Bold claim: smartphones could soon screen for early dopamine loss in Parkinson’s disease without radiation-heavy brain scans. And here’s how this could transform early detection. In a recent NPJ Digital Medicine study, researchers paired everyday smartphone movement data with standard clinical motor scores to predict dopaminergic deficit, offering a potentially accessible, noninvasive screening pathway for prodromal and manifest Parkinson’s disease.
Parkinson’s disease (PD) disrupts dopamine pathways in the nigrostriatal region, leading to motor difficulties. Confirming dopamine deficiency traditionally relies on advanced imaging methods that can be costly, expose patients to radiation, and aren’t universally available. The study explores using widely available smartphones alongside clinical assessments to gauge motor function and infer dopamine status.
How PD is typically diagnosed
Dopamine transporter (DaT) imaging with single-photon emission computed tomography (SPECT) is commonly used to confirm dopamine deficiency. The resulting striatal binding ratio (SBR) reflects DaT levels in key brain areas like the caudate nucleus and putamen. A lower SBR signals greater dopaminergic neuron loss and worse motor function, correlating with motor scores on the Movement Disorder Society‑Unified Parkinson’s Disease Rating Scale Part III (MDS-UPDRS-III) and symptoms such as slowed movement, posture and gait changes, speech, and facial expression loss.
PD is classified as an alpha-synucleinopathy, a brain disorder driven by abnormal accumulation of the alpha-synuclein protein. Diagnosing prodromal PD is crucial because early intervention can shape outcomes. For example, isolated REM sleep behavior disorder (iRBD) raises the annual risk of progressing to overt PD or dementia with Lewy bodies (DLB), and a substantial portion of iRBD patients show early nigrostriatal dopaminergic deficiency.
Digital tools are increasingly used to screen for PD, including the eight‑minute Oxford Parkinson’s Disease Centre (OPDC) smartphone app. Prior work from the same group showed OPDC can differentiate healthy individuals, iRBD patients, and PD patients, and can even predict MDS-UPDRS-III scores.
What this study adds
The researchers examined whether machine learning models can use smartphone-derived movement data, in combination with MDS-UPDRS-III scores, to predict DaT status and SBR. If successful, this approach could triage people who might have an abnormal DaT scan, prioritizing access to further imaging when needed and reducing unnecessary testing.
Study design and findings
- Participants: 93 individuals with iRBD, PD, or neither, who had undergone both a DaT scan and a smartphone-based motor assessment within the preceding year.
- Method: Machine learning models were trained on smartphone movement data to predict whether the DaT scan would be positive or negative.
- Results: Using 100 unique DaT scans, the smartphone-based model achieved about 80% discrimination, comparable to models built on MDS-UPDRS-III scores. Combining smartphone data with MDS-UPDRS-III raised the area under the curve (AUC) to about 85%. The logistic regression model using MDS-UPDRS-III alone achieved 82% AUC, while combining both data sources yielded the highest performance at 85% AUC. Predictions of SBR from these models were most accurate for gait, manual dexterity, and tremor.
- Interpretation: The high-frequency, multi-dimensional smartphone sampling captures subtle motor features that may be missed during routine exams, enabling more sensitive detection of subclinical tremors and dopamine deficiency. Integrating digital measurements with clinical scores offers a potentially more accurate and scalable approach than either source alone.
Important caveats
- When analyses focused only on milder PD cases, model performance diminished, suggesting motor-based assessments alone may be less reliable for tracking early progression.
- The study’s sample size is modest, so findings should be validated in larger, diverse cohorts before clinical adoption.
Implications and takeaways
If validated, a combined digital-clinical framework could serve as a cost-effective, widely accessible pre-screening tool for DaT imaging. This could enable earlier intervention and more frequent monitoring, empowering patients and clinicians by bringing closer-to-imaging insights into routine, noninvasive visits.
Journal reference
Gunter, K. M., Groenewald, K., Aubourg, T., et al. (2025). Smartphone-based prediction of dopaminergic deficit in prodromal and manifest Parkinson’s disease. NPJ Digital Medicine. DOI: 10.1038/s41746-025-02148-2. https://www.nature.com/articles/s41746-025-02148-2
