Mayo Clinic Proceedings. Digital health最新文献

{"title":"An Automated Mobile Cognitive Test for the Identification of Cognitive Impairment: A Cross-sectional Feasibility and Diagnostic Study","authors":"Louis Y. Tee MD, PhD ,&nbsp;Li Feng Tan MBBS ,&nbsp;Santhosh Seetharaman MBBS ,&nbsp;Lian Leng Low MBBS ,&nbsp;Zhi Peng Ong BS ,&nbsp;Munirah Bashil BS ,&nbsp;Hock Hai Teo PhD","doi":"10.1016/j.mcpdig.2025.100252","DOIUrl":"10.1016/j.mcpdig.2025.100252","url":null,"abstract":"<div><h3>Objective</h3><div>To develop Digital Processing Speed Test (DPST), a free, automated, multilingual, artificial intelligence–based cognitive testing application, with the aim to enhance recognition of cognitive impairment in underserved communities by leveraging mobile health to improve cognitive testing’s accessibility.</div></div><div><h3>Patients and Methods</h3><div>In this cross-sectional feasibility and diagnostic study, we determined the test performance of DPST for the identification of mild cognitive impairment (MCI) and dementia, compared with traditional cognitive tests, such as Mini-Mental State Examination (MMSE) and Montreal Cognitive Assessment (MoCA). The study was conducted from January 19, 2021, to November 12, 2023. In total, 476 adult participants were recruited by consecutive sampling at waiting areas of primary and secondary care clinics. The participants completed MMSE and MoCA with trained assessors and then performed DPST independently on a mobile device. The reference standard was a clinical diagnosis of MCI/dementia by a memory specialist blinded to the DPST score.</div></div><div><h3>Results</h3><div>Area under the receiver operating characteristic curve analyses showed that area under the curves were similar for the 3 tests (MMSE, 0.862; MoCA, 0.888; DPST, 0.861). Likewise, sensitivity (DPST, 85.2%; MMSE, 85.2%; MoCA, 90.2%), negative likelihood ratio (DPST, 0.197; MMSE, 0.193; MoCA, 0.129), specificity (DPST, 75.0%; MMSE, 76.5%; MoCA, 76.2%), and positive likelihood ratio (DPST, 3.41; MMSE, 3.62; MoCA, 3.79) were similar.</div></div><div><h3>Conclusion</h3><div>Digital Processing Speed Test, a free, automated, multilingual cognitive test conducted on a mobile device, has similar test performance to MMSE and MoCA. Nonetheless, DPST does not capture the multidomain cognitive deficits that characterize MCI/dementia. Moreover, test-retest reliability and interrater agreement of artificial intelligence–based handwriting recognition needs further confirmation.</div></div>","PeriodicalId":74127,"journal":{"name":"Mayo Clinic Proceedings. Digital health","volume":"3 3","pages":"Article 100252"},"PeriodicalIF":0.0,"publicationDate":"2025-07-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144756917","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Designing Artificial Intelligence-Powered Health Care Assistants to Reach Vulnerable Populations: A Discrete Choice Experiment Among South African University Students","authors":"Amy Zheng MPH ,&nbsp;Lawrence Long PhD ,&nbsp;Caroline Govathson MSc ,&nbsp;Candice Chetty-Makkan PhD ,&nbsp;Sarah Morris BS ,&nbsp;Dino Rech MBA ,&nbsp;Matthew P. Fox DSc ,&nbsp;Sophie Pascoe PhD","doi":"10.1016/j.mcpdig.2025.100248","DOIUrl":"10.1016/j.mcpdig.2025.100248","url":null,"abstract":"<div><h3>Objective</h3><div>To understand what preferences are important to university students in South Africa when engaging with a hypothetical artificial intelligence-powered health care assistant (AIPHA) to access health information using a discrete choice experiment.</div></div><div><h3>Patients and Methods</h3><div>We conducted an unlabeled, forced choice discrete choice experiment among adult South African university students through Prolific, an online research platform, from June 26, 2024 to August 31, 2024. Each choice option described a hypothetical AIPHA using 8 attribute characteristics (cost, confidentiality, security, health care topics, language, persona, access, and services). Participants were presented with 10 choice sets each comprised of 2 choice options and asked to choose between the 2. A conditional logit model was used.</div></div><div><h3>Results</h3><div>Three hundred participants were recruited and enrolled. Most participants were Black, born in South Africa, heterosexual, working for a wage, and had a mean age of 26.5 years (SD, 6.0). Language, security, and receiving personally tailored advice were the most important attributes for AIPHA. Participants strongly preferred the ability to communicate with the AIPHA in any South African language of their choosing instead of only English and receive information about health topics specific to their context including information on clinics geographically near them. The results were consistent when stratified by sex and socioeconomic status.</div></div><div><h3>Conclusion</h3><div>Participants had strong preferences for security and language, which is in line with previous studies where successful uptake and implementation of such health interventions clearly addressed these concerns. These results build the evidence base for how we might engage young adults in health care through technology effectively.</div></div>","PeriodicalId":74127,"journal":{"name":"Mayo Clinic Proceedings. Digital health","volume":"3 3","pages":"Article 100248"},"PeriodicalIF":0.0,"publicationDate":"2025-07-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144748792","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Comparing Machine Learning and Nurse Predictions for Hospital Admissions in a Multisite Emergency Care System","authors":"Jonathan Nover MBA, RN ,&nbsp;Matthew Bai MD ,&nbsp;Prem Tismina MS ,&nbsp;Ganesh Raut MS ,&nbsp;Dhavalkumar Patel MS ,&nbsp;Girish N. Nadkarni MD, MPH ,&nbsp;Benjamin S. Abella MD, MPhil ,&nbsp;Eyal Klang MD ,&nbsp;Robert Freeman DNP, RN, NE-BC","doi":"10.1016/j.mcpdig.2025.100249","DOIUrl":"10.1016/j.mcpdig.2025.100249","url":null,"abstract":"<div><h3>Objective</h3><div>To prospectively compare nurse predictions with a machine learning (ML) model for hospital admissions and evaluate whether adding the nurse prediction to ML outputs enhances predictive performance.</div></div><div><h3>Patients and Methods</h3><div>In this prospective, observational study at 6 hospitals in a large mixed quaternary/community emergency department (ED) system (annual ED census ∼500,000), triage nurses recorded a binary admission prediction for adult patients. These predictions were compared with an ensemble ML model (XGBoost + Bio-Clinical BERT) trained on structured data (demographics, vital signs, and medical history) and triage text. Nurse predictions were similarly analyzed and then integrated with the ML output to assess for improvement in predictive accuracy.</div></div><div><h3>Results</h3><div>The ensemble ML model (XGBoost + Bio-Clinical BERT) was trained on 1.8 million historical ED visits (January 2019 to December 2023). It was then tested on 46,912 prospective ED visits with recorded nurse predictions (September 1, 2024 to October 31, 2024). In the prospective arm, nurse predictions yielded an accuracy of 81.6% (95% CI, 81.3-81.9), a sensitivity of 64.8% (63.7-65.8), and a specificity of 85.7% (85.3-86.0). At a 0.30 probability threshold, the ML model attained an accuracy of 85.4% (85.0-85.7) and a sensitivity of 70.8% (69.8-71.7). Combining nurse predictions with the ML output did not improve accuracy beyond the model alone.</div></div><div><h3>Conclusion</h3><div>Machine learning-based predictions outperformed triage nurse estimates for hospital admissions. These findings suggest that an admission prediction system anchored by ML can perform reliably using data available at triage.</div></div>","PeriodicalId":74127,"journal":{"name":"Mayo Clinic Proceedings. Digital health","volume":"3 3","pages":"Article 100249"},"PeriodicalIF":0.0,"publicationDate":"2025-07-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144739648","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Youth Uptake of Digital Sexual and Reproductive Health Services Across Sociodemographic Groups (2018-2022): A Total Population Study from Stockholm, Sweden","authors":"Lovisa Hellsten MPH ,&nbsp;Viktor H. Ahlqvist PhD ,&nbsp;Anna M. Nielsen RM, PhD ,&nbsp;Gunnar Brandén PhD ,&nbsp;Anna Mia Ekström MD, PhD ,&nbsp;Kyriaki Kosidou MD, PhD","doi":"10.1016/j.mcpdig.2025.100251","DOIUrl":"10.1016/j.mcpdig.2025.100251","url":null,"abstract":"<div><h3>Objective</h3><div>To examine uptake of in-person and digital sexual and reproductive health (SRH) services among adolescents and young adults, quantify uptake across time, and explore whether the introduction of digital services affected the sociodemographic composition of users.</div></div><div><h3>Patients and Methods</h3><div>This Swedish total population study included all Stockholm residents aged 12-22 years between January 1st 2018 and December 31st 2022. The primary outcome was in-person or digital visits (chat and video) of SRH services within a year, identified using regional health care registries. Sociodemographic predictors included sex, age, migrant background, parental education, and household income, analyzed with repeated-measures multivariable regressions.</div></div><div><h3>Results</h3><div>Among the 454,405 individuals, 23.96% had at some point used SRH services (80.01% women) between 2018 and 2022. In-person visits remained the predominant mode of contact. Women had higher annual utilization rate of both in-person (women: 15.27%; 95% CI, 15.13-15.40; men: 1.75%; 95% CI, 1.72-1.78) and digital visits (women: 2.23%; 95% CI, 2.16-2.30; men: 0.12%; 95% CI, 0.11-0.13). Significantly lower uptake was also observed in the lowest income quintile (digital: adjusted odds ratio [aOR], 0.34; 95% CI, 0.31-0.36; in-person: aOR, 0.43; 95% CI, 0.42-0.45) compared with the highest quintile (reference group). Among digital visits, chat was more equitably used than video consultations across sociodemographic groups, including smaller differences between the highest and lowest income quintiles (chat: aOR, 0.59; 95% CI, 0.54-0.65; video: aOR, 0.25; 95% CI, 0.23-0.27). Only modest reductions in socioeconomic disparities were observed after the introduction of digital services.</div></div><div><h3>Conclusions</h3><div>Sociodemographic disparities in utilization were not alleviated by the introduction of digital visits; in-person users were also the primary digital users. Chat could be more equitable than video, but further research is needed.</div></div>","PeriodicalId":74127,"journal":{"name":"Mayo Clinic Proceedings. Digital health","volume":"3 3","pages":"Article 100251"},"PeriodicalIF":0.0,"publicationDate":"2025-07-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144724163","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Virtual Health Care in Hospital-at-Home Models for Patients With Acute Infections: A Scoping Review","authors":"Maria Normand Larsen MD ,&nbsp;Tatjana Sandreva Dreisig MD ,&nbsp;Maja Kjær Rasmussen MSc ,&nbsp;Anders N.Ø. Schultz MD ,&nbsp;Thyge Lynghøj Nielsen MD, PhD ,&nbsp;Thea K. Fischer MD, DMSc","doi":"10.1016/j.mcpdig.2025.100250","DOIUrl":"10.1016/j.mcpdig.2025.100250","url":null,"abstract":"<div><div>Given the imbalance between high care demand and strained hospital capacity, hospital-at-home (HaH) models offer a potential solution by providing hospital-level care in patients’ homes. This scoping review maps the literature on hospital-led virtual health care within HaH models for acute infections, focusing on intervention characteristics and evaluation designs. Following Johanna Briggs Institute guidelines and PRISMA-ScR, we included studies on virtual and hybrid HaH models using telemedicine for remote monitoring and interventions. The literature searches were performed from October 3, 2022 to October 22, 2022, and updated on July 11, 2024 and identified 15,062 potentially relevant records. From these, 79 studies met the inclusion criteria, highlighting the diversity of HaH models and their evaluations. Hybrid models provided broader treatment options, but many studies lacked detailed intervention descriptions, complicating implementation and meta-analyses. Most studies evaluated patient outcomes, with limited attention to health care staff and relatives. Nearly 45,000 participants were assessed, but only 254 participated in randomized controlled trials, indicating a need for more high-level evidence. Relevant gaps remain, including model heterogeneity and inconsistent reporting.</div></div>","PeriodicalId":74127,"journal":{"name":"Mayo Clinic Proceedings. Digital health","volume":"3 3","pages":"Article 100250"},"PeriodicalIF":0.0,"publicationDate":"2025-07-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144748793","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Predicting Tolerance to Anthracycline Chemotherapy Using Electrocardiogram-Based Artificial Intelligence in Sarcoma","authors":"Jack B. Korleski MD ,&nbsp;Regina M. Koch MD ,&nbsp;Thanh P. Ho MD ,&nbsp;Steven I. Robinson MBBS ,&nbsp;Scott H. Okuno MD ,&nbsp;Joerg Herrmann MD ,&nbsp;Brittany L. Siontis MD","doi":"10.1016/j.mcpdig.2025.100247","DOIUrl":"10.1016/j.mcpdig.2025.100247","url":null,"abstract":"<div><h3>Objective</h3><div>The objective of this study was to understand the utility of artificial intelligence-enabled electrocardiogram (AI-ECG) to assess the tolerability of anthracycline chemotherapy.</div></div><div><h3>Patients and Methods</h3><div>From December 18, 2006 to October 15, 2020, we identified adults with sarcoma who were treated with anthracycline chemotherapy at our institution who had an ECG within 1 year prior to treatment initiation. Utilizing previously defined AI-ECG nomograms, we obtained age and ejection fraction (EF) predictions. Changes in AI-ECG age were correlated with chemotherapy tolerance (the rates of dose reductions, treatment delays, and early discontinuation). We measured the sensitivity and specificity of the ECG to predict an EF of less than 50% or 35% prior to treatment and compared how changes in the AI-ECG EF prediction related to changes in echocardiography-based EF.</div></div><div><h3>Results</h3><div>Forty patients met the eligibility criteria. Sixty-eight percent of the patients were men. The median age was 56.5 years (18-76 years). We did not find differences in chemotherapy tolerance between patients who had an elevated or decreased ECG age. There was a trend `toward higher rates of dose reductions in patients with high ECG aging (odds ratio, 5.13; <em>P</em>=.32). The AI-ECG low EF prediction had a sensitivity of 100% and a specificity of 94% to isolate patients with an EF of less than 50% prior to treatment. Two patients’ EF decreased more than 10% after treatment, and both cases showed significant increases in the low EF prediction.</div></div><div><h3>Conclusion</h3><div>Overall, AI-based predictions on ECG tracings could be a way to monitor for decreases in EF during treatment with anthracycline chemotherapy. We recommend further studies to evaluate AI-ECG aging as a marker for chemotherapy tolerance.</div></div>","PeriodicalId":74127,"journal":{"name":"Mayo Clinic Proceedings. Digital health","volume":"3 3","pages":"Article 100247"},"PeriodicalIF":0.0,"publicationDate":"2025-07-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144686358","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Artificial Intelligence and Multi-Omics in Pharmacogenomics: A New Era of Precision Medicine","authors":"Mike Zack MD, PhD, MPH,&nbsp;Danil N. Stupichev MSc,&nbsp;Alex J. Moore BSc,&nbsp;Ioan D. Slobodchikov MSc,&nbsp;David G. Sokolov MSc,&nbsp;Igor F. Trifonov MSc,&nbsp;Allan Gobbs MSc","doi":"10.1016/j.mcpdig.2025.100246","DOIUrl":"10.1016/j.mcpdig.2025.100246","url":null,"abstract":"<div><div>Pharmacogenomics is entering a transformative phase as high-throughput “omics” techniques become increasingly integrated with state-of-the-art artificial intelligence (AI) methods. Although early successes in single-gene pharmacogenetics reported clear clinical benefits, many drug response phenotypes are governed by intricate networks of genomic variants, epigenetic modifications, and metabolic pathways. Multi-omics approaches address this complexity by capturing genomic, transcriptomic, proteomic, and metabolomic data layers, offering a comprehensive view of patient-specific biology. Advanced AI models, including deep neural networks, graph neural networks, and representation learning techniques, further enhance this landscape by detecting hidden patterns, filling gaps in incomplete data sets, and enabling in silico simulations of treatment responses. Such capabilities not only improve predictive accuracy but also deepen mechanistic insights, revealing how gene–gene and gene–environment interactions shape therapeutic outcomes. At the same time, real-world data from diverse patient populations is broadening the evidence base, underscoring the importance of inclusive datasets and population-specific algorithms to reduce health disparities. Despite challenges related to data harmonization, interpretability, and regulatory oversight, the synergy between multi-omics integration and AI-driven analytics holds relevant promise for revolutionizing clinical decision-making. In this review, we highlighted key technological advances, discussed current limitations, and outlined future directions for translating multi-omics plus AI innovations into routine personalized medicine.</div></div>","PeriodicalId":74127,"journal":{"name":"Mayo Clinic Proceedings. Digital health","volume":"3 3","pages":"Article 100246"},"PeriodicalIF":0.0,"publicationDate":"2025-06-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144662090","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Artificial Intelligence Image-Diagnosis for Female Genital Schistosomiasis","authors":"Jiayuan Zhu MSc ,&nbsp;J. Alison Noble DPhil ,&nbsp;Mireille Gomes DPhil","doi":"10.1016/j.mcpdig.2025.100245","DOIUrl":"10.1016/j.mcpdig.2025.100245","url":null,"abstract":"<div><h3>Objective</h3><div>To introduce a novel, artificial Intelligence (AI), deep learning-based application for automated diagnosis of female genital schistosomiasis (FGS), a disease estimated to affect around 56 million women and girls in sub-Saharan Africa.</div></div><div><h3>Patients and Methods</h3><div>This study focused on cervical images collected from a high endemic FGS area in Cameroon, from August 1, 2020 to August 31, 2021. We applied the You Only Look Once deep learning model and employed a 5-fold cross-validation approach, accompanied by sensitivity analysis, to optimize model performance.</div></div><div><h3>Results</h3><div>The model achieved a sensitivity of 0.96 (76/78) and an accuracy of 0.78 (97/125), demonstrating improved performance over an existing, non-AI-based, computerized image diagnostic method, which has a sensitivity of 0.94 (73/78) but an accuracy of 0.58 (73/125) on the same dataset. In addition, the AI model significantly reduced processing time, decreasing from 47 minutes to under 90 seconds for testing 250 images.</div></div><div><h3>Conclusion</h3><div>This study highlights the potential of deep learning-based models for automated diagnosis for FGS while reducing the reliance on specialized clinical expertise. It also underscores the need for further work to address current limitations of such AI-based methods for FGS diagnosis.</div></div>","PeriodicalId":74127,"journal":{"name":"Mayo Clinic Proceedings. Digital health","volume":"3 3","pages":"Article 100245"},"PeriodicalIF":0.0,"publicationDate":"2025-06-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144605925","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Healthy Heart Assistant, a WhatsApp-Based Generative Pretrained Transformer Technology, for Self-Care in Hypertensive Patients","authors":"Samuel E. Antia MD, MSc ,&nbsp;Collins N. Ugwu MD, MSc ,&nbsp;Vishal Ghodka BE ,&nbsp;Babangida S. Chori MSc ,&nbsp;Muhammad S. Nazir MD, MSc ,&nbsp;Chizoba A. Odili PhD ,&nbsp;Godsent C. Isiguzo MD, PhD ,&nbsp;Sri Vasireddy MS, MBA ,&nbsp;Augustine N. Odili MD, PhD","doi":"10.1016/j.mcpdig.2025.100243","DOIUrl":"10.1016/j.mcpdig.2025.100243","url":null,"abstract":"<div><h3>Objective</h3><div>To evaluate the feasibility, usability, and efficacy of innovative generative pretrained transformer chatbot in improving self-care in hypertensive patients in a resource-limited setting.</div></div><div><h3>Patients and Methods</h3><div>A single-arm nonblinded clinical trial was deployed in a busy cardiology clinic in a low-resource setting. Artificial intelligence–enabled chatbot (Healthy Heart Assistant) was activated in smartphones of 50 adults on treatment for hypertension. Participants were trained on how to use the Healthy Heart Assistant including setting medication and appointment reminders. Baseline questionnaires were administered at enrollment and 30 days later to explore acceptability, feasibility and usability of the bot. We used chatbot usability questionnaire and self-made Healthy Heart Assistant satisfaction questionnaire to assess bot usability and patients’ satisfaction, respectively. The study began on April 5, 2024, through July 15, 2024.</div></div><div><h3>Results</h3><div>Of 200 hypertensive clinic attendees, 70 (35%) had internet-enabled bot-compatible cell phones, of which 50 hypertensive patients were recruited to participate in the study. Among 50 participants, 2 (4%) were lost to follow-up; 19 (39.6%) were women; and 40 (83.3%) had attained tertiary level of education. Mean time of training to use bot was 5.7 minutes, with 35 (70.8%) of participants being able to use the bot within 5 minutes. The median frequency of chats for participants within the timeframe was an average of 1.5 chats/day. Chatbot usability questionnaire score was 69.5%, whereas self-made Healthy Heart Assistant satisfaction questionnaire score was 90%.</div></div><div><h3>Conclusion</h3><div>This proof-of-concept study shows that generative artificial intelligence can be applied with reasonable success in hypertension self-care in low-resource settings and has potential for being effective.</div></div>","PeriodicalId":74127,"journal":{"name":"Mayo Clinic Proceedings. Digital health","volume":"3 3","pages":"Article 100243"},"PeriodicalIF":0.0,"publicationDate":"2025-06-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144549292","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Artificial Intelligence in Digital Self-Diagnosis Tools: A Narrative Overview of Reviews","authors":"Aikaterini Mentzou PhD ,&nbsp;Amy Rogers MD ,&nbsp;Edzia Carvalho PhD ,&nbsp;Angela Daly PhD ,&nbsp;Maeve Malone HDip ,&nbsp;Xaroula Kerasidou PhD","doi":"10.1016/j.mcpdig.2025.100242","DOIUrl":"10.1016/j.mcpdig.2025.100242","url":null,"abstract":"<div><div>Digital self-diagnosis tools, or symptom checkers, many of which incorporate artificial intelligence technology, are intended to provide diagnostic information and triage advice to lay users. This narrative overview of reviews explores the common themes and issues raised by existing evidence synthesis literature on these tools to establish a common ground for interdisciplinary research. We searched 3 bibliographic databases (PubMed, Scopus and Web of Science) and Google Scholar using keyword combinations of <em>artificial</em>, <em>self-diagnosis</em>, <em>intelligence</em>, and <em>machine learning</em> for publications from 2019 to 2023. We included systematic reviews, meta-analyses, scoping reviews, narrative syntheses, and opinion pieces that discussed tools where users proactively entered personal health information to acquire a predicted diagnosis of their symptoms or triage advice. This overview reveals significant gaps in understanding the key areas of development, implementation, impact, and oversight of digital self-diagnosis tools. Additionally, the terminology used to describe these tools and their underlying technologies varies widely, encompassing technologies ranging from simple branching logic algorithms to complex deep neural networks. Our interdisciplinary analysis identified gaps and critical areas for future research across all stages of the lifecycles of these tools. The diverse challenges uncovered highlight the necessity for multiagency and multidisciplinary efforts promoting responsible development and implementation.</div></div>","PeriodicalId":74127,"journal":{"name":"Mayo Clinic Proceedings. Digital health","volume":"3 3","pages":"Article 100242"},"PeriodicalIF":0.0,"publicationDate":"2025-06-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144534516","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}