Multidimensional analysis of frailty: the E.O. Galliera case through Bio-Signal and Muscle Quality telemonitoring

This thesis is part of the project ’RAISE - Robotics and AI for Socio-economic Empowerment’ (ECS00000035), funded by the NextGenerationEU initiative and the Ministry of University and Research (MUR), under the National Recovery and Resilience Plan (NRRP), Mission 4, Component 2, Investment 1.5. The project aims to develop intelligent devices and technologies for telemonitoring frail individuals in collaboration with the Italian Institute of Technology and the Geriatric Hospital E.O. Galliera. With increasing life expectancy and declining birth rates, the ageing population faces a rise in age-related diseases affecting motor skills and autonomy. A key approach for assessing frailty in geriatric patients is the Comprehensive Geriatric Assessment, using multidimensional evaluation indices as the Multidimensional Prognostic Index (MPI) to evaluate the individual’s clinical condition. This study targets people over 65, focusing on sarcopenia, which impairs mobility, increases fall risks and reduces autonomy. The system is designed for home use, allowing early detection of decline and optimized treatment management, reducing strain on healthcare systems and improving care effectiveness. For remote monitoring, advanced sensors were integrated: Body Cardio smart scale uses bioelectrical impedance analysis to estimate body composition and Pulse Wave Velocity (PWV), which assesses arterial stiffness and cardiovascular risk. It measures body mass, fat mass, muscle mass, bone mass, and hydration status; BPM Core, an advanced sphygmomanometer with ECG, digital stethoscope, and blood pressure sensor, detects atrial fibrillation and measures blood pressure; lastly, Sleep Analyzer provides detailed sleep cycle analysis using radar technology and motion sensors, without direct contact. To optimise data management, an advanced API-based procedure was developed to extract the raw data from the sensors without going through the proprietary Withings platform, by routing it directly to the MiaCare platform used in the project. Through the use of Python and JavaScript, it was possible to fully automate the process, eliminating the need for manual intervention. This approach ensures a continuous and accurate download of data, with the possibility to configure predefined acquisition intervals, thus ensuring a constant and timely flow of information for remote monitoring, while respecting all authentication and privacy protocols, guaranteeing the security and protection of sensitive patient data. All the sensor signals were analysed. Interval and amplitude parameters were extracted from the ECG signal, and spectral analysis of heart rate variability (HRV) provided information on modulation of the autonomic nervous system. Signals were cross-correlated with those of the other sensors for a multidimensional health assessment. A Matlab code was developed to interpret the data and generate graphs illustrating the trend of monitored parameters over time. Additionally, the platform’s front-end was designed with graphical programming, providing doctors with intuitive reports and visualisations of the patient’s condition. Finally, to verify the accuracy and ensure the reliability of the proposed sensors, their data were compared with the hospital sensors usually used during patient screening and assessment clinical procedures. Future work includes adding 3D-printed devices for muscle strength monitoring, crucial for diagnosing sarcopenia, enhancing the platform’s ability to offer extensive and customized monitoring.