To investigate whether keystroke dynamics (KD) were able to monitor symptoms in people with multiple sclerosis (pwMS).

To get a better insight into disease progression in pwMS, tools that allow continuously, remotely, and passively monitoring of disease progression are needed. Typing on smartphone keyboards is a promising tool since typing behavior, and thus KD is known to be affected by motor and non-motor symptoms.

101 MS patients underwent a maximum of five clinical visits during one year where the following clinical constructs were measured: fatigue (Modified Fatigued Impact Scale, Checklist Individual Strength, Fatigue Severity Scale), cognition (Symbol Digit Modality Test, California Verbal Learning Test, Brief Visuospatial Memory Test), manual dexterity (9-Hole Peg Test) and clinical disability (MS Functional Composite, Expanded Disability Status Scale). KD were collected by the intelligent Neurokeys keyboard on a patient’s own phone, pre-processed and aggregated seven days before and after clinical assessments. To determine the associations between KD and clinical measures, canonical correlation analyses were conducted to analyze outcome measures in a multivariate construct rather than on an individual basis.

For all constructs, one significant canonical variate (CV) pair was found, indicating the relationship with KD features. Analyses showed eight processing- and two typing speed KD features as predictors for the cognition construct (r=.71, p<0.001). Manual dexterity was predicted by one typing speed and three processing KD features (r=.60, p<0.001), and clinical disability were related to two typing speed features and one processing KD feature (r=.48, p<0.001). Fatigue was predicted by three mental processing KD features (r=.28, p<0.001). In general, slower typing speed and/or processing reflect worse clinical performance.

Typing behavior characteristics reflect distinct motor- and non-motor symptoms in people with MS. Keyboard interactions; therefore, represent a promising avenue to un unobtrusively and passively monitoring of MS symptoms in daily life.