JMIR Formative Research最新文献

Background: Commercial wearable devices allow for continuous heart rate (HR) monitoring in daily life. Their accuracy under ecologically valid conditions, however, remains insufficiently independently tested, especially during irregular activity, cognitive stress, and variable climates.

Objective: This study evaluated the HR accuracy of 10 commercially available wearables under controlled variations in physical activity, cognitive stress, and temperature. We hypothesized that physical activity irregularity, cognitive stress, and thermal climate conditions would affect measurement accuracy.

Methods: Forty-five healthy adults (21-68, mean 34, SD 12 y) completed a standardized protocol in climate-controlled chambers simulating neutral (23 °C), hot (36 °C), and cold (10 °C) conditions. Tasks included rest, cognitive stress (Montreal Imaging Stress Task), steady walking, and intermittent walking. Each of the 10 devices (Fitbit Charge 6, Fitbit Inspire 3, Garmin Vivosmart 5, Garmin Vivoactive 5, Apple Watch SE, Google Pixel Watch 2, Polar Ignite 3, Polar Pacer, Xiaomi Watch 2, and Oura Ring Gen 3) was compared against electrocardiogram-derived HR from a Zephyr BioHarness chest strap. Accuracy was assessed using mean absolute error (MAE), mean absolute percentage error (MAPE), repeated-measures concordance correlation coefficient (CCC), and Bland-Altman analysis.

Results: Significant variability across the devices was observed. Fitbit Charge 6 (MAE 4.5 bpm, MAPE 5.5%, CCC 0.93) and Google Pixel Watch 2 (MAE 4.9 bpm, MAPE 6.7%, CCC 0.87) showed strong agreement with the gold standard. In contrast, Fitbit Inspire 3, Polar Ignite 3, Polar Pacer, and Oura Ring displayed larger errors (MAE 9-14 bpm, MAPE 11%-16%) and lower CCC values (0.45-0.66). The climate conditions did not significantly affect the measurement accuracy of the test devices. The activity type, however, did have a significant effect: intermittent walking increased errors for multiple devices.

Conclusions: Wearable HR measurement accuracy is device-specific and context-dependent. Moderate climates did not impair performance, but irregular movement reduced accuracy. Fitbit Charge 6 and Google Pixel Watch 2 demonstrated the highest reliability, supporting their use in health and sports monitoring. Careful device selection and context-aware interpretation remain critical for applied and clinical applications.

Background: The development of efficient, scalable, and precise tools to assess knowledge of evidence-based parenting strategies is critical, particularly as increased parenting knowledge is a core target of many intervention programs.

Objective: This study aimed to develop and evaluate a computerized adaptive testing version of the Knowledge of Effective Parenting Test-Internalizing module (KEPT-I CAT).

Methods: Using computerized adaptive testing simulations from a large (n=1000) national dataset, we compared the performance of the KEPT-I CAT to both the full-length Knowledge of Effective Parenting Test-Internalizing module and a 10-item static short form (KEPT-I Brief).

Results: Results indicated that the KEPT-I CAT achieved comparable efficiency to the KEPT-I Brief (10 items), while demonstrating superior psychometric properties and modestly reducing the potential for practice effects.

Conclusions: Given these advantages, the KEPT-I CAT is well-suited for post-intervention assessment and may facilitate research examining how increases in parenting knowledge relate to changes in behavior and reductions in child internalizing symptoms.

Background: Health care providers (HCPs) in public health facilities in low- and middle-income countries, including Nepal, often lack adequate training to manage mental health problems effectively.

Objective: This study evaluated the impact of structured mental health training on the knowledge, attitudes, confidence, and psychosocial support skills of nonspecialist HCPs in Madhesh Province, Nepal.

Methods: This study is a nested substudy within a larger domestic violence (DV) intervention trial and used a mixed method, pre-post intervention design with a comparison group. A total of 46 nonspecialist HCPs were randomized into 2 groups: group 1 (n=24) received a 10-day comprehensive mental health and violence prevention training; group 2 (n=22) received a 3-day training focused on ethical considerations, the link between intimate partner violence (IPV) or DV and mental health, and available referral services. The training was based on the World Health Organization's Problem Management Plus model, with augmented modules on safety planning and psychosocial support. Changes in knowledge and attitude scores were assessed at baseline, immediately post-training, and at 3-month follow-up. In-depth interviews with participants from group 1 were thematically analyzed.

Results: At baseline, nearly 90% of nonspecialist HCPs had not received any prior formal mental health training. Both groups demonstrated significant improvements in mental health knowledge, with a greater increase observed in group 1 (mean score 41.33-48.41) compared to group 2 (41.18-44.27). Attitudes toward individuals with mental health problems also improved in both groups, reflected in reductions in social distance and perceived dangerousness scores. Thematic analysis of interviews indicated enhanced confidence and psychosocial support skills, particularly in managing mental health concerns among women experiencing IPV or DV.

Conclusions: Structured mental health training significantly improved both knowledge and attitudes among nonspecialist HCPs in public health facilities in Madhesh Province. Participants also reported increased confidence in addressing common mental health concerns. This training model has potential for scale-up in other resource-limited settings to build frontline capacity in managing mental health problems and supporting women experiencing IPV or DV.

Background: Lately, big data studies have shown promise in using patient characteristics to rank the likelihood of retention of antiseizure medications (ASMs), a measure indicating tolerability as well as effect. How such results can be integrated into clinical practice has yet to be studied. We developed EPstat, a noncommercial tool that provides physicians with real-world treatment retention data from 33,998 patients with epilepsy.

Objective: This study investigated the user experience of EPstat after its pilot launch.

Methods: EPstat was developed in an iterative process with first a prototype and then a final version accessible on the health care region intranet. EPstat was launched in 2022 through emails and information meetings at neurology departments. After 1 year, an online questionnaire was distributed to physicians in our health service region's neurology clinics (5 hospitals). Descriptive statistics and thematic analysis were used to summarize responses. To supplement the survey, 3 semistructured workshops or group interviews with neurologists and residents were used to gather further feedback.

Results: Of the 27 survey respondents, 19 (70%) were aware of EPstat and 10 (37%) had used it. Users rated EPstat highly for ease of use (median 5, IQR 4-5) and applicability in clinical practice (median 4, IQR 4-4). Two of the 10 respondents who had used it indicated that the platform had influenced their choice of ASM. Workshop participants advocated for expanding the platform to include retention data on newer ASMs and general information relevant to epilepsy management.

Conclusions: The notion of using big data to improve ASM selection was well received. However, there were barriers to the initial use, and users requested a more comprehensive resource that also incorporated other information related to epilepsy. EPstat is now being updated with more recent ASM statistics, including information on newer ASMs. Mobile access, more information for physicians, and mentioning the tool in regional guidelines are some possible measures to increase use. Linking multinational statistics could also increase the precision of the presented data and, thus, increase usefulness. Study of EPstat will continue and should include thematic analysis of representative and rigorously sampled workshop participants. Such studies are also likely to provide information on how physicians and health services receive web-based tools, which are likely to soon be driven by artificial intelligence. In similar projects, we recommend greater participatory involvement of both health care providers and patients already at the design stage.