Sequence-based dynamic handwriting analysis for Parkinson’s disease detection with one-dimensional convolutions and BiGRUs

Parkinson’s disease (PD) is commonly characterized by several motor symptoms, such as bradykinesia, akinesia, rigidity, and tremor. The analysis of patients’ fine control, particularly handwriting, a powerful tool to support PD assessment. Over the years, various dynamic attributes pen pressure, stroke speed, in-air time, etc., which can be captured with help online handwriting acquisition tools, have been evaluated for identification PD. Motion events, their associated spatio-temporal properties in enable effective classification patients through unique sequential patterns. This paper proposes novel model based on one-dimensional convolutions Bidirectional Gated Recurrent Units (BiGRUs) assess potential information identifying Parkinsonian symptoms. One-dimensional are applied raw sequences well derived features; resulting then fed BiGRU layers achieve final classification. proposed method outperformed state-of-the-art approaches PaHaW dataset achieved competitive results NewHandPD dataset.