Diseases like Parkinson's affect a non-negligible percentage of elder people worldwide. Affected people life quality strictly depends on the disease progression and counteractions that medical doctors decide to apply, according to periodical clinical evaluations. The possibility to have continuous monitoring of people with the help of electronic devices is a desirable opportunity for two main reasons: i) to reduce queues for medical doctors for in-person clinical assessment, ii) provide them with real-time and continuous monitoring data produced by assisted people in their home environment. To this end, the paper proposes the development and physical realization of an all-in-one low-cost device, able to both measure and process movement data in real-time, adopting miniaturized inertial sensors and machine learning capabilities. To accomplish the task, a pre-validated movement simulator is adopted to generate data concerning different kinds of movement disorders, specifically focusing on tremors related to Parkinson's disease specific medical tests. The simulator takes into consideration the metrological features of the platform later adopted in the miniaturized device. The generated data are then used to train a machine learning tool to recognize such disorders and provide an estimation of the disease's progression status. Once completed the tuning procedure, the whole process has been transferred to the developed physical device and validation tests have been carried out to prove its suitability for the described purpose. Obtained performance (minimum 98.5% mean accuracy with the emulated data) and low-energy consumption (maximum 27 mWh) allow stating as such prototype can be a robust basis for the following engineering and release phases to create a wearable object. It can be helpful for medical doctors as assisting support for correct diagnosis and prompt intervention in case of sudden worsening of people disease status.
A Low-Cost Edge Computing Device for Real-Time Detection of Motor Symptoms in Neurodegenerative Diseases Using Machine Learning
Carissimo, ChiaraMembro del Collaboration Group
;Cerro, GianniMembro del Collaboration Group
;Komici, KlaraMembro del Collaboration Group
;
2024-01-01
Abstract
Diseases like Parkinson's affect a non-negligible percentage of elder people worldwide. Affected people life quality strictly depends on the disease progression and counteractions that medical doctors decide to apply, according to periodical clinical evaluations. The possibility to have continuous monitoring of people with the help of electronic devices is a desirable opportunity for two main reasons: i) to reduce queues for medical doctors for in-person clinical assessment, ii) provide them with real-time and continuous monitoring data produced by assisted people in their home environment. To this end, the paper proposes the development and physical realization of an all-in-one low-cost device, able to both measure and process movement data in real-time, adopting miniaturized inertial sensors and machine learning capabilities. To accomplish the task, a pre-validated movement simulator is adopted to generate data concerning different kinds of movement disorders, specifically focusing on tremors related to Parkinson's disease specific medical tests. The simulator takes into consideration the metrological features of the platform later adopted in the miniaturized device. The generated data are then used to train a machine learning tool to recognize such disorders and provide an estimation of the disease's progression status. Once completed the tuning procedure, the whole process has been transferred to the developed physical device and validation tests have been carried out to prove its suitability for the described purpose. Obtained performance (minimum 98.5% mean accuracy with the emulated data) and low-energy consumption (maximum 27 mWh) allow stating as such prototype can be a robust basis for the following engineering and release phases to create a wearable object. It can be helpful for medical doctors as assisting support for correct diagnosis and prompt intervention in case of sudden worsening of people disease status.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
