The rapid growth of the Internet of Medical Things (IoMT) has enabled continuous monitoring of physiological signals and data-driven healthcare services, supporting predictive analytics, personalized treatment, and remote patient monitoring. However, conventional centralized machine learning approaches require sensitive patient data to be aggregated on centralized servers, raising concerns regarding privacy, security, regulatory compliance, scalability, and communication overhead. This systematic review critically evaluates federated learning (FL) approaches for IoMT-enabled predictive healthcare and remote patient monitoring, focusing on FL architectures, aggregation strategies, privacy-preserving mechanisms, communication efficiency, scalability, and practical deployment challenges. This systematic review followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 guidelines. Relevant literature was collected from IEEE Xplore, ScienceDirect, SpringerLink, ACM Digital Library, PubMed, Scopus, and Web of Science. Following screening and eligibility assessment, 69 studies published between 2020 and 2026 were included. The selected studies were evaluated using a structured quality assessment framework covering dataset validity, experimental rigor, reproducibility, scalability evaluation, privacy protection, and reporting quality. The evidence was synthesized to compare FL architectures, aggregation techniques, security mechanisms, communication performance, and healthcare applications. The findings indicate that FL enables collaborative model training without directly sharing raw patient data. Hierarchical and edge-assisted FL architectures can reduce communication overhead and improve scalability in distributed healthcare environments. Differential privacy, secure aggregation, homomorphic encryption, secure multiparty computation, and blockchain integration were identified as privacy-preserving mechanisms used in the reviewed studies. Many studies reported predictive accuracies exceeding 90
Digital health application use is shaped by motivation, technology, beliefs about their health and privacy. When considered simultaneously, the constructs examined within individual theories may overlap or lose their independent contribution. This study used an exploratory cross-theoretical framework to identify which factors emerged as the most stable correlates of future behavioral intention toward digital health application use. A cross-sectional survey was conducted (n = 156) among adults with prior or recent experience using digital health applications. Constructs derived from UTAUT2/TAM, the Health Belief Model, Privacy Calculus Theory, and Self-Determination Theory were measured using a multi-staged structural equation modeling approach. Separate measurement for each theory to examine the relationships was conducted, followed by a full integrated exploratory model. Constructs were retained in a reduced model using explicit rule-based retention criteria based on construct validity, theoretical distinctness, independent structural contributions, and parsimony. Sensitivity analyses using latent-variable and composite-score models were subsequently performed. Several constructs were associated with behavioral intention within their respective theory-specific models, but most associations were significantly reduced after simultaneous cross-theoretical adjustment. The only constructs, Relatedness and Self-Efficacy, were retained in the primary reduced model and remained independently associated with behavioral intention, with Relatedness showing the stronger association (β = 0.516, p < .001) followed by Self-Efficacy (β = 0.376, p < .001). The reduced model described 69.5
Personalized social media platforms rely on algorithmic curation to increase user engagement, yet the physiological mechanisms underlying this process are not well understood. This study investigated whether personalized TikTok content influences ocular markers of engagement, specifically pupil size and spontaneous eye blink rate, in a healthy sample. A within-subject design was used in which participants viewed both personalized videos (from their own TikTok feed) and non-personalized videos (from a novel account). Eye tracking was employed to record pupil size and spontaneous eye blink rate during video viewing. Differences between conditions were analyzed using within-subject statistical comparisons. Exploratory analyses examined associations between physiological measures and individual differences, including habitual TikTok use, daily usage duration, follower count, and self-reported enjoyment. Pupil size was significantly larger during personalized compared to non-personalized video viewing, indicating increased arousal-related engagement for personalized content. In contrast, spontaneous eye blink rate did not differ between conditions. Exploratory analyses did not reveal consistent associations between physiological measures and individual usage variables. Viewing personalized TikTok content was associated with larger pupil size than viewing non-personalized content, consistent with increased arousal-related engagement. The absence of differences in eye blink rate suggests that reward-related processes may not be differentially affected by personalization, or that this measure may lack sensitivity in this context. Because personalized and non-personalized feeds inherently differ in the content selected for each user, the present design cannot distinguish the contributions of specific content characteristics, such as emotional intensity and familiarity, to the observed physiological response. With these limitations in mind, the findings highlight the potential of physiological measures, particularly pupillometry, for studying engagement in digital media environments.
Colorectal cancer (CRC) remains a leading cause of cancer-related mortality in the UK. Early detection and removal of adenomatous polyps are essential to improving outcomes. Artificial intelligence-assisted colonoscopy tools, such as GI Genius (GIG), may enhance adenoma detection rates (ADR) and reduce the likelihood of missed lesions. A patient-level simulation cost-effectiveness model was developed to estimate the economic and clinical impact of GIG-assisted colonoscopy compared with standard colonoscopy alone in a symptomatic for suspected CRC population in England. The model adopted a National Health Service and Personal Social Services perspective, with a lifetime time horizon. Key model inputs were informed by the NAIAD trial, which evaluated the impact of GIG-assisted colonoscopy on ADR, compared with standard colonoscopy. Base case and probabilistic sensitivity analysis (PSA) were conducted to estimate the incremental cost-effectiveness ratio and the probability of cost effectiveness. Base case results suggest that GIG is a cost-effective intervention, associated with per-person cost savings of £69 and quality-adjusted life year (QALY) gains of 0.046. This was primarily due to increased ADR and subsequent reduced CRC incidence. Scenario analyses testing uncertainty around the sensitivity of colonoscopy, clinical outcomes for those initially undiagnosed, differences in outcomes across experienced and non-experienced endoscopists, and the total costs of GIG produced findings that were consistent with the base case and did not materially change the estimated costs or QALYs. PSA results indicate that GIG is cost effective in 53
Under-reporting of adverse drug reactions (ADRs) to regulatory authorities remains a major challenge for pharmacovigilance, despite the substantial patient safety and financial implications of ADRs encountered in clinical practice. Reporting by healthcare professionals is persistently low; in 2017, only 19
