Can typing patterns detect Parkinson's disease?

Small studies have found that the timing of keystrokes can separate people with early Parkinson's disease from people without it, with AUC scores from 0.81 to 0.98 ^{1 2}. A typing pattern is not a diagnosis, though. The sections below explain why, and what the studies did and did not show. Our typing research guide sets out the evidence.

Typing timing separated early Parkinson's from controls in small studies

A 2016 study recorded 42 patients with early Parkinson's and 43 controls ¹, each typing a folk tale on their own computer as they normally would at home ³. Analysis of key hold times separated the two groups with an AUC of 0.81 ¹. AUC stands for area under the curve. Higher scores mean better separation, and a score at chance level is no better than a coin toss.

A 2017 study recorded keystroke timing from 103 people on a computer keyboard, 32 of them with mild Parkinson's ⁴. It reached an AUC of 0.98 across two separate groups of participants ². A 2018 study used touchscreen typing on smartphones rather than a keyboard, in matched groups of 18 patients and 15 controls, and reached an AUC of 0.92 ⁵.

These are high numbers, but they describe the groups each study chose. Whether a typing measure works the same way across ordinary people has not been shown.

A typing score is not a diagnosis

The studies are not a diagnostic test that a clinician can apply to one person. The National Library of Medicine's patient article in MedlinePlus Magazine puts it plainly: "There’s no single test for Parkinson’s disease" ⁶. It adds that doctors usually diagnose the condition based on medical history and a neurological examination ⁷. The diagnosis rests on the history and the exam, not on one measurement.

Each study used its own participants, so their scores cannot be compared as if they came from one test. We found no larger, independent study that repeats the keyboard results on a broad population. Until one does, a typing score is a research signal.

Typing timing also changes with fatigue, injury and distraction

A typing timing signal picks up many things besides disease. Typing rhythm varies with injury, fatigue or distraction, according to a 2013 survey of keystroke biometrics ⁸. That matters for interpreting these studies. A timing difference between two groups is hard to attribute to disease unless the groups were matched on those factors. Only the 2018 study describes its groups as matched. We did not confirm how the 2016 and 2017 studies chose their controls, so read each result with that gap in mind.

What the research has not yet shown

These are the gaps we found, based on the studies we read:

• Performance in unselected people. The studies compared chosen patients with chosen controls. How a typing measure performs across everyone who types is not established.
• Independent replication. The keyboard results come from separate small groups, and we found no larger study that repeats them.
• Added value. Whether a typing measure adds anything to what a clinician already assesses is not shown by these studies.

Symptoms are a reason to see a doctor, not a typing test

A typing test is not a reason to self-diagnose. If you notice tremor, slowness, stiffness, changes in balance or walking, or changes in handwriting or speech, talk to a doctor. The diagnosis is made in a clinic from the history and an examination, and the doctor is the right person to weigh any other test.

If typing itself causes pain in your hands or wrists, the question on whether typing can cause carpal tunnel covers that separate concern.

Related reading

• The question on whether typing rhythm can identify you uses the same timing data for a different purpose, identity rather than health.
• The keystroke dynamics entry defines the timing measures these studies use.
• Speech is another channel that researchers study in Parkinson's. Our guide on speech recognition for speech impairments covers that side.
• The question on whether typing speed correlates with IQ covers another claim that typing is linked to a trait.

The typing research guide places this question among the other questions it covers, and the typing and writing speed overview lists the rest.