Nevin Manimala

Eur J Surg Oncol. 2026 May 6;52(7):111869. doi: 10.1016/j.ejso.2026.111869. Online ahead of print.

ABSTRACT

BACKGROUND: In 2020, immunotherapy entered first-line care for hepatocellular carcinoma (HCC). It remained uncertain whether population-level survival improved thereafter and whether adding immunotherapy to chemotherapy in routine practice was associated with additional benefit.

METHODS: A population-based SEER analysis was performed among HCC diagnosed in 2018-2022. Comparative analyses were performed between cases in the immunotherapy era (IEC) versus cases in the pre-immunotherapy era (PIEC). A prespecified chemotherapy subset compared chemotherapy + immunotherapy with chemotherapy alone. Competing-risks methods (Gray’s test; Fine-Gray) were applied with cancer-related death (CRD) as the event.

RESULTS: In total, 12056 patients were included (PIEC 7291; IEC 4765). After matching (n = 8796), better survival was observed in IEC versus PIEC (HR 0.905, 95% CI 0.846-0.968; P = 0.004). In the matched cohort, a lower CRD risk was also observed for IEC (sHR 0.807, 95% CI 0.750-0.868; P = 7.33 × 10^-9) with a matched-strata Gray’s P = 0.0105. In the chemotherapy subset (n = 3381; PSM n = 1884), the addition of immunotherapy was not associated with a statistically significant OS advantage after adjustment (HR 0.917; P = 0.166) or after matching (HR 0.914; P = 0.218), a pattern that remained consistent in the non-surgical subgroup (matched HR 0.885; P = 0.106). The matched CRD comparison was not significant (Gray’s P = 0.96), whereas an unmatched Fine-Gray model suggested a protective association (sHR 0.803, 95% CI 0.701-0.919; P ≈ 0.001).

CONCLUSIONS: Diagnosis in the post-approval immunotherapy era was associated with modestly improved OS and lower CRD at the population level. However, because this was an era-based rather than regimen-level comparison, the observed association cannot be attributed solely to immunotherapy uptake.

PMID:42119196 | DOI:10.1016/j.ejso.2026.111869

JMIR Mhealth Uhealth. 2026 May 12;14:e82276. doi: 10.2196/82276.

ABSTRACT

BACKGROUND: Nutrition apps offer scalable opportunities to support dietary behavior change and prevent chronic diseases. Their success depends on sustained user engagement, which is essential yet challenging to achieve and, consequently impacts the long-term effectiveness of these digital tools. Engagement strategies have been widely explored in digital health, but a comprehensive synthesis focusing on nutrition apps for adults is lacking.

OBJECTIVE: This scoping review aimed to map the current engagement approaches and metrics implemented in nutrition apps targeting adults and to identify how user engagement is defined across studies.

METHODS: We conducted a search of the PubMed, Scopus, Cochrane, and Web of Science databases for relevant studies published from January 1, 2013, to June 30, 2024. The inclusion criteria included original adult interventional or observational studies that evaluated nutrition apps and reported user‑engagement strategies or metrics. Two reviewers independently screened records in Covidence, with discrepancies resolved by a third reviewer. Data were charted across study characteristics, engagement strategies, and engagement metrics and then synthesized narratively.

RESULTS: A total of 59 studies that used apps to improve dietary behaviors were included in our analysis, including randomized controlled trials, observational trials, and mixed methods studies. Most of these apps were designed for adults who were overweight and obese. The studies were primarily conducted in North America and Europe and were randomized controlled trials or nonrandomized intervention studies, with varying durations and sample sizes. Engagement strategies varied widely, and engagement was typically measured by frequency of specific function use and frequency of app use, followed by retention rate. The most common engagement strategies reported in studies were push notifications (n=29, 49%), behavioral theory integration (n=24, 41%), personalization and customization (n=19, 32%), and goal‑setting features (n=18, 31%). Only 31% (n=18) of studies provided an explicit definition of “user engagement,” and definitions were highly heterogeneous. Engagement measurement was dominated by quantitative system‑recorded metrics, including time and frequency of using specific functions (n=38, 64%), app use frequency (n=34, 58%), and retention (n=17, 29%). Few studies assessed qualitative or long‑term engagement dimensions, and long‑duration studies rarely integrated adaptive or contextualized engagement mechanisms. Research apps more frequently used theory‑driven strategies compared with commercial apps, which tended to emphasize streamlined user experience.

CONCLUSIONS: Although several engagement strategies are commonly used, their implementation is inconsistent and often lacks grounding in conceptual frameworks. Research in the future needs to prioritize the use of common definitions for user engagement and measurement criteria while implementing user-centered design methods and using multiple research approaches to study the complex patterns of user engagement. The evidence base for engagement strategies needs strengthening because it will support the development of sustainable nutrition mobile health interventions.

PMID:42119138 | DOI:10.2196/82276

JMIR Med Inform. 2026 May 12;14:e84326. doi: 10.2196/84326.

ABSTRACT

BACKGROUND: The widespread adoption of electronic health records has resulted in the generation of large volumes of clinical notes. Learning algorithms and large language models can be trained on these resources, but they are susceptible to noise-irrelevant or noninformative data. This sensitivity can lead to significant challenges, including performance degradation and the generation of inaccurate predictions or “hallucinations.” This study addresses a critical challenge in clinical informatics: efficiently filtering millions of documents for relevance before advanced language model processing, particularly in resource-constrained environments.

OBJECTIVE: We present a novel framework for determining document relevance in clinical settings using a chronic obstructive pulmonary disease (COPD) dataset.

METHODS: We developed a novel framework using weak supervision and domain-expert heuristics to generate “silver standard” labels for training data and gold standard expert-annotated labels, creating 2 datasets to optimize the model during the development phase and subsequent testing phase. Various text representation techniques (bag of words, term frequency-inverse document frequency, lightweight document embeddings, compression-based features, and Unified Medical Language System concept extraction) were evaluated. These representations were used to train random forest, extreme gradient boosting, and k-nearest neighbor classifiers. Models were optimized on a small expert-annotated dataset and evaluated on a held-out test set.

RESULTS: The combination of lightweight document embedding with a random forest classifier demonstrated the best performance, achieving a precision of 0.73, recall of 0.86, and F1-score of 0.80 (95% CI 0.76-0.87) for identifying relevant COPD documents. This significantly outperformed baseline heuristics (precision=0.70; recall=0.38; F1-score=0.50, 95% CI 0.43-0.56) and other tested methods.

CONCLUSIONS: Our study presents a novel framework for identifying COPD-relevant clinical documents using lightweight embedding and machine learning. This approach effectively filters pertinent documents, enhancing information retrieval precision. The framework’s scalability and minimal annotation needs make it promising for diverse health care applications, potentially optimizing clinical outcomes through efficient document selection for data-driven decision support systems.

PMID:42119137 | DOI:10.2196/84326

Stud Health Technol Inform. 2026 May 7;335:236-241. doi: 10.3233/SHTI260090.

ABSTRACT

BACKGROUND: Machine learning models for predicting major adverse cardiovascular events (MACE) often generalize poorly across populations, and multinational development is limited by data-sharing constraints.

OBJECTIVES: We investigate whether federated learning (FL) can reduce the generalization gap of MACE prediction models across international clinical cohorts while preserving data privacy.

METHODS: Using harmonized electronic medical record (EMR) data from Austria, Brazil, and the USA, we train federated and local XGBoost and multilayer perceptron (MLP) models and evaluate performance using AUROC.

RESULTS: Our preliminary results show that the performance of local models degrades substantially on external cohorts, particularly when trained on smaller datasets. FL reduces this gap, with the greatest gains observed when compared to models trained on smaller cohorts and evaluated on the largest cohort. Local models performed best in-country, and XGBoost consistently outperformed MLPs.

CONCLUSION: Federated learning improves cross-site generalizability of MACE prediction models, with trade-offs between global robustness and local performance.

PMID:42119126 | DOI:10.3233/SHTI260090

Stud Health Technol Inform. 2026 May 7;335:222-229. doi: 10.3233/SHTI260088.

ABSTRACT

BACKGROUND: Digital health applications (DiGA) are regulated medical software intended to support prevention and therapy.

OBJECTIVES: To assess acceptance, potential application areas, information needs, and implementation-related criteria and concerns regarding DiGA among registered dietitians in Austria.

METHODS: An anonymous cross-sectional online survey was conducted in 2025.

RESULTS: 105 eligible dietitians provided data (completion rate 88.6 %). Familiarity with the DiGA concept was moderate. DiGA were perceived as very useful or useful across most dietetic practice areas. Usability, seamless integration into patients’ everyday routines, and data protection and security were the most important criteria for recommending DiGA. Respondents expressed a strong interest in profession-specific continuing education and supported a clearly defined role of dietitians in the use of DiGA.

CONCLUSION: Austrian dietitians demonstrate openness towards DiGA while emphasizing a strong interest for competence development. Early involvement of dietitians, alongside clear quality, interoperability, and data protection standards may support the integration of DiGA into routine care.

PMID:42119124 | DOI:10.3233/SHTI260088

Stud Health Technol Inform. 2026 May 7;335:197-202. doi: 10.3233/SHTI260084.

ABSTRACT

BACKGROUND: The Austrian Electronic Health Record (ELGA) aims to enhance healthcare coordination and patient empowerment, yet public uptake remains limited.

OBJECTIVE: This study explored citizens’ motivations and barriers toward ELGA use and reflected on the potential of science communication events to foster dialogue on digital health.

METHODS: During the Long Night of Science at UMIT TIROL, 35 participants anonymously shared on a whiteboard their reasons for using or not using ELGA. Statements were thematically analyzed using the Context-Mechanism-Outcome (CMO) framework.

RESULTS: Major barriers were concerns about data security and privacy, login complexity, and perceived lack of necessity. Facilitators included fast access to medical data, reduced paperwork, and improved continuity of care.

CONCLUSION: Participants balanced digital convenience with privacy concerns. Public events such as the Long Night of Science provide valuable opportunities not only to inform citizens about digital health but to let them actively participate in science, exchange perspectives, and learn from their lived experiences.

PMID:42119120 | DOI:10.3233/SHTI260084

Stud Health Technol Inform. 2026 May 7;335:191-196. doi: 10.3233/SHTI260083.

ABSTRACT

BACKGROUND: Adherence to movement precautions following sternotomy is essential for sternal healing, but patients often find it difficult to maintain the correct behavior.

METHODS: This study introduces a real-time computer vision-based system to evaluate movement compliance with the post-sternotomy precautions protocol. A YOLOv11 object detection model was trained using a dataset comprising real images and AI-generated images. A secondary monitoring algorithm was developed to use YOLO inference results to classify the whole action rather than a single frame.

RESULTS: The integration of synthetic data significantly enhanced YOLO model performance, achieving a mAP50 of 80.3%. In real-time validation, the monitoring algorithm correctly classified 85% of non-compliant actions without misclassifying any compliant actions.

CONCLUSION: This work demonstrates the feasibility of a low-cost, non-invasive solution for monitoring post-sternotomy precautions. Furthermore, the use of Generative AI proved effective in overcoming data scarcity.

PMID:42119119 | DOI:10.3233/SHTI260083

Stud Health Technol Inform. 2026 May 7;335:163-168. doi: 10.3233/SHTI260077.

ABSTRACT

Standardized care plans are essential for improving care quality, enabling consistent care delivery and seamless data exchange across care settings. However, the extent to which existing care planning tools implement standards, like HL7 FHIR, remains unclear. This study evaluates existing care planning tools through a fit-gap analysis assessing their alignment with standardized, interoperable care plan requirements and identify gaps hindering standardization and system integration. Tools were evaluated based on standard compliance, particularly FHIR, output capabilities, geographical origin, and accessibility patterns. Nine tools demonstrated full FHIR compliance, two partial implementation, and five no explicit FHIR support. Geographic analysis revealed US and Australian tools exhibited higher FHIR implementation rates than European solutions. Only seven tools achieved the “fit” criterion of combining care planning functionality with full FHIR compliance. No tools from Asia, Africa, or South America were identified, suggesting global gaps in access to standardized care planning technologies. These findings highlight geographic disparities in interoperability standards adoption and a need for policy interventions, particularly in Europe and underrepresented regions.

PMID:42119113 | DOI:10.3233/SHTI260077

Stud Health Technol Inform. 2026 May 7;335:153-154. doi: 10.3233/SHTI260074.

ABSTRACT

Recruiting patients for medical research requires a balance between ethical transparency and practical feasibility. We examined a two-stage re-consenting process for patients in the context of biobanks, capturing trends in patient engagement and preferences regarding future data use. Participation declined mainly at the opt-in stage, highlighting early procedural barriers, while among consenting, most participants allowed broad secondary use of their data without additional recontacting. In general, the results obtained support a transparent opt-out solution for data donation and secondary use of health data.

PMID:42119110 | DOI:10.3233/SHTI260074

Stud Health Technol Inform. 2026 May 7;335:131-136. doi: 10.3233/SHTI260069.

ABSTRACT

BACKGROUND: Serious games (SGs) with hand-tracking solutions are increasingly adopted in neurological rehabilitation. The Leap Motion Controller (LMC) is widely used in clinical and research contexts, while software-based systems, such as MediaPipe (MP), are emerging.

OBJECTIVES: This study compared the usability of MP- and LMC-based SGs.

METHODS: Seven SGs were developed to evaluate Grasp and Pinch gestures and tested with fifty healthy volunteers. Usability was assessed with the System Usability Scale (SUS) after each device, and an ad-hoc comparative questionnaire (LMC, MP, Neutral) collected preferences on several aspects.

RESULTS: Both systems achieved excellent SUS scores (MP: 88/100; LMC: 90/100), with no statistical difference in usability. The comparative questionnaire showed a stronger preference for LMC (63%) versus MP (17%), with 20% neutral.

CONCLUSION: Although both solutions demonstrated high usability, LMC was more frequently preferred in direct comparison. The SUS results of MP encourage further investigation, including large-scale tests with neurological patients.

PMID:42119105 | DOI:10.3233/SHTI260069