Background: High staff turnover is a widespread issue across nearly all hospital departments, often exceeding 20% annually. This constant flux disrupts continuity of care and creates a recurring challenge: how to rapidly integrate new employees into complex clinical environments, both physically and functionally. Traditional onboarding methods struggle to meet this demand, particularly in services operating 24/7, such as emergency departments (EDs).
Objective: This formative study presents the design and implementation of a web-based 3D gamified simulation platform aimed at improving staff onboarding in clinical environments. The paper outlines both the technical architecture-with guidance for hospital IT departments-and the acceptability and usability for permanent staff, who play a key role in ensuring onboarding continuity. We sought to assess whether such a tool could be autonomously managed and well received by health care professionals.
Methods: The intervention consisted of 2 linked components: a real-time, browser-based 3D simulation replicating the hospital's ED and a web-based quest editor allowing nontechnical staff to update training content. The system supports self-paced onboarding through location-based tasks, object searches, quizzes, and simulated staff interactions. Two preliminary usability studies were conducted: one with 37 ED staff members testing the 3D simulation and another with 9 users exploring the quest editor. Feedback was gathered through anonymous questionnaires and a descriptive analysis.
Results: Early results showed high feasibility and acceptability. Among 3D simulation testers (n=37), 90% (33/37) found the tool helpful for understanding the department's structure, and 81% (30/37) believed it would be useful for new staff. The inclusion of personal anecdotes and gamified tasks was viewed as engaging and motivating. The quest editor (n=9) was positively rated by 91% (8/9) of users, who appreciated the ability to autonomously update content without IT support. These findings support the dual promise of the platform (ie, pedagogical flexibility and technical sustainability).
Conclusions: This work demonstrates the feasibility of a gamified simulation platform designed for high-turnover clinical environments. It highlights both the operational deployment framework and the early acceptability among key staff members. While further validation with actual new hires is needed, this formative study shows promising potential for generalization beyond emergency care. The modular and editable nature of the system makes it a viable solution for scalable onboarding in other hospital departments.
Background: Physical activity is a simple, low-risk intervention that could be integrated into daily life to improve glycemic control in individuals with prediabetes and early-stage type 2 diabetes mellitus (T2DM). However, maintaining physical activity remains challenging, even when its benefits are well understood. Although digital peer support has the potential to promote and maintain physical activity, its effectiveness has not yet been sufficiently established.
Objective: This study examined the impact of a digital peer-supported app on daily step goal achievement and average daily step counts among individuals with prediabetes and early-stage T2DM.
Methods: This 3-month, prospective, nonrandomized controlled trial recruited participants aged 40-79 years with prediabetes or early-stage T2DM. The participants were divided into a digital peer-supported app group and a control group. The digital peer-supported app group tracked their daily steps, shared their progress with small peer groups, and received real-time feedback and support within the app. The control group tracked their steps individually using a pedometer. The primary outcome was the achievement rate of daily step goals. Secondary outcomes included the average daily step count, BMI, glycosylated hemoglobin A1c level, blood pressure, and self-reported lifestyle behaviors.
Results: A total of 32 participants (digital peer-supported app group: n=18 and control group: n=14) completed the study. The digital peer-supported app group reported a significantly higher median daily step goal achievement rate (57.2%, IQR 32.2%-90% vs 26.7%, IQR 10%-64.4%; P=.04) and daily step count (6854, IQR 4846-10388 steps vs 3946, IQR 3176-6832 steps; P<.03) compared to the control group. No significant differences were observed in glycosylated hemoglobin A1c levels, blood pressure, BMI, or lifestyle behaviors.
Conclusions: Our findings inform research in this field by suggesting that a digital peer-supported app may support daily step goal achievement and increase step counts among individuals with prediabetes and early-stage T2DM over the 3-month study period. The digital peer-supported app facilitated real-time feedback, peer approval, and continuous engagement to support participation in light physical activity.
Trial registration: UMIN-CTR UMIN000039466; https://center6.umin.ac.jp/cgi-open-bin/ctr_e/ctr_view.cgi?recptno=R000044999.
Unlabelled: Artificial intelligence (AI) has the capacity to transform health care by improving clinical decision-making, optimizing workflows, and enhancing patient outcomes. However, this potential remains limited by a complex set of technological, human, and ethical barriers that constrain its safe and equitable implementation. This paper argues for a holistic, systems-based approach to AI integration that addresses these challenges as interconnected rather than isolated. It identifies key technological barriers, including limited explainability, algorithmic bias, integration and interoperability issues, lack of generalizability, and difficulties in validation. Human factors such as resistance to change, insufficient stakeholder engagement, and education and resource constraints further impede adoption, whereas ethical and legal challenges related to liability, privacy, informed consent, and inequity compound these obstacles. Addressing these issues requires transparent model design, diverse datasets, participatory development, and adaptive governance. Recommendations emerging from this synthesis are as follows: (1) establish standardized international regulatory and governance frameworks; (2) promote multidisciplinary co-design involving clinicians, developers, and patients; (3) invest in clinician education, AI literacy, and continuous training; (4) ensure equitable resource allocation through dedicated funding and public-private partnerships; (5) prioritize multimodal, explainable, and ethically aligned AI development; and (6) focus on long-term evaluation of AI in real-world settings to ensure adaptive, transparent, and inclusive deployment. Adopting these measures can align innovation with accountability, enabling health care systems to harness AI's transformative potential responsibly and sustainably to advance patient care and health equity.
Background: South Korea has the highest suicide rate among the Organisation for Economic Co-operation and Development nations, with particularly elevated figures among persons with disabilities. Research has shown a strong correlation between suicidal ideation and suicide attempts.
Objective: This study aimed to investigate the factors that contribute to suicidal ideation among persons with disabilities in South Korea, utilizing machine learning methods based on national survey data.
Methods: We employed data from the 2020 National Survey on Persons with Disabilities in South Korea, which included 7025 respondents. The primary variable of interest was the answer to the question, "have you thought about taking your own life in the past year?" The dataset was divided into training (80%) and test (20%) subsets. Because the survey contained too many questions (n=1394), feature selection was conducted using random forest variable importance to identify the top 100 features. Subsequently, 5 machine learning models were trained to predict suicidal ideation based on the selected features: logistic regression, support vector machine, random forest, Extreme Gradient Boosting (XGBoost), and feed-forward neural network.
Results: A total of 6832 persons with disabilities responded to the suicidal ideation question and were included in the study. The most common types of primary disability were physical disability (n=1773, 26.0%) and hearing disability (n=979, 14.3%). Of the 6832 persons with disabilities, 12.1% (n=829) indicated they had had suicidal thoughts in the past year. Significant factors that impacted suicidal ideation included intense feelings of sadness, difficulties associated with their disabilities, and overall health satisfaction. Among the models tested, the random forest model exhibited the best predictive performance with a median area under the receiver operating characteristic curve of 0.905 (IQR 0.895-0.913), a median precision of 0.592 (IQR 0.561-0.616), and a median recall of 0.588 (IQR 0.564-0.620).
Conclusions: This study highlights critical predictors of suicidal ideation in persons with disabilities in South Korea, underscoring the necessity for focused mental health interventions. The results demonstrate the potential of machine learning to identify these factors, which can aid in the development of future suicide prevention strategies. Future work is warranted to investigate if the factors identified in this study are causal.
Background: Experience sampling methodology (ESM) is an assessment method used in psychosis research. Symptom severity and gender may be associated with ESM engagement. Exploring qualitative experiences of using ESM among people with psychosis should aid developing more relevant, accessible digital assessments.
Objective: This study aimed to examine factors that could affect engagement with ESM, such as associations of completion rates with age, ethnicity, gender, and clinical severity. It also aimed to explore qualitatively service users' experiences of using this data collection method.
Methods: Data from 134/207 AVATAR2 trial (ISRCTN55682735) participants were used to evaluate associations between demographic variables, symptom severity, and ESM completion rates. Trial participants were purposively sampled to participate in an interview to discuss their experiences of using ESM or to discuss reasons why they chose not to use it.
Results: Multiple regression analyses of 134 participants found that age, gender, ethnicity, and clinical severity were not associated with ESM completion rates (F5,128=0.548; P=.74). A thematic analysis of 17 participant interviews found 3 overarching themes: Factors affecting engagement with ESM, Perceived benefits of ESM, and Suggestions for improvement. These themes described how ESM has multiple benefits for people with psychosis, including increasing knowledge and awareness of mental health. ESM was straightforward and easy to use; however, engaging in other activities, experiencing positive symptoms, little experience using technology, and trial involvement impacted engagement. Participant's decision to use ESM could be influenced by concerns about security and privacy.
Conclusions: Recommendations are made on how engagement with ESM can be improved, making it easier to use this method with this population, including providing increased support or training when using digital-based assessment or intervention as well as providing information on how digital data are used and recorded.
Background: Early identification of the etiology of spontaneous intracerebral hemorrhage (ICH) could significantly contribute to planning a suitable treatment strategy. A notable radiomics-based artificial intelligence (AI) model for classifying causes of spontaneous ICH from brain computed tomography scans has been previously proposed.
Objective: This study aimed to externally validate and assess the utility of this AI model.
Methods: This study used 69 computed tomography scans from a separate cohort to evaluate the AI model's performance in classifying nontraumatic ICHs into primary, tumorous, and vascular malformation related. We also assessed the accuracy, sensitivity, specificity, and positive predictive value of clinicians, radiologists, and trainees in identifying the ICH causes before and after using the model's assistance. The performances were statistically analyzed by specialty and expertise levels.
Results: The AI model achieved an overall accuracy of 0.65 in classifying the 3 causes of ICH. The model's assistance improved overall diagnostic performance, narrowing the gap between nonradiology and radiology groups, as well as between trainees and experts. The accuracy increased from 0.68 to 0.72, from 0.72 to 0.76, from 0.69 to 0.74, and from 0.72 to 0.75 for nonradiologists, radiologists, trainees, and specialists, respectively. With the model's support, radiology professionals demonstrated the highest accuracy, highlighting the model's potential to enhance diagnostic consistency across different levels.
Conclusions: When applied to an external dataset, the accuracy of the AI model in categorizing spontaneous ICHs based on radiomics decreased. However, using the model as an assistant substantially improved the performance of all reader groups, including trainees and radiology and nonradiology specialists.
Background: The application of large language models (LLMs) in medicine is rapidly advancing. However, evaluating LLM capabilities in specialized domains such as traditional Chinese medicine (TCM), which possesses a unique theoretical system and cognitive framework, remains a sizable challenge.
Objective: This study aimed to provide an empirical evaluation of different LLM types in the specialized domain of TCM stroke.
Methods: The Traditional Chinese Medicine-Stroke Evaluation Dataset (TCM-SED), a 203-question benchmark, was systematically constructed. The dataset includes 3 paradigms (short-answer questions, multiple-choice questions, and essay questions) and covers multiple knowledge dimensions, including diagnosis, pattern differentiation and treatment, herbal formulas, acupuncture, interpretation of classical texts, and patient communication. Gold standard answers were established through a multiexpert cross-validation and consensus process. The TCM-SED was subsequently used to comprehensively test 2 representative LLM models: GPT-4o (a leading international general-purpose model) and DeepSeek-R1 (a large model primarily trained on Chinese corpora).
Results: The test results revealed a differentiation in model capabilities across cognitive levels. In objective sections emphasizing precise knowledge recall, DeepSeek-R1 comprehensively outperformed GPT-4o, achieving an accuracy lead of more than 17% in the multiple-choice section (96/137, 70.1% vs 72/137, 52.6%, respectively). Conversely, in the essay section, which tested knowledge integration and complex reasoning, GPT-4o's performance notably surpassed that of DeepSeek-R1. For instance, in the interpretation of classical texts category, GPT-4o achieved a scoring rate of 90.5% (181/200), far exceeding DeepSeek-R1 (147/200, 73.5%).
Conclusions: This empirical study demonstrates that Chinese-centric models have a substantial advantage in static knowledge tasks within the TCM domain, whereas leading general-purpose models exhibit stronger dynamic reasoning and content generation capabilities. The TCM-SED, developed as the benchmark for this study, serves as an effective quantitative tool for evaluating and selecting appropriate LLMs for TCM scenarios. It also offers a valuable data foundation and a new research direction for future model optimization and alignment.
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：