JMIR Formative Research最新文献

Background: Maternal mental health disorders are prevalent, yet many individuals do not receive adequate support due to stigma, financial constraints, and limited access to care. Digital interventions, particularly chatbots, have the potential to provide scalable, low-cost support, but few are tailored specifically to the needs of perinatal individuals.

Objective: This study aimed to (1) design and develop Moment for Parents, a tailored chatbot for perinatal mental health education and support, and (2) assess usability through engagement, usage patterns, and user experience.

Methods: This study used a human-centered design to develop Moment for Parents, a rules-based chatbot to support pregnant and postpartum individuals. In phase 1, ethnographic interviews (n=43) explored user needs to inform chatbot development. In phase 2, a total of 108 pregnant and postpartum individuals were recruited to participate in a pilot test and had unrestricted access to the chatbot. Engagement was tracked over 8 months to assess usage patterns and re-engagement rates. After 1 month, participants completed a usability, relevance, and satisfaction survey, providing key insights for refining the chatbot.

Results: Key themes that came from the ethnographic interviews in phase 1 included the need for trusted resources, emotional support, and better mental health guidance. These insights informed chatbot content, including mood-based exercises and coping strategies. Re-engagement was high (69/108, 63.9%), meaning users who had stopped interacting for at least 1 week returned to the chatbot at least once. A large proportion (28/69, 40.6%) re-engaged 3 or more times. Overall, 28/30 (93.3%) found the chatbot relevant for them, though some noted repetitive content and limited response options.

Conclusions: The Moment for Parents chatbot successfully engaged pregnant and postpartum individuals with higher-than-typical retention and re-engagement patterns. The findings underscore the importance of flexible, mood-based digital support tailored to perinatal needs. Future research should examine how intermittent chatbot use influences mental health outcomes and refine content delivery to enhance long-term engagement and effectiveness.

Background: Wearable sensor bracelets have gained interest for their ability to detect symptomatic and presymptomatic infections through alterations in physiological indicators. Nevertheless, the use of these devices for public health surveillance among attendees of large-scale events such as hajj, the Islamic religious mass gathering held in Saudi Arabia, is currently in a nascent phase.

Objective: This study aimed to explore hajj stakeholders' perspectives on the use of wearable sensor bracelets for disease detection.

Methods: We conducted a qualitative, theoretically informed, interview-based study from March 2022 to October 2023 involving a diverse sample of hajj stakeholders, including technology experts, health care providers, and hajj service providers. The study was guided by the task-technology fit model and the unified theory of acceptance and use of technology to provide a comprehensive understanding of the factors influencing the acceptance and use of the technology. Semistructured in-depth interviews were used to capture perspectives on using wearable sensor bracelets for infectious disease detection during hajj. Thematic analysis of interview transcripts was conducted.

Results: A total of 14 individuals were interviewed. In total, 4 main themes and 13 subthemes emerged from the study, highlighting crucial challenges, considerations, recommendations, and opportunities in the use of wearable sensor bracelets for the presymptomatic detection of infectious diseases during hajj. Implementing wearable sensor bracelets for disease detection during hajj faces obstacles from multiple perspectives, encompassing users, implementing stakeholders, and technological factors. Hajj stakeholders were concerned about the substantial financial and operational barriers. The motivation of implementing stakeholders and users is essential for the acceptance and uptake of devices during hajj. Successful integration of wearables into the hajj surveillance system depends on several factors, including infrastructure, device features, suitable use cases, training, and a smooth organizational integration process.

Conclusions: This study provides valuable insights into the potential opportunities and challenges of adopting wearable sensor bracelets for disease detection during hajj. It offers essential factors to consider and important suggestions to enhance comprehension and ensure the effective implementation of this technology.

Background: Graves disease (GD) is an autoimmune thyroid disorder characterized by hyperthyroidism and autoantibodies. The COVID-19 pandemic has raised questions about its potential relationship with autoimmune diseases like GD.

Objective: This study aims to investigate the causal association between COVID-19 and GD through Mendelian randomization (MR) analysis and assess the impact of COVID-19 on GD.

Methods: We conducted an MR study using extensive genome-wide association study data for GD and COVID-19 susceptibility and its severity. We used stringent single nucleotide polymorphism selection criteria and various MR methodologies, including inverse-variance weighting, MR-Egger, and weighted median analyses, to assess causal relationships. We also conducted tests for directional pleiotropy and heterogeneity, as well as sensitivity analyses.

Results: The MR analysis, based on the largest available dataset to date, did not provide evidence supporting a causal relationship between COVID-19 susceptibility (odds ratio [OR] 0.989, 95% CI 0.405-2.851; P=.93), COVID-19 hospitalization (OR 0.974, 95% CI 0.852-1.113; P=.70), COVID-19 severity (OR 0.979, 95% CI 0.890-1.077; P=.66), and GD. Tests for directional pleiotropy and heterogeneity, as well as sensitivity analyses, supported these findings.

Conclusions: This comprehensive MR study does not provide sufficient evidence to support a causal relationship between COVID-19 and the onset or exacerbation of GD. These results contribute to a better understanding of the potential association between COVID-19 and autoimmune diseases, alleviating concerns about a surge in autoimmune thyroid diseases due to the pandemic. Further research is warranted to explore this complex relationship thoroughly.

Background: Popularized by ChatGPT, large language models (LLMs) are poised to transform the scalability of clinical natural language processing (NLP) downstream tasks such as medical question answering (MQA) and automated data extraction from clinical narrative reports. However, the use of LLMs in the health care setting is limited by cost, computing power, and patient privacy concerns. Specifically, as interest in LLM-based clinical applications grows, regulatory safeguards must be established to avoid exposure of patient data through the public domain. The use of open-source LLMs deployed behind institutional firewalls may ensure the protection of private patient data. In this study, we evaluated the extraction performance of a locally deployed LLM for automated MQA from surgical pathology reports.

Objective: We compared the performance of human reviewers and a locally deployed LLM tasked with extracting key histologic and staging information from surgical pathology reports.

Methods: A total of 84 thyroid cancer surgical pathology reports were assessed by two independent reviewers and the open-source FastChat-T5 3B-parameter LLM using institutional computing resources. Longer text reports were split into 1200-character-long segments, followed by conversion to embeddings. Three segments with the highest similarity scores were integrated to create the final context for the LLM. The context was then made part of the question it was directed to answer. Twelve medical questions for staging and thyroid cancer recurrence risk data extraction were formulated and answered for each report. The time to respond and concordance of answers were evaluated. The concordance rate for each pairwise comparison (human-LLM and human-human) was calculated as the total number of concordant answers divided by the total number of answers for each of the 12 questions. The average concordance rate and associated error of all questions were tabulated for each pairwise comparison and evaluated with two-sided t tests.

Results: Out of a total of 1008 questions answered, reviewers 1 and 2 had an average (SD) concordance rate of responses of 99% (1%; 999/1008 responses). The LLM was concordant with reviewers 1 and 2 at an overall average (SD) rate of 89% (7%; 896/1008 responses) and 89% (7.2%; 903/1008 responses). The overall time to review and answer questions for all reports was 170.7, 115, and 19.56 minutes for Reviewers 1, 2, and the LLM, respectively.

Conclusions: The locally deployed LLM can be used for MQA with considerable time-saving and acceptable accuracy in responses. Prompt engineering and fine-tuning may further augment automated data extraction from clinical narratives for the provision of real-time, essential clinical insights.

Background: HIV index case testing (ICT) aims to identify people living with HIV and their contacts, engage them in HIV testing services, and link them to care. ICT implementation has faced challenges in Malawi due to limited counseling capacity among lay health care workers (HCWs). Enhancing capacity through centralized face-to-face training is logistically complex and expensive. A decentralized blended learning approach to HCW capacity-building, combining synchronous face-to-face and asynchronous digital modalities, may be an acceptable way to address this challenge.

Objective: The objective of this analysis is to describe factors influencing HCW anticipated acceptability of blended learning using the Technology Acceptance Model (TAM).

Methods: This formative qualitative study involved conducting 26 in-depth interviews with HCWs involved in the ICT program across 14 facilities in Machinga and Balaka, Malawi (November-December 2021). Results were analyzed thematically using TAM. Themes were grouped into factors affecting the 2 sets of TAM constructs: perceived usefulness and perceived ease of use.

Results: A total of 2 factors influenced perceived usefulness. First, HCWs found the idea of self-guided digital learning appealing, as they believed it would allow for reinforcement, which would facilitate competence. They also articulated the need for opportunities to practice and receive feedback through face-to-face interactions in order to apply the digital components. In total, 5 factors influenced perceived ease of use. First, HCWs expressed a need for orientation to the digital technology given limited digital literacy. Second, they requested accessibility of devices provided by their employer, as many lacked personal devices. Third, they wished for adequate communication surrounding their training schedules, especially if they were going to be asynchronous. Fourth, they wished for support for logistical arrangements to avoid work interruptions. Finally, they wanted monetary compensation to motivate learning, a practice comparable with offsite trainings.

Conclusions: A decentralized blended learning approach may be an acceptable method of enhancing ICT knowledge and skills among lay HCWs in Malawi, although a broad range of external factors need to be considered. Our next step is to integrate these findings into a blended learning package and examine perceived acceptability of the package in the context of a cluster randomized controlled trial.