Association between wrist-worn actigraphy and the MDS-UPDRS Parkinson’s disease rating scale through machine learning: an exploratory study

By
Gent Ymeri
Sara Caramaschi
Alban Haton
Carl Magnus Olsson
Myrthe Wassenburg
Per Svenningsson
Dario Salvi
July 13, 2026

Frontiers In Digital Health

Overview

This study investigates the relationship between wrist-worn actigraphy and MDS-UPDRS scores in Parkinson's disease.

Background

Parkinson's disease (PD) is characterized by fluctuating motor and non-motor symptoms that are often inadequately captured during episodic clinical assessments. The Movement Disorder Society-Unified Parkinson's Disease Rating Scale (MDS-UPDRS) is a standard tool for evaluating PD severity, but it may not reflect daily symptom variability. Continuous monitoring through wearable sensors like wrist-worn accelerometers offers a potential solution for more comprehensive symptom assessment in real-world settings.

Data Highlights

MDS-UPDRS Part	Mean Absolute Error (MAE)	Correlation (r)
Part I	3.0	0.60
Part II	2.7	0.61
Part III	8.2	0.47
Part IV	1.6	0.83
Total Score	13.3	0.49

Key Findings

The strongest estimation performance was for MDS-UPDRS Part IV with a MAE of 1.6 and correlation of 0.83.
Non-embedding features performed best for MDS-UPDRS Part II (MAE = 2.7, r = 0.61).
The combined feature set yielded the best results for MDS-UPDRS Part I (MAE = 3.0, r = 0.60) and Part III (MAE = 8.2, r = 0.47).
Estimation accuracy varies depending on the MDS-UPDRS target assessed.
Wrist-worn actigraphy may provide complementary information to traditional clinical assessments.

Clinical Implications

The findings suggest that wrist-worn actigraphy could enhance the understanding of PD severity in daily life. Clinicians may consider integrating wearable technology into routine assessments to capture symptom fluctuations more effectively.