Objective: To determine whether a smartphone-based alternating tapping task can reliably differentiate Parkinson’s disease (PD) from healthy controls and to identify the tapping metrics that best capture bradykinesia.​​​​

Background: Parkinson’s disease is a progressive neurodegenerative condition marked by both motor and non-motor symptoms [1]. Assessing bradykinesia objectively remains challenging with conventional rating scales such as the MDS-UPDRS [2]. The global increase in PD prevalence [3] underscores the need for digital solutions that provide precise and repeatable measures in clinical and telemedicine settings.

Method: We recruited 115 individuals with PD (ON, OFF or dyskinetic stages) and 726 healthy controls. Participants performed a 10-second smartphone-based alternating tapping test using a dedicated mobile application [Figure 1]. We extracted multiple tapping parameters—inter-tap interval, tapping location dispersion, error taps, holding duration, rhythmicity, and temporal unpredictability. Statistical comparisons were made using Mann-Whitney U tests. A multivariate logistic regression analysis identified significant predictors of PD, and the model was evaluated via ROC-AUC.

Results: A total of 490 PD recordings and 726 control recordings were analyzed. Nearly all tapping metrics differed significantly (p<0.05) between PD and controls, except for inter-tap interval (p=0.212) and correct taps interval (p=0.075). The extracted tapping features are illustrated in [Figure 2]. Regression analysis identified standardized correct tapping score, tapping location dispersion, temporal unpredictability, holding duration, and inter-tap interval variability as significant predictors [Figure 3]. The final model achieved an ROC-AUC of 0.775 [Figure 4]. Heatmaps of normalized tapping locations demonstrate differences in spatial tapping patterns between PD and controls [Figure 5].

Conclusion: Smartphone-based alternating tapping offers a simple and objective method for detecting bradykinesia-related dexterity impairments in PD. Inaccuracy and variability of spatiotemporal tapping metrics were key indicators of PD-related motor impairment. These findings support the potential of multi-faceted tapping parameters as valuable digital biomarkers for monitoring disease progression and enabling early detection.

Screenshots of the PD-Plus smartphone application

Tapping feature comparison in PD vs. controls

Forest plot of odds ratios for tapping parameters

ROC curve for PD classification model

Heatmaps of tap locations in PD and controls

References: [1] Jankovic J. Parkinson’s disease: clinical features and diagnosis. J Neurol Neurosurg Psychiatry. 2008.
[2] Goetz CG, et al. MDS-UPDRS scale for PD. Mov Disord. 2008.
[3] Bhidayasiri R, Sringean J, Phumphid S, et al. The rise of Parkinson’s disease is a global challenge, but efforts to tackle this must begin at a national level: a protocol for national digital screening and “eat, move, sleep” lifestyle interventions to prevent or slow the rise of non-communicable diseases in Thailand. Front Neurol. 2024.

« Back to 2025 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/feature-extraction-and-classification-using-smartphone-based-alternating-tapping-tasks-for-distinguishing-parkinsons-disease-from-healthy-individuals/