Health Informatics Journal最新文献

Background: Lupus nephritis (LN) flares raise the risks of renal failure and mortality in systemic lupus erythematosus (SLE) patients, making risk stratification and individualized care crucial. Our goal was to develop machine learning (ML) models to predict LN flares.

Methods: A total of 1546 SLE patients were enrolled from a hospital-based cohort. Electronic health record (EHR), single nucleotide polymorphism (SNP), and polygenic risk score (PRS) were combined to construct ML models. SHapley Additive exPlanation (SHAP) values were calculated to assess each feature's contribution.

Results: Within 5 years, 448 patients developed LN. Of the 686,354 SNPs, 375 were used for PRS computation. The model combining EHR, SNP, and PRS achieved the highest AUROC of 0.9512 and AUPRC of 0.8902 in validation, while the XGB-based hybrid model reached an AUPRC of 0.9021 in testing. The SHAP summary plot highlighted the top 20 features predicting LN flares.

Conclusions: This hybrid model combining SNP, PRS, and EHR predicts active LN and requires validation.

Objectives: To obtain insights about inexperienced clinical researchers' hypothesis quality and associated factors. The findings inform the development of informatics tools to aid the hypothesis generation process. Methods: We analyzed an existing dataset collected through a randomized controlled study, focusing on individual hypotheses and participants. We invited clinical researchers to analyze datasets and develop hypotheses using the think-aloud method. Participants' screen activity and audio were recorded, transcribed, coded, and analyzed to measure the time and cognitive events (a granular unit of thought processes used by the participants while generating hypotheses). Hypotheses were rated by an expert panel. Here we analyzed (1) the top 5-rated hypotheses, (2) the bottom 5-rated hypotheses, and (3) the participants who generated them. Results: Participants who generated the top 5-rated hypotheses utilized fewer cognitive events and a shorter range of time per hypothesis; their hypotheses presented a higher valid rate, and they were more experienced. Conclusion: Having more experience is positively associated with higher quality and valid rates of the generated hypotheses. The higher-rated hypotheses seem to be positively associated with slightly fewer cognitive events and shorter time. The effect may not be linear. These analyses provide evidence for customized study designs or tool development based on these associated factors.

Healthcare big data has raised expectations for secondary use in research and information-based management. This case study explores limitations of using electronic health record (EHR) data from a hospital data lake by deriving indicators on opioid prescribing. A multi-staged method to calculate indicators of rational opioid use was developed covering both inpatient orders and outpatient prescriptions. The process included data selection, editing, organization, and validation. Visual Basic was employed to calculate indicators and to semi-quantify data limitations. Data (2015-2019) covered 3.3 million patients with 179,853 opioid and 22,415 benzodiazepine orders. Data quality issues, including unstructured, irregular, and invalid entries, limited analysis to indicators on opioid use and contraindications. In conclusion, secondary use should be considered in EHR system development. Data should be recorded in structured and unambiguous format, and methods for data quality measurement should be developed to ensure high-quality data is readily available in data lakes.

Multimodal artificial intelligence (MMAI) integrates and interprets diverse data types, such as images, text, video, and audio, and offers new opportunities for clinical decision support systems (CDSSs). Traditional CDSSs rely on unimodal data, which limits their predictive accuracy and coverage. The incorporation of MMAI holds promise for more accurate diagnosis, treatment optimization, and personalized patients care by synthesizing heterogeneous data sources. This narrative review explores the growing role of MMAI in improving diagnostic sensitivity, personalizing treatment, and improving healthcare delivery through the integration of heterogeneous data sources. It examines the evolution of MMAI technologies, such as large language models, large vision models, vision-language models, and large multimodal models, and their practical applications in clinical settings. The review also addresses key ethical, technical, and infrastructure challenges, such as data quality, model interpretability, bias, and system interoperability. Finally, it provides strategic recommendations for clinicians, researchers, and policy makers to promote responsible adoption of MMAI in healthcare. While recent developments show significant promise, addressing current limitations is essential to fully realize the transformative potential of MMAI in modern medicine.

Objective: To assesses the transformation of UK Clinical Practice Research Datalink (CPRD) databases into the Observational Medical Outcomes Partnership (OMOP) Common Data Model (CDM) version 5.3.1. Methods: A systematic approach was used to generate medical code lists and compare prevalent and incident counts between the source and OMOP CDM versions. Results: The results showed, for CPRD General Practitioner Online Database (GOLD) database, 89.5% of clinical events had no or very small differences in prevalent and incident event counts between the two versions of the database. The differences for CPRD Aurum were even smaller, with 97.4% of events showing no or very small differences in counts between the source and OMOP versions. Some observed discrepancies were due to codes being mapped into different tables. Conclusion: The study findings confirm the consistency of the OMOP transformation and provide confidence in analyses that query CPRD OMOP-transformed data.

Background: The preparation phase of an IT implementation may have consequences for the outcomes. To increase the chance of a successful implementation, many organisations invest in preparing the organisation through organisational readiness projects and similar activities. Objective: To determine how hospital staff in Central Norway were prepared for a new electronic health record (EHR) system and how expectations were met during the go-live and early use phase. Methods: Semi-structured interviews with the implementation team and staff members of a University hospital. Results: Preparations were carefully planned, but the staff considered them insufficient. The staff were positive towards the aim of the new EHR but found the training and practical preparations inadequate, resulting in unmet expectations and frustrations. Conclusion: Readiness activities in socio-technical implementations should emphasise to users that the information system presents opportunities, not necessarily fixed-use recipes, and is continuously evolving. Contextual information can help users align their expectations with the possibilities.

Objective: This study aims to conduct a multidimensional evaluation of non-small cell lung cancer (NSCLC)-related videos on social media platforms in China (TikTok, Bilibili, and Red).Methods: Validated tools were used to evaluate video quality (DISCERN instrument), reliability (Journal of the American Medical Association [JAMA] benchmarks), understandability, and actionability (Patient Education Materials Assessment Tool [PEMAT]).Results: This study included 96 videos, primarily created by medical professionals (n = 63). The median DISCERN score was 30.0 (IQR 28.5-34.4), indicating poor quality overall. Compared to videos rated as "good", the "poor" videos had significantly shorter durations (P = 0.040). The overall median understandability and actionability scores were 81.8% (IQR 75.0-90.9%) and 0% (IQR 0.0-66.7%), respectively, indicating good understandability but extremely poor actionability. Only one met all four JAMA benchmarks. TikTok videos with the shortest durations garnered the highest numbers of "likes", "comments", and "bookmarks", while Bilibili videos exhibited a relatively high overall quality.Conclusions: To guide the public in making informed medical decisions, Chinese NSCLC videos need improvement in various aspects.

Objective: With rising digital tool use and lack of standard evaluation criteria in medical education, students should learn to assess the quality of medical apps. We designed a workshop to equip medical students with app evaluation skills. This study describes the workshop and experiences of students and faculty with the Mobile App Rubric for Learning (MARuL) instrument. Methods: The study used qualitative design and consecutive sampling, including all third-year medical students enrolled in a medical informatics course. Faculty facilitators introduced students to the MARuL instrument and trained them on its use. Students, grouped based on their familiarity with medical apps, conducted evaluations, and feedback was collected through an online form. The data were analyzed using thematic content analysis. Faculty also reflected on the workshop's implementation and the effectiveness of the instrument. Results: A total of 275 students participated, identifying challenges such as the instrument's length, complex terminology, scoring difficulties, limited answer options, and the lack of open-ended questions. Faculty reported difficulties in explaining the MARuL and conducting the workshop effectively. Conclusion:The findings suggest the importance of integrating standard medical apps evaluation tools into the medical curriculum. These tools should be time-efficient, easy to understand, and simple to interpret for their utilization in an educational setting.

Background: Mobile health applications for the management of chronic conditions may be an effective method of supporting self-management. Objective: This report aims to describe the design process of a novel mHealth sleep retraining program for people with low back pain (LBP). Methods: This report includes a detailed description of the design and implementation process which utilized a four-stage framework: (i) concept development, (ii) therapeutic/design approaches, and app features/functionality), (iii) implementation metrics, and (iv) clinical outcomes in end-users. Results: The content of MySleepSolutions which is user-centered, personalized and evidence-based, is grounded in Cognitive Behavior Therapy for Insomnia (CBT-I) techniques and follows best practice guidelines. Conclusions: This detailed description of the design and development of MySleepSolutions ensures the program is comprehensive and has a sound theoretical basis while providing transparency and supporting future research into the development of mHealth retraining programs for sleep disturbance associated with LBP.

Objective: This paper demonstrates a data quality assurance (DQA) process as a means to identify and handle flaws in data, and hence improve the accuracy of an investigation into the prevalence of harmful versus non-harmful/near-miss incident reports in a single NHS acute provider.Methods: The three-step DQA process consists of an initial univariate data quality analysis, followed by a bivariate missingness analysis, and concluding with the design of appropriate multiple imputation techniques. With data quality established, the acuity and incident data were aggregated and aligned to the Ward-Month level for the period August 2015 to December 2020 inclusive. The final analysis was performed using binary regression, pooling results via Reuben's Rule.Results: The application of our three-step quality assurance process was able to detect and correct for common data quality issues. The resulting analysis identified a Ward dependency for the effect of Covid-19 lockdown measures on incident reporting culture which would have been missed without the applied imputation strategy.Conclusions: Our approach outlines a replicable methodology for understanding and fixing data quality issues in operational data. As daily operational decisions are being guided by data, it is important to leverage appropriate imputation techniques and ensure an optimal decision is reached.