Objective: Seizure prediction could greatly improve the quality of life of people suffering from epilepsy. Modern prediction systems leverage artificial intelligence (AI) techniques to automatically analyze neurophysiological data, most commonly the electroencephalogram (EEG), in order to anticipate upcoming epileptic events. However, the performance of these systems is normally assessed using randomized splitting methods, which can suffer from data leakage and thus result in an optimistic evaluation. In this review, we systematically surveyed the available scientific literature looking for research approaches that adopted more stringent assessment methods based on patient-independent testing. Approach: We queried three scientific databases (PubMed, Scopus, and Web of Science), focusing on AI techniques based on non-invasive EEG recorded from human subjects. We first summarize a standardized signal processing pipeline that could be deployed for the development and testing of cross-patient seizure prediction systems. We then analyze the research work that meets our selection criteria. Main results: 21 articles adopted patient-independent validation methods, constituting only 4% of the published work in the entire field of epileptic seizure prediction. Among eligible articles, the most common approach to deal with cross-patient scenarios was based on source domain adaptation techniques, which allow to fine-tune the predictive model on a limited set of data recorded from a set of independent target patients. Significance: Overall, our review indicates that epileptic seizure prediction remains an extremely challenging problem and significant research efforts are still needed to develop automated systems that can be deployed in realistic clinical settings. Our review protocol is based on the preferred reporting items for systematic review and meta-analysis protocols 2020 guidelines for conducting systematic reviews, considering NHLBI and ROBIS tools to mitigate the risk of bias, and it was pre-registered in PROSPERO (registration number: CRD4202452317).

Background: Chronic Obstructive Pulmonary Disease (COPD) is a major global health burden in which acute exacerbations accelerate progression and increase hospitalizations. Emerging technologies, such as wearable biosensors, artificial intelligence (AI), and digital health tools, enable more proactive disease management. Objectives: This umbrella review synthesized evidence from systematic reviews and meta-analyses on (1) AI-driven prediction of COPD exacerbations using low-cost wearable biosignals, and (2) the effectiveness of digital health interventions on disease management, quality of life, and medication adherence. Methods: A systematic search of PubMed, Scopus, and Web of Science (2015–2025) identified eligible reviews. Methodological quality was assessed using AMSTAR-2, and study overlap was quantified with the Corrected Covered Area (CCA). A narrative synthesis was conducted across two research questions. Protocol registered in PROSPERO (CRD420251164450). Results: Twenty-seven reviews met the inclusion criteria. AI models demonstrated promising internal predictive accuracy but lacked external validation and clinical integration. Digital health interventions, such as mHealth applications and telerehabilitation, showed small to moderate improvements in quality of life and physical function. Reported effects varied considerably (OR = 0.20–2.37; I² = 0–94%), indicating substantial heterogeneity across studies. Evidence for improvements in medication adherence and exacerbation reduction was inconsistent, and most included reviews were rated “Low” or “Critically Low” in methodological quality, limiting the generalizability of findings. Conclusions: AI and digital tools show strong promise for proactive COPD management, particularly through wearable-derived biosignals, outperforming traditional static assessments. However, their clinical readiness remains limited due to small-scale studies, interpretability challenges, inconsistent outcome measures, and a lack of external validation. To support real-world translation and regulatory adoption, future research must prioritize large-scale, rigorous, and equitable studies with standardized methodologies and robust generalizability testing.

Chronic psychological stress contributes to allostatic load and is associated with cardiovascular, metabolic, and mental health disorders. Wearable devices enable continuous, noninvasive monitoring of autonomic signals such as heart rate variability (HRV), creating new opportunities for real-time stress assessment. Large language models (LLMs) are increasingly explored as interfaces for interpreting such data, but it remains unclear whether their predictions reflect physiologically meaningful patterns or rely on superficial heuristics. In this study, we assess whether LLM-derived stress predictions are physiologically coherent and how this varies with model scale. Using a longitudinal wearable dataset collected in naturalistic conditions (35 participants; 5,100 five-minute windows with HRV and contextual features), we obtained stress pseudoprobabilities from three models in the Mistral 3 family (675B, 14B, 3B) via zero-shot prompting. To make model behavior interpretable, we trained surrogate models to approximate LLM outputs and analyzed feature–response relationships using SHAP. Our results indicate that surrogate models closely reproduced LLM predictions (R² up to 0.915; Cohen's k up to 0.941), enabling high-fidelity characterization of decision patterns and providing a practical framework for auditing the physiological coherence of LLM-derived predictions. Physiological coherence increased with model scale: the largest model exhibited near complete alignment with established HRV stress responses, together with stable, predominantly monotonic feature effects and a balanced integration of physiological and contextual information. This pattern weakened at smaller scales, with the mid scale model showing partial alignment and the smallest model displaying reduced stability, greater feature concentration, and more irregular, non monotonic relationships. These findings indicate that larger LLMs encode more physiologically consistent representations of stress, whereas smaller models rely on simplified and less stable strategies, and highlight the value of surrogate based analysis as a practical framework for evaluating LLM behavior in biomedical applications and supporting their responsible integration into wearable health analytics.

Background: Digital microinterventions are increasingly used to support stress self-management in everyday life. While previous research has examined their feasibility, adherence, and preliminary effectiveness, much less is known about how these interventions are experienced over extended periods of real-world use. Objective: This study used collaborative autoethnography (CAE) to examine the everyday experience of living with MINT (Micro-Interventions for Navigating Tension), a Telegram-based chatbot for stress management. Our goal was not to evaluate clinical outcomes but to understand how scheduled microinterventions fit into everyday life, when they were ignored, and when people relied on other ways of coping instead, and to identify implications for the design of future digital mental health tools. Methods: We conducted a 4-week CAE involving 5 digital mental health researchers, who used the MINT chatbot as both participants and researchers. The experience was documented through multiple qualitative and quantitative sources, including reflective diaries, ecological momentary assessments (EMAs), workshop discussions, and wearable data, and analyzed collaboratively using reflexive thematic analysis (RTA). Results: Six interconnected themes described how MINT was experienced in everyday life: (T1) the importance of timing and emotional receptivity, (T2) the need for reciprocity and relational continuity, (T3) the tendency to favor existing coping strategies over suggested exercises, (T4) the meaning of nonengagement during periods of high stress, (T5) the collapse of self-initiated use after reminders were removed, and (T6) how everyday experiences with consumer wearables and general-purpose large language models reshaped expectations of digital mental health systems. Conclusions: Our findings show that CAE complements conventional evaluations of digital mental health interventions by revealing aspects of everyday use that are difficult to capture through feasibility studies, large-scale trials, or looking at usage data. Beyond the specific case of MINT, they suggest that future stress-management systems should better account for users’ receptivity, provide greater continuity across interactions, build on existing coping strategies, and help users move from externally prompted engagement to autonomous self-management.

Background: Chronic Obstructive Pulmonary Disease (COPD) is a major global health burden. Acute exacerbations (AE) significantly accelerate disease progression, reduce quality of life, and increase hospitalization rates. Emerging technologies, such as wearable biosensors, artificial intelligence (AI), and digital health interventions, offer new possibilities for proactive management. Objectives: This umbrella review synthesized evidence from systematic reviews and meta-analyses on (1) AI-driven prediction of COPD exacerbations using low-cost wearable biosignals, and (2) the effectiveness of digital health interventions on disease management, quality of life, and medication adherence. Methods: A systematic search of PubMed, Scopus, and Web of Science (2015–2025) identified eligible reviews. Methodological quality was assessed using AMSTAR-2, and study overlap was quantified with the Corrected Covered Area (CCA). A narrative synthesis was conducted across two research questions. Results: Twenty-seven reviews were included. Results from reviews suggest that AI models show promising internal predictive performance but often lack external validation, calibration, and clinical integration. Digital health interventions, including mHealth apps, telemonitoring, and telerehabilitation, instead seem to lead to small to moderate improvements in quality of life and physical function. Effects on medication adherence and exacerbation reduction were inconsistent. However, most included reviews were rated “Low” or “Critically Low” in methodological quality, and presented high outcome heterogeneity, limiting comparability and generalization of results. Conclusion: While AI and digital tools offer promising avenues for improving COPD management, their clinical readiness is limited by methodological weaknesses and insufficient validation. Future research should focus on addressing these gaps and real-world implementation of technological solutions for COPD care, prioritizing equity in access, and standardized outcome reporting.