Handwriting, touchscreen dexterity and bradykinesia measures in Parkinson's disease: a feature selection study.

Abstract

Introduction: Bradykinesia affects handwriting and smartphone usage in patients with Parkinson's disease (pwPD).

Objective: To assess handwriting, hand dexterity, smartphone usage, and bradykinesia in pwPD and identify features that best describe upper-limb alterations in pwPD.

Methods: Forty pwPD and 30 age/sex-matched healthy controls were included. We used standard handwriting/dexterity tests: Manual-Ability-Measure-36, Purdue-Pegboard-Test (PPT) and copy of a text on paper. Spatiotemporal handwriting parameters were assessed using tests on a tablet: copy of text and pre-writing tasks. To obtain objective data on movement speed and amplitude on the smartphone, we developed tests involving the most commonly used gestures (tap, swipe, and slide). Bradykinesia during a finger-tapping task was evaluated using electromagnetic sensors. Sequential feature selection models were used to identify the parameters best distinguishing pwPD and healthy controls.

Results: PwPD relative to healthy controls showed reduced manual ability and dexterity. They showed reduced movement amplitude and speed in smartphone tests and signs of micrographia during handwriting tests. Moreover, kinematic parameters correlated with both PPT and Movement Disorder Society-Unified Parkinson's Disease Rating Scale III. Each feature selection model demonstrated a good accuracy, particularly when including standard handwriting/dexterity tests (R² = 0.90), tests on smartphone (R² = 0.94) and all the features together (R² = 0.97). The best features were self-reported manual abilities, PPT, tap and swipe speed/amplitude on smartphone.

Conclusions: This study showed that technological assessments can be added to standard evaluations to provide quantitative measures of handwriting, dexterity, and bradykinesia that will be useful to assess PD progression and the effects of interventions in pwPD.