AI accurately detects Parkinson’s in Bengali speakers
Summary and headline written by AI from the source article. How we work
Researchers developed BenSParX, a new machine learning framework and dataset that accurately identifies Parkinson’s disease from spoken Bengali.
The system addresses a gap in healthcare technology, as existing voice analysis tools for Parkinson’s largely focus on major languages like English and lack resources for Bengali, spoken by over 230 million people primarily in Bangladesh and India.
The framework combines analysis of acoustic features in speech with a technique called SHAP, which explains how the AI reaches its diagnosis by quantifying the importance of different vocal characteristics. The BenSParX framework not only created the first Bengali speech dataset for Parkinson’s detection but also demonstrated high performance. It achieved 95.67% accuracy, a 95.62% F1 score, and an area under the curve of 0.990 in tests.
Importantly, the researchers also tested the framework on Parkinson’s datasets from other languages and found it consistently surpassed existing AI methods. This research could improve early diagnosis in regions where access to specialized neurological care is limited. By analyzing speech, the system offers a non-invasive and affordable way to screen for Parkinson’s, potentially leading to earlier interventions and better patient outcomes. The researchers made the labelled dataset publicly available to encourage further development and wider access to this technology.
