Mobile App-Based Assessment Predicts Dopaminergic Deficiency in Parkinson’s Disease

Overview

This study demonstrates that brief smartphone-based motor assessments can effectively predict dopamine transporter (DaT) SPECT scan abnormalities and striatal binding ratios in Parkinson’s disease (PD) and isolated REM sleep behaviour disorder (iRBD). Machine learning models using smartphone-derived features achieved good classification performance, supporting their potential as scalable, cost-effective alternatives to DaT imaging.

Background

Parkinson’s disease is characterized by dopaminergic neuron loss leading to motor symptoms, with DaT SPECT imaging serving as a key diagnostic tool by quantifying dopamine transporter availability. Despite its clinical utility, DaT imaging is costly, requires specialized equipment, and involves radiation exposure, limiting frequent use. Prodromal markers like iRBD show early nigrostriatal dysfunction detectable by DaT SPECT, but scalable, accessible screening methods are needed. Smartphone-based digital assessments have shown promise in differentiating PD, iRBD, and controls, and in predicting motor severity scores, motivating their evaluation for predicting dopaminergic deficiency.

Data Highlights

Key Findings

• Smartphone-derived motor features can classify DaT scan status with AUC up to 0.80 using XGBoost models.
• Combining smartphone features with MDS-UPDRS-III scores improves classification performance (AUC up to 0.85 with logistic regression).
• Regression models predict striatal binding ratios with moderate accuracy (RMSE = 0.49, R² = 0.56).
• Gait, tremor, and dexterity features are the most predictive smartphone-derived markers of dopaminergic deficiency.
• Smartphone assessments offer a scalable, non-invasive complement to clinical evaluations and DaT imaging.

Clinical Implications

Smartphone-based motor assessments provide a practical, accessible tool to screen for dopaminergic deficiency in PD and prodromal populations, potentially reducing reliance on costly and less accessible DaT SPECT imaging. Incorporating these digital assessments alongside clinical scales like MDS-UPDRS-III may enhance diagnostic accuracy and facilitate earlier identification and monitoring of dopaminergic loss. However, larger independent validation studies are necessary before widespread clinical implementation.

Conclusion

Smartphone-derived motor features can effectively predict dopaminergic deficiency as measured by DaT SPECT, supporting their role as scalable adjuncts to clinical evaluation in Parkinson’s disease. These findings encourage further validation to establish digital assessments as accessible tools for dopaminergic status characterization.

References

1. Oxford Parkinson’s Disease Centre Study -- Smartphone-based motor assessments in PD
2. Yang et al. 4-year longitudinal study -- Correlation of SBR and MDS-UPDRS-III
3. Kerstens et al. -- Negative correlation between MDS-UPDRS-III and striatal binding
4. Parkinson’s At Risk Study -- Prodromal PD and DaT SPECT alterations