| KEYWORD |
Area Engineering
Development of an innovative ML algortim for PWV measurement on single signal sampling site
keywords CARDIOVASCULAR DISEASES, MACHINE LEARNING, PHYSIOLOGICAL SIGNALS, PWV
Reference persons DANILO DEMARCHI
External reference persons Irene Buraioli (irene.buraioli@polito.it)
Research Groups MiNES (Micro&Nano Electronic Systems)
Thesis type EXPERIMENTAL
Description Cardiovascular disease (CVD) is the leading cause of mortality in the world, accounting for about 17.9 million deaths a year, corresponding to 32 percent of total annual deaths. Early diagnosis and treatment of CVD can dramatically reduce the chance of premature mortality and ensure an everyday life. Cardiovascular risk has been widely correlated in the literature with Pulse Wave Velocity (PWV), an index with high predictive content. PWV represents the speed of pressure wave propagation within arteries and is usually assessed between the carotid and femoral sites. Precisely, it is calculated as the ratio of the distance between the two sites to the time elapsed to travel that length. Nowadays, PWV estimation is performed with instruments based mainly on applanation tonometry, but the high cost has severely limited their use in the clinical setting. The need for two sampling sites complicates the development of wearable solutions that allow continuous and automatic monitoring of this highly predictive parameter of cardiovascular risk.
This thesis aims to develop an ML algorithm for measuring PWV from pulse wave signals taken from a single acquisition site. The starting dataset has already been constructed and enriched by the research group collaborating with a medical team referring to acquisitions on volunteer subjects.
The proposed thesis work will be structured as follows:
1) The candidate will be required to perform a state-of-the-art ML algorithm analysis for pulse wave signals.
2) Next, the candidate will be required to analyze the dataset provided and implement a ML algorithm that allows reliable estimation of PWV.
3) Finally, the candidate should demonstrate the effectiveness of the proposed solution by evaluating its performance.
This process will lead the candidate to produce an algorithm for PWV estimation applicable in next-generation devices in clinical settings for cardiovascular risk monitoring and assessment.
Required skills Signal processing, programming in Matlab/Python, basics of ML
Deadline 14/12/2024
PROPONI LA TUA CANDIDATURA