wenxiao Wu, Jiayi Li, Xiaohan Zhao, Yunsong Liu, Yongsheng Zhou","doi":"10.2196/63961","DOIUrl":"10.2196/63961","url":null,"abstract":"&lt;p&gt;&lt;strong&gt;Background: &lt;/strong&gt;Virtual reality (VR) simulators are increasingly used in dental education, offering advantages such as repeatable practice and immediate feedback. However, evidence comparing their efficacy to traditional phantom heads for veneer preparation training remains limited.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Objective: &lt;/strong&gt;This study aimed to compare the effectiveness of 2 widely used VR simulators (Unidental and Simodont) against traditional phantom heads for veneer tooth preparation training and evaluate the impact of training sequence (simulator-first vs phantom-head-first) on skill acquisition.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Methods: &lt;/strong&gt;A randomized controlled trial was conducted with 80 fourth-year dental students from Peking University School of Stomatology. Participants were stratified by gender and academic performance, then equally allocated to 8 groups. Groups 1-3 trained exclusively using Unidental, Simodont, or phantom heads, respectively, while groups 4-8 followed hybrid sequences combining simulator and phantom-head training. Each participant performed veneer preparations on a maxillary central incisor. Preparations were evaluated by a blinded instructor using a validated 100-point rubric assessing marginal integrity (30%), preparation depth (25%), proximal contour (25%), and surface smoothness (20%). Posttraining questionnaires (100-point scale) compared user perceptions of simulator realism, haptic feedback, and educational value.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Results: &lt;/strong&gt;There were no statistically significant differences in the preparation quality among groups using different training methods (Unidental: 88.9, SD 3.6; Simodont: 88.6, SD 1.6; phantom heads: 89.4, SD 2.8; P=.81) or different training methodologies (simulator-first vs phantom-head-first) (simulator first: P=.18; phantom head first: P=.09, different sequences of Unidental: P=.16; different sequences of Simodont: P=.11). However, significant differences were observed between the evaluations of the 2 simulators in terms of realism of the odontoscope's reflection (Simodont: 55.6, SD 33.7; Unidental: 87.5, SD 13.9; P&lt;.001), force feedback (Simodont: 66.2, SD 22.4; Unidental: 50.8, SD 18.9; P=.007), and simulation of the tooth preparation process (Simodont: 64.4, SD 16.0; Unidental: 50.6, SD 16.6; P=.003). Evaluation results showed no statistical differences between the 2 simulators in display effect (Simodont: 77.43, SD 21.58; Unidental: 71.68, SD 20.70; P=.24), synchronism of virtual and actual dental instruments (Simodont: 67.86, SD 19.31; Unidental: 59.29, SD 20.10; P=.11), and dental bur operation simulation (Simodont: 63.32, SD 19.99; Unidental: 55.79, SD 19.62; P=.16). The Unidental simulator was rated better than the Simodont simulator in terms of the realism of odontoscope's reflection. In all other aspects, Simodont was superior to Unidental. There was no significant difference in the students' attitudes towards the 2 simulators (improve skills: P=.19; inspire to learn: P=","PeriodicalId":16337,"journal":{"name":"Journal of Medical Internet Research","volume":"27 ","pages":"e63961"},"PeriodicalIF":5.8,"publicationDate":"2025-05-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12121536/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144119958","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Patient Reactions to Artificial Intelligence-Clinician Discrepancies: Web-Based Randomized Experiment.","authors":"Farrah Madanay, Laura S O'Donohue, Brian J Zikmund-Fisher","doi":"10.2196/68823","DOIUrl":"https://doi.org/10.2196/68823","url":null,"abstract":"&lt;p&gt;&lt;strong&gt;Background: &lt;/strong&gt;As the US Food and Drug Administration (FDA)-approved use of artificial intelligence (AI) for medical imaging rises, radiologists are increasingly integrating AI into their clinical practices. In lung cancer screening, diagnostic AI offers a second set of eyes with the potential to detect cancer earlier than human radiologists. Despite AI's promise, a potential problem with its integration is the erosion of patient confidence in clinician expertise when there is a discrepancy between the radiologist's and the AI's interpretation of the imaging findings.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Objective: &lt;/strong&gt;We examined how discrepancies between AI-derived recommendations and radiologists' recommendations affect patients' agreement with radiologists' recommendations and satisfaction with their radiologists. We also analyzed how patients' medical maximizing-minimizing preferences moderate these relationships.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Methods: &lt;/strong&gt;We conducted a randomized, between-subjects experiment with 1606 US adult participants. Assuming the role of patients, participants imagined undergoing a low-dose computerized tomography scan for lung cancer screening and receiving results and recommendations from (1) a radiologist only, (2) AI and a radiologist in agreement, (3) a radiologist who recommended more testing than AI (ie, radiologist overcalled AI), or (4) a radiologist who recommended less testing than AI (ie, radiologist undercalled AI). Participants rated the radiologist on three criteria: agreement with the radiologist's recommendation, how likely they would be to recommend the radiologist to family and friends, and how good of a provider they perceived the radiologist to be. We measured medical maximizing-minimizing preferences and categorized participants as maximizers (ie, those who seek aggressive intervention), minimizers (ie, those who prefer no or passive intervention), and neutrals (ie, those in the middle).&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Results: &lt;/strong&gt;Participants' agreement with the radiologist's recommendation was significantly lower when the radiologist undercalled AI (mean 4.01, SE 0.07, P&lt;.001) than in the other 3 conditions, with no significant differences among them (radiologist overcalled AI [mean 4.63, SE 0.06], agreed with AI [mean 4.55, SE 0.07], or had no AI [mean 4.57, SE 0.06]). Similarly, participants were least likely to recommend (P&lt;.001) and positively rate (P&lt;.001) the radiologist who undercalled AI, with no significant differences among the other conditions. Maximizers agreed with the radiologist who overcalled AI (β=0.82, SE 0.14; P&lt;.001) and disagreed with the radiologist who undercalled AI (β=-0.47, SE 0.14; P=.001). However, whereas minimizers disagreed with the radiologist who overcalled AI (β=-0.43, SE 0.18, P=.02), they did not significantly agree with the radiologist who undercalled AI (β=0.14, SE 0.17, P=.41).&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Conclusions: &lt;/strong&gt;Radiologists who recommend less testing than AI may face decrease","PeriodicalId":16337,"journal":{"name":"Journal of Medical Internet Research","volume":"27 ","pages":"e68823"},"PeriodicalIF":5.8,"publicationDate":"2025-05-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144127817","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":"Identifying Kidney Stone Risk Factors Through Patient Experiences With a Large Language Model: Text Analysis and Empirical Study.","authors":"Chao Mao, Jiaxuan Li, Patrick Cheong-Iao Pang, Quanjing Zhu, Rong Chen","doi":"10.2196/66365","DOIUrl":"https://doi.org/10.2196/66365","url":null,"abstract":"<p><strong>Background: </strong>Kidney stones, a prevalent urinary disease, pose significant health risks. Factors like insufficient water intake or a high-protein diet increase an individual's susceptibility to the disease. Social media platforms can be a valuable avenue for users to share their experiences in managing these risk factors. Analyzing such patient-reported information can provide crucial insights into risk factors, potentially leading to improved quality of life for other patients.</p><p><strong>Objective: </strong>This study aims to develop a model KSrisk-GPT, based on a large language model (LLM) to identify potential kidney stone risk factors from web-based user experiences.</p><p><strong>Methods: </strong>This study collected data on the topic of kidney stones on Zhihu in the past 5 years and obtained 11,819 user comments. Experts organized the most common risk factors for kidney stones into six categories. Then, we use the least-to-most prompting in the chain-of-thought prompting to enable GPT-4.0 to think like an expert and ask GPT to identify risk factors from the comments. Metrics, including accuracy, precision, recall, and F₁-score, were used to evaluate the performance of such a model.</p><p><strong>Results: </strong>Our proposed method outperforms other models in identifying comments containing risk factors with 95.9% accuracy and F₁-score, with a precision of 95.6% and a recall of 96.2%. Out of the 863 comments identified with risk factors, our analysis showed the most mentioned risk factors for kidney stones in Zhihu user discussions, mainly including dietary habits (high protein, high calcium intake), insufficient water intake, genetic factors, and lifestyle. In addition, new potential risk factors were discovered with GPT, such as excessive use of supplements like vitamin C and calcium, laxatives, and hyperparathyroidism.</p><p><strong>Conclusions: </strong>Comments from social media users offer a new data source for disease prevention and understanding patient journeys. Our method not only sheds light on using LLMs to efficiently summarize risk factors from social media data but also on LLMs' potential to identify new potential factors from the patient's perspective.</p>","PeriodicalId":16337,"journal":{"name":"Journal of Medical Internet Research","volume":"27 ","pages":"e66365"},"PeriodicalIF":5.8,"publicationDate":"2025-05-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144127816","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":"Temporal Association Between ChatGPT-Generated Diarrhea Synonyms in Internet Search Queries and Emergency Department Visits for Diarrhea-Related Symptoms in South Korea: Exploratory Study.","authors":"Jinsoo Kim, Ansun Jeong, Juseong Jin, Sangjun Lee, Do Kyoon Yoon, Soyeoun Kim","doi":"10.2196/65101","DOIUrl":"https://doi.org/10.2196/65101","url":null,"abstract":"<p><strong>Background: </strong>Diarrhea, a common symptom of gastrointestinal infections, can lead to severe complications and is a major cause of emergency department (ED) visits.</p><p><strong>Objective: </strong>This study explored the temporal association between internet search queries for diarrhea and its synonyms and ED visits for diarrhea-related symptoms.</p><p><strong>Methods: </strong>We used data from the National Emergency Department Information System (NEDIS) and NAVER (Naver Corporation), South Korea's leading search engine, from January 2017 to December 2021. After identifying diarrhea synonyms using ChatGPT, we compared weekly trends in relative search volumes (RSVs) for diarrhea, including its synonyms and weekly ED visits. Pearson correlation analysis and Granger causality tests were used to evaluate the relationship between RSVs and ED visits. We developed an Autoregressive Integrated Moving Average with Exogenous Variables (ARIMAX) model to further predict these associations. This study also examined the age-based distribution of search behaviors and ED visits.</p><p><strong>Results: </strong>A significant correlation was observed between the weekly RSV for diarrhea and its synonyms and weekly ED visits for diarrhea-related symptoms (ranging from 0.14 to 0.51, P<.05). Weekly RSVs for diarrhea synonyms, such as \"upset stomach,\" \"watery diarrhea,\" and \"acute enteritis,\" showed stronger correlations with weekly ED visits than weekly RSVs for the general term \"diarrhea\" (ranging from 0.20 to 0.41, P<.05). This may be because these synonyms better reflect layperson terminology. Notably, weekly RSV for \"upset stomach\" was significantly correlated with weekly ED visits for diarrhea and acute diarrhea at 1 and 2 weeks before the visit (P<.05). An ARIMAX model was developed to predict weekly ED visits based on weekly RSVs for diarrhea synonyms with lagged effects to capture their temporal influence. The age group of <50 years showed the highest activity in both web-based searches and ED visits for diarrhea-related symptoms.</p><p><strong>Conclusions: </strong>This study demonstrates that weekly RSVs for diarrhea synonyms are associated with weekly ED visits for diarrhea-related symptoms. By encompassing a nationwide scope, this study broadens the existing methodology for syndromic surveillance using ED data and provides valuable insights for clinicians.</p>","PeriodicalId":16337,"journal":{"name":"Journal of Medical Internet Research","volume":"27 ","pages":"e65101"},"PeriodicalIF":5.8,"publicationDate":"2025-05-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144127928","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":"Evaluation of the Aspects of Digital Interventions That Successfully Support Weight Loss: Systematic Review With Component Network Meta-Analysis.","authors":"Michael Nunns, Samantha Febrey, Rebecca Abbott, Jill Buckland, Rebecca Whear, Liz Shaw, Alison Bethel, Kate Boddy, Jo Thompson Coon, G J Melendez-Torres","doi":"10.2196/65443","DOIUrl":"10.2196/65443","url":null,"abstract":"&lt;p&gt;&lt;strong&gt;Background: &lt;/strong&gt;Obesity is a chronic complex disease associated with increased risks of developing several serious and potentially life-threatening conditions. It is a growing global health issue. Pharmacological treatment is an option for patients living with overweight or obesity. Digital technology may be leveraged to support patients with weight loss in the community, but it is unclear which of the multiple digital options are important for success.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Objective: &lt;/strong&gt;This systematic review and component network meta-analysis aimed to identify components of digital support for weight loss interventions that are most likely to be effective in supporting patients to achieve weight loss goals.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Methods: &lt;/strong&gt;We searched MEDLINE, Embase, APA PsycInfo, and Cochrane Central Register of Controlled Trials from inception to November 2023 for randomized controlled trials using any weight loss intervention with digital components and assessing weight loss outcomes in adults with BMI ≥25 kg/m&lt;sup&gt;2&lt;/sup&gt; (≥23 kg/m&lt;sup&gt;2&lt;/sup&gt; for Asian populations). Eligible trials were prioritized for synthesis based on intervention relevance and duration, and the target population. Trial arms with substantial face-to-face elements were deprioritized. Prioritized trials were assessed for quality using the Cochrane Risk of Bias Tool v1. We conducted intervention component analysis to identify key digital intervention features and a coding framework. All prioritized trial arms were coded using this framework and were included in component network meta-analysis.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Results: &lt;/strong&gt;Searches identified 6528 reports, of which 119 were included. After prioritization, 151 trial arms from 68 trials were included in the synthesis. Nine common digital components were identified from the 151 trial arms: provision of information or education, goal setting, provision of feedback, peer support, reminders, challenges or competitions, contact with a specialist, self-monitoring, and incentives or rewards. Of these, 3 components were identified as \"best bets\" because they were consistently and numerically, but not usually significantly, most likely to be associated with weight loss at 6 and 12 months. These were patient information, contact with a specialist, and incentives or rewards. An exploratory model combining these 3 components was significantly associated with successful weight loss at 6 months (-2.52 kg, 95% CI -4.15 to -0.88) and 12 months (-2.11 kg, 95% CI -4.25 to 0.01). No trial arms used this specific combination of components.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Conclusions: &lt;/strong&gt;Our findings indicate that the design of digital interventions to support weight loss should be carefully crafted around core components. On their own, no single digital component could be considered essential for success, but a combination of information, specialist contact, and incentives warrants further examination.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Trial registration: &lt;/","PeriodicalId":16337,"journal":{"name":"Journal of Medical Internet Research","volume":"27 ","pages":"e65443"},"PeriodicalIF":5.8,"publicationDate":"2025-05-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144127704","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":"Application of AI Chatbot in Responding to Asynchronous Text-Based Messages From Patients With Cancer: Comparative Study.","authors":"Xuexue Bai, Shiyong Wang, Yuanli Zhao, Ming Feng, Wenbin Ma, Xiaomin Liu","doi":"10.2196/67462","DOIUrl":"https://doi.org/10.2196/67462","url":null,"abstract":"&lt;p&gt;&lt;strong&gt;Background: &lt;/strong&gt;Telemedicine, which incorporates artificial intelligence such as chatbots, offers significant potential for enhancing health care delivery. However, the efficacy of artificial intelligence chatbots compared to human physicians in clinical settings remains underexplored, particularly in complex scenarios involving patients with cancer and asynchronous text-based interactions.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Objective: &lt;/strong&gt;This study aimed to evaluate the performance of the GPT-4 (OpenAI) chatbot in responding to asynchronous text-based medical messages from patients with cancer by comparing its responses with those of physicians across two clinical scenarios: patient education and medical decision-making.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Methods: &lt;/strong&gt;We collected 4257 deidentified asynchronous text-based medical consultation records from 17 oncologists across China between January 1, 2020, and March 31, 2024. Each record included patient questions, demographic data, and disease-related details. The records were categorized into two scenarios: patient education (eg, symptom explanations and test interpretations) and medical decision-making (eg, treatment planning). The GPT-4 chatbot was used to simulate physician responses to these records, with each session conducted in a new conversation to avoid cross-session interference. The chatbot responses, along with the original physician responses, were evaluated by a medical review panel (3 oncologists) and a patient panel (20 patients with cancer). The medical panel assessed completeness, accuracy, and safety using a 3-level scale, whereas the patient panel rated completeness, trustworthiness, and empathy on a 5-point ordinal scale. Statistical analyses included chi-square tests for categorical variables and Wilcoxon signed-rank tests for ordinal ratings.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Results: &lt;/strong&gt;In the patient education scenario (n=2364), the chatbot scored higher than physicians in completeness (n=2301, 97.34% vs n=2213, 93.61% for fully complete responses; P=.002), with no significant differences in accuracy or safety (P&gt;.05). In the medical decision-making scenario (n=1893), the chatbot exhibited lower accuracy (n=1834, 96.88% vs n=1855, 97.99% for fully accurate responses; P&lt;.001) and trustworthiness (n=860, 50.71% vs n=1766, 93.29% rated as \"Moderately trustworthy\" or higher; P&lt;.001) compared with physicians. Regarding empathy, the medical review panel rated the chatbot as demonstrating higher empathy scores across both scenarios, whereas the patient review panel reached the opposite conclusion, consistently favoring physicians in empathetic communication. Errors in chatbot responses were primarily due to misinterpretations of medical terminology or the lack of updated guidelines, with 3.12% (59/1893) of its responses potentially leading to adverse outcomes, compared with 2.01% (38/1893) for physicians.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Conclusions: &lt;/strong&gt;The GPT-4 chatbot performs comparably to physicians in patient ","PeriodicalId":16337,"journal":{"name":"Journal of Medical Internet Research","volume":"27 ","pages":"e67462"},"PeriodicalIF":5.8,"publicationDate":"2025-05-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144119843","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}