Language modeling screens Parkinson's disease with self-reported questionnaires

Parkinson's disease (PD) is a growing public health challenge as populations age. Current diagnosis relies on motor symptoms and invasive procedures, making early detection difficult. Here we present Quest2Dx, a generalized artificial intelligence model that analyzes health questionnaires to enable low-cost, non-invasive PD screening. Conventional models are built for each questionnaire, and missing responses degrade their performance. Quest2Dx uses language modeling to encode the meaning of each question and answer, allowing it to transfer across questionnaires while enhancing interpretability. On the Parkinson's Progression Markers Initiative (PPMI) Online and Fox Insight datasets, Quest2Dx achieved areas under the receiver operating characteristic curve (AUROCs) of 0.977 and 0.974, respectively, outperforming existing methods. Additionally, cross-questionnaire validation achieved AUROCs of 0.920 and 0.952, respectively, from PPMI Online to Fox Insight and vice versa. Quest2Dx also identified key predictors from the questions to provide further insights. These results indicate a promising path to PD screening in primary care.