Multimodal Remote Digital Phenotyping for Detecting and Tracking Early Parkinsonian Change in LRRK2 Carriers
Published in Research Square (Preprint), 2026
We present a remote, multimodal video framework for identifying prodromal Parkinson’s disease among LRRK2 mutation carriers, analyzing $\mathbf{829}$ participants, including $\mathbf{158}$ LRRK2 carriers ($36$ with manifest PD, $122$ non-manifest). The AI model distinguished non-manifest carriers from controls with $\mathbf{92.9\%}$ accuracy (AUROC $\mathbf{0.92}$, AUPRC $\mathbf{0.82}$), and a continuous “PD Weigh-In” score captured clinical decline in carriers who subsequently developed PD, correlating strongly with expert ratings (Pearson $r = 0.77$, Spearman $\rho = 0.79$) in the held-out LRRK2 test cohort.
Recommended citation: Adnan, T.M.T., Abdelkader, A., Islam, M.S., Pawlik, M., Hartman, E., Hoque, E. et al., 2026. Multimodal Remote Digital Phenotyping for Detecting and Tracking Early Parkinsonian Change in LRRK2 Carriers. Research Square, rs.3.rs-9474934.
Download Paper