Estimation of motor severity scales in Parkinson’s disease by linear models of bimanual non-alternating index finger tapping features

Abstract

Background and ObjectiveClinical scales used by well-trained clinicians to assess motor symptoms in patients with Parkinson’s Disease (PD) allow to establish the patients’ medical therapy and follow-up their response. However, these assessments are subjective and their application to patients requires experienced and qualified operators. This study analyzes the role of the kinematic patient’s features, captured by a simple computer keyboard paradigm, in predicting the scores prescribed by an experienced neurologist. MethodsA total of 47 patients in their ON medication state participated in this study. Their motor capacity was assessed by an experienced neurologist with several standardized clinical scales. The patients also performed 5 consecutive trials of 10 s of a computerized finger tapping task by pressing with their index the space bar, first with their dominant hand and then with the other hand. 270 tapping-related features were extracted from the tapping task data for each participant and linear regression multivariate models for each clinical variable were built by using these features. ResultsThe best resulting models were for the motor capacity (Unified Parkisnon Disase Scale Revised – MDS-UPDRS Part III), years from disease onset and balance scores (Limit of Stability – LoS), with root mean squared errors (RMSE) of 0.268, 0.254 and 0.150, respectively, all bellow their corresponding minimal clinically important differences. Those models included variables from both hands and from all trials, mainly regarding slow and fast tapping-related variables in different degrees. ConclusionsA simple bimanual non-alternating finger tapping task has shown to foresee motor capacity and balance scores by using statistical and machine learning methods. This easy and quick task could be performed periodically in the medical office or at home helping the clinician to know the patients’ motor state and temporary alterations in that way and to make finer clinical decisions about the proper pharmacological treatment of every patient.