ParkMAE: a cross-linguistic masked autoencoder framework for robust Parkinson’s disease detection from speech
Angelika Ando, Adrien Lesage, Marc De Gennes, Rachid Riad · 2025
Read ParkMAE: a cross-linguistic masked autoencoder framework for robust Parkinson’s disease detection from speech by Angelika Ando, Adrien Lesage, Marc De Gennes, Rachid Riad (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/pmc-12780117.txt.
Description
Angelika Ando, Adrien Lesage, Marc De Gennes, and Rachid Riad present ParkMAE in 2025, a speech model that spots Parkinson's disease across languages. Voice change can be an early sign of the illness, but tools often work in one language only. The team gathered about seven hundred fifty hours of speech to train on, across four languages.
Their model learns by hiding and rebuilding parts of speech, with no labels needed. It beat classic sound markers by a wide margin on cross language diagnosis, while using far fewer parts than big speech models. It also held up on a language it had never trained on. Staging the illness from voice proved harder.
Catalog record
| CREATOR | Angelika Ando, Adrien Lesage, Marc De Gennes, Rachid Riad |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 53 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |