Smartphone motor tests can predict dopamine deficiency in Parkinson’s disease without brain scans

By pairing everyday smartphone motion data with established clinical scores, researchers demonstrate a promising path toward accessible, radiation-free screening for early dopamine loss in Parkinson’s disease and related prodromal states.

Study: Smartphone-based prediction of dopaminergic deficit in prodromal and manifest Parkinson’s disease. Image Credit: mozakim / Shutterstock.com

Parkinson’s disease (PD) is a neurodegenerative disease characterized by disruption of dopamine-dependent pathways in the nigrostriatal pathway, a region of the brain that regulates coordinated voluntary movement. Dopamine deficiency is currently confirmed using advanced methods that are often expensive, involve radiation exposure, and have restricted accessibility.

A recent study published in the journal NPJ Digital Medicine explores the use of smartphones coupled with clinical scores to evaluate motor function and predict dopamine deficiency.

How is PD diagnosed?

Dopamine transporter (DaT) single-photon emission computed tomography (SPECT) is often used to establish the diagnosis of dopamine deficiency in PD. SPECT subsequently quantifies the striatal binding ratio (SBR), which reflects DaT levels in key regions of interest (ROIs) such as the caudate nucleus and putamen.

The SBR predicts several aspects of PD progression, as a low SBR indicates greater loss of dopaminergic neurons and motor dysfunction. This correlates with the contralateral Movement Disorder Society-Unified Parkinson’s Disease Rating Scale Part III (MDS-UPDRS-III) motor scores in PD, as well as with symptoms such as bradykinesia, posture, gait, speech, and loss of facial expression.

PD is considered an alpha-synucleinopathy, a brain disorder caused by the abnormal accumulation and misfolding of the alpha-synuclein protein within neurons and other brain cells. As a neurodegenerative disease, it is crucial to diagnose prodromal forms of PD to prevent or reduce its severity through appropriate interventions.

The presence of isolated REM sleep behavior disorder (iRBD) is associated with a 6% annual risk that an existing subclinical alpha-synucleinopathy may convert to overt PD or dementia with Lewy Bodies (DLB). Over 60% of people with iRBD exhibit early signs of nigrostriatal dopaminergic deficiency, 30% of whom subsequently develop alpha-synucleinopathy within three years.

Digital tools are widely used for PD screening, which includes the eight-minute Oxford Parkinson’s Disease Centre (OPDC) smartphone application. Previously, the researchers of the current study reported that OPDC can accurately discriminate among healthy individuals, those with iRBD, and PD patients, and also predict MDS-UPDRS-III motor scores.

The current study builds on the correlation between DaT and motor scores to determine whether machine learning models can use smartphone-based data to predict DaT status and SBR accurately. If successful, this inexpensive and easily accessible approach could be adopted to triage individuals at a greater risk of an abnormal DaT scan.

Study findings

The current study included 93 patients with iRBD, PD, or neither, all of whom had previously undergone both a DaT scan and a smartphone-based assessment within the following year. Machine learning models were trained on smartphone-collected data to predict whether the DaT would be positive or negative.

Using 100 unique DaT scans, the smartphone data model achieved a discrimination value of 80%, which is comparable to the model based on MDS-UPDRS-III scores. When both data sources were incorporated into the model, an area under the curve (AUC) value of 85% was observed.

The logistic regression model based on MDS UPDRS-III, or on both, performed slightly better overall, with AUCs of 82% and 85%, respectively. Regression models predicted SBR with modest accuracy, with the highest predictive value for gait, manual dexterity, and tremor.

The smartphone-based assessment uses repeated high-frequency sampling of movement to measure it across multiple dimensions. Thus, this method captures clinical features that healthcare workers may miss, enabling more sensitive detection of subclinical tremors that often characterize early dopamine deficiency.

Based on these findings, the authors postulated that the ability to identify greater detail, combined with standardized motor scores, could significantly improve the accuracy of predicting SBR. Both the smartphone-based model and MDS-UPDRS III scores were comparable in their ability to discriminate between classes. However, the clinical score performed better with logistic regression than with the more complex smartphone-based model.

This underscores the added value of integrating digital assessments while highlighting the importance of model selection based on data complexity and dimensionality.”

All models were less effective when subjected to an additional logistic regression analysis that included only milder PD cases, suggesting that motor-based assessments alone may be less reliable for predicting PD progression during early disease. Despite the small sample size, the study findings confirm the feasibility of combining smartphone-based motor assessment with clinical MDS-UPDRS-III scores to predict DaT scan status in people with iRBD and PD.

If confirmed, this combined clinical and digital framework could provide a cost-effective and widely accessible pre-screening tool for DaT imaging – bringing the potential for earlier intervention and more frequent monitoring into the hands of patients and clinicians alike.”

Adblock test (Why?)

Article