Objective: To define motor progression subphenotypes in Parkinson’s disease (PD) and evaluates the predictive power of clinical and smartphone data in distinguishing slow from fast progressors at baseline.

Background: PD exhibits heterogeneous motor progression, complicating treatment strategies. While clinical assessments like the Movement Disorder Society–Unified Parkinson’s Disease Rating Scale (MDS-UPDRS) are standard, digital tools such as smartphone sensors offer novel insights into real-time motor changes. Few studies integrate these modalities to define subphenotypes or predict trajectories, a gap this analysis addresses using trial data.

Method: We conducted an exploratory analysis of 103 PD participants from the Exenatide-PD trial, tracked over 96 weeks in ON- and OFF-medication states. Adjusted mean changes in MDS-UPDRS Part III scores from baseline to week 96 were compared between exenatide and placebo across prespecified subgroups (sex, age, disease duration, motor subphenotypes like tremor-dominant or akinetic-rigid) and data-driven clusters (slow vs fast progressors). Predictive models integrated baseline MDS-UPDRS Part III scores with smartphone-derived metrics (e.g., gait, tremor). Accuracy was evaluated using area under the receiver operating characteristic curve (AUC) with 95% confidence intervals (CIs).

Results: In ON states, exenatide vs placebo differences in MDS-UPDRS Part III changes ranged from -0.03 (95% CI, -2.97 to 2.91) in females to 9.87 (95% CI, 8.10 to 11.64) in fast progressors. In OFF states, differences spanned 1.71 (95% CI, -0.92 to 4.34) in akinetic-rigid participants to 8.23 (95% CI, 5.91 to 10.55) in fast progressors. Data-driven fast progressors exhibited significantly greater decline (P<.001). Baseline prediction with MDS-UPDRS Part III alone yielded AUCs of 0.54 (95% CI, 0.00 to 0.93) ON and 0.53 (95% CI, 0.21 to 0.85) OFF; integrating smartphone data improved AUCs to 0.77 (95% CI, 0.39 to 1.00) ON and 0.76 (95% CI, 0.47 to 0.98) OFF. A simulated digital protocol combining both classifiers achieved AUCs of 0.74 (95% CI, 0.61 to 0.86) ON and 0.80 (95% CI, 0.67 to 0.91) OFF.

Conclusion: Motor progression in PD varies across subphenotypes, with data-driven fast progressors showing marked decline. Combining clinical and smartphone data enhances baseline prediction of motor trajectories, offering potential for personalized PD management.

Figure 1

Figure 2

Table 2

Table 3

Table 1

« Back to 2025 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/predicting-parkinsons-disease-motor-progression-using-clinical-and-digital-data/