The Early Detection of Cardiac Fatigue: Could the HRV Be Used as a Physiological Biomarker by AI?

Physical activity is vital for promoting health and rehabilitation,
and ensuring cardiovascular safety during such activities is paramount. Electrocardiography (ECG) and its longitudinal monitoring remain crucial for the early detection of
cardiac diseases. Recent advancements in nonlinear RR analysis and machine learning offer
promising approaches to identifying subtle precursors of cardiac pathologies in monitoring
systems using simple heart rate (HR) wearable sensors. Therefore, using HR sensors in
human activity recognition (HAR) is recommendable. After defining fatigue in a cardiological context, and focusing on an AI-based methods suite for HAR, the main research
question of this scoping review is as follows: “Can RR time series be successfully used
as physiological biomarkers for the early detection of cardiac fatigue?” The reported data
on assessment of fatigue are focused on the last two decades. The aim of this scoping
review was to collect, present and discuss the existing literature on the effectiveness of
AI-based methods for processing RR time series as a predictive biomarker for cardiac
fatigue compared to commonly used questionnaires for this outcome in adult populations.
Methods: Queries were conducted in the PubMed, Scopus and Google Scholar databases
for the time period 2005–2025. Only research articles and review papers were considered
suitable candidates. Results: Data from 10 papers were considered, related to the information researched. Conclusions: Information on HRV-based objective measures is quite
scarce and there is an urgent need to adopt a multidisciplinary approach and to improve
advanced AI-based nonlinear analyses to differentiate cardiac physiological status from
cardiac pathological status.