The objective of this project is to gather and analyse speech samples from PD patients, prodromal cohorts, and healthy age-matched controls to identify potential speech and language parameters that could indicate and predict the development of PD. The speech-derived parameters will include analyses that reflect motor function (e.g. voice quality), cognition (e.g. word retrieval difficulties), and emotional (e.g. monotone speech), as well as motivational (e.g. reduced spontaneous speech) functions. Following initial sample size calculation and patient consent, speech samples will be collected using an automated phone call. By conducting regular, automated phone calls in the prodromal stage of PD, it may be possible to identify specific speech characteristics that predict the development of PD. Such a system could enable more proactive and targeted healthcare strategies for those at high risk of developing this debilitating disease. Following the collection of clinical, speech, and genetic data from PD patients, prodromal cohorts, and healthy controls, we will evaluate: 1. cognitive function through speech analysis, focusing on elements such as speech rate, hesitations, and word-finding difficulties. 2. emotional and motivational aspects of speech, including prosody (pitch, rhythm, and stress), speech monotonicity, and spontaneous speech volume and rate. 3. analyse motor function (specifically dysarthria) through the examination of articulatory precision, speech sound distortions, and voice quality. 4. explore respiratory function through the analysis of breath support for speech, including breath group duration and speech pausing patterns. The collection of speech data will be accomplished using the Mili tool developed by KIE. Furthermore, the collected speech will be analysed using state-of-the-art feature extraction and AI/ML algorithms and XAI techniques using the proprietary software of KIE. DC5 will work closely together with DCs4 and 6 due to the common focus on speech biomarkers. This project complements their work by focusing on distinct disease symptoms.
This project is part of the "Digital Health" work package.
Artificial Intelligence in Parkinson’s Disease