copy-typed text
[Research] A new NLP task: diagnosing (classifying) Parkinson's disease from copy-typed text
The paper contains three datasets collected from three studies, each involving between doses of PD positive subjects and a similar amount of control subjects. The results demonstrate that there is a significant signal that separates the two groups. All subjects are asked to evaluate their typing proficiency. The hypothesis is that the model developed in the paper, is able to detect, with much higher specificity and sensitivity, early onset of PD as it operates on the type of errors you make as a typist but also the speed with which you do those errors. This is very different from how current tests are done.