Frameworks for human activity recognition (HAR) can be applied in the clinical environment for monitoring patients’ motor and functional abilities either remotely or within a rehabilitation program. Deep Learning (DL) models can be exploited to perform HAR by means of raw data, thus avoiding time-demanding feature engineering operations. Most works targeting HAR with DL-based architectures have tested the workflow performance on data related to a separate execution of the tasks. Hence, a paucity in the literature has been found with regard to frameworks aimed at recognizing continuously executed motor actions. In this article, the authors present the design, development, and testing of a DL-based workflow targeting continuous human activity recognition (CHAR). The model was trained on the data recorded from ten healthy subjects and tested on eight different subjects. Despite the limited sample size, the authors claim the capability of the proposed framework to accurately classify motor actions within a feasible time, thus making it potentially useful in a clinical scenario.

Abnormal locomotor patterns may occur in case of either motor damages or neurological conditions, thus potentially jeopardizing an individual’s safety. Pathological gait recognition (PGR) is a research field that aims to discriminate among different walking patterns. A PGR-oriented system may benefit from the simulation of gait disorders by healthy subjects, since the acquisition of actual pathological gaits would require either a higher experimental time or a larger sample size. Only a few works have exploited abnormal walking patterns, emulated by unimpaired individuals, to perform PGR with Deep Learning-based models. In this article, the authors present a workflow based on convolutional neural networks to recognize normal and pathological locomotor behaviors by means of inertial data related to nineteen healthy subjects. Although this is a preliminary feasibility study, its promising performance in terms of accuracy and computational time pave the way for a more realistic validation on actual pathological data. In light of this, classification outcomes could support clinicians in the early detection of gait disorders and the tracking of rehabilitation advances in real time.

Human gait is a repetitive motor action that involves the coordination of various motor patterns controlled by a network of kinematic relations. Gait disorders, which can arise from neurological, muscular, or skeletal impairments, affect this coordination and lead to abnormal walking patterns. Pathological gait recognition (PGR) employs artificial intelligence methods to identify abnormal gaits by discriminating healthy from pathological motor patterns, thus helping clinicians to identify and monitor abnormal locomotor patterns. However, redundant input data, which may derive from similar motor patterns, can negatively impact on the model performance. Feature selection techniques are typically adopted to optimize the input dataset of AI-based frameworks and may rely on correlation analysis to identify redundant inputs. Graph theory can be used to provide a better understanding of the intricate relations and dependencies among input data. A paucity has been found in the literature about the exploitation of complex network analysis (CNA) as a method for optimizing the input dataset of a DL model oriented to PGR. This study employs CNA to optimize the performance of a PGR-oriented framework based on a DL architecture fed by inertial sensors. Despite the limited sample size, the model outcomes seems to confirm the effectiveness of complex network analysis in improving the performance in terms of prediction time by means of input data optimization.

Artificial intelligence algorithms have become extensively utilized in survival analysis for high-dimensional, multi-source data. However, due to their complexity, these methods often yield poorly interpretable outcomes, posing challenges in the analysis of several conditions. One of these conditions is obstructive sleep apnea, a sleep disorder characterized by the simultaneous occurrence of comorbidities. Survival analysis provides a potential solution for assessing and categorizing the severity of obstructive sleep apnea, aiding personalized treatment strategies. Given the critical role of time in such scenarios and considering limitations in model interpretability, time-dependent explainable artificial intelligence algorithms have been developed in recent years for direct application to basic Machine Learning models, such as Cox regression and survival random forest. Our work aims to enhance model selection in OSA survival analysis using time-dependent XAI for Machine Learning and Deep Learning models. We developed an end-to-end pipeline, training several survival models and selecting the best performers. Our top models—Cox regression, Cox time, and logistic hazard—achieved good performance, with C-index scores of 0.81, 0.78, and 0.77, and Brier scores of 0.10, 0.12, and 0.11 on the test set. We applied SurvSHAP methods to Cox regression and logistic hazard to investigate their behavior. Although the models showed similar performance, our analysis established that the results of the log hazard model were more reliable and useful in clinical practice compared to those of Cox regression in OSA scenarios.

Background: Frailty is highly prevalent among older adults with cardiovascular disease (CVD) and strongly predicts disability and mortality after cardiac events. Although cardiac rehabilitation (CR) improves prognosis, frail older patients often face barriers to participating in in-person programs. eHealth-based, home-delivered CR programs incorporating tele-rehabilitation and remote monitoring may improve accessibility, yet evidence regarding their effectiveness on frailty status remains limited. Methods: We designed a multicenter, randomized, parallel-group trial enrolling people ≥65 years recently hospitalized for acute heart failure (AHF) and/or acute coronary syndrome (ACS). Participants were randomized 1:1 to an eHealth home-based tele-rehabilitation program or the usual care. The primary endpoint is frailty prevalence at follow-up, defined by an Essential Frailty Toolset (EFT) score ≥3, with co-primary outcomes being between-group differences in the mean levels of EFT and Short Physical Performance Battery (SPPB) scores after 3–6 months. Secondary endpoints include mortality and hospitalization, among others. Results: The full protocol and study procedures are reported. Between May 2024 and December 2025, 589 patients were screened at the two Italian centers involved; 442 met eligibility criteria and 209 were enrolled and randomized. Baseline characteristics were largely comparable between groups. The mean age was 77 ± 9 years, 70% were male, and 55% had ACS. Lower-than-expected enrollment was mainly attributable to refusal related to difficulties in using digital devices. Conclusions: This randomized trial will evaluate whether a multidomain, eHealth-based CR intervention can reduce the prevalence or degree of frailty in older people after AHF or ACS. We report the study protocol and baseline characteristics of the enrolled cohort, highlighting the challenge of digital illiteracy in contemporary older populations.