Peer-reviewed research

High-accuracy detection of early Parkinson's Disease using multiple characteristics of finger movement while typing

Warwick R. Adams, PLOS ONE, 2017 · DOI 10.1371/journal.pone.0188226

Read on October 16, 2026

What it is: Keystroke timing was recorded from 103 people, 32 of them with mild Parkinson's disease, while they typed on a computer keyboard over an extended period. A machine-learning ensemble separated early-stage patients from controls on two separate groups of participants.
How much weight it carries: Peer-reviewed. The sample is small and recruited for the study, and the reported accuracy is an estimate on these groups. Replication in larger, independent cohorts is needed before the accuracy can be expected in routine screening.

Facts taken from this source

In a 2017 study, keystroke timing was captured from 103 people, 32 with mild Parkinson's disease and the rest non-Parkinson's controls, while they typed on a computer keyboard.
“keystroke timing information from 103 subjects (comprising 32 with mild PD severity and the remainder non-PD controls) was captured as they typed on a computer keyboard over an extended period” (Abstract)
In a 2017 study, a machine-learning ensemble using keystroke timing separated early-stage Parkinson's from controls with an AUC of 0.98 across two separate groups of participants.
“this approach was able to successfully discriminate between early-PD subjects and controls with 96% sensitivity, 97% specificity and an AUC of 0.98” (Abstract)