Predicting in-hospital indicators from wearable-derived signals for cardiovascular and respiratory disease monitoring: an in silico study

Cardiovascular and respiratory diseases (CVRD) are the leading causes of death worldwide. The construction of health digital twins for patient monitoring is becoming a fundamental tool to reduce invasive procedures, lower healthcare costs, minimize patient hospitalization, design clinical trials and personalize therapies. The aim of this study is to investigate the feasibility of machine learning-based monitoring of healthy subjects and CVRD patients in an in silico context. In particular, a population of virtual subjects, both healthy and with CVRD, was created using a comprehensive zero-dimensional global closed-loop model. Then, we trained Gaussian process regression (GPR) models, informed by wearable-acquired data, to predict variables normally acquired with invasive or operator-dependent methods. Presented results demonstrate, in an in silico setting, the feasibility of GRP-based prediction of in-hospital variables from wearable-derived indexes. Author summaryCardiovascular and respiratory diseases are major global health concerns. Effective remote monitoring is essential for early detection of complications and improved patient care, especially for people with chronic conditions. Wearable devices provide a non-invasive way to track health indicators, but they do not directly measure certain key physiological parameters that doctors typically assess in hospitals. In this study, we explore how machine learning approaches can help bridge this gap. By using virtually-generated data, we trained Gaussian process regression models to estimate critical cardiovascular and respiratory indexes, such as cardiac output and oxygen levels. In particular, we created a virtual population of simulated patients and used their data to train and test our model. Our findings suggest that this approach can be a valuable tool for remote monitoring, providing healthcare professionals with accurate insights, without the need for invasive procedures and enabling earlier detection of complications. However, further testing with real patient data is necessary to fully assess its clinical potential.