Models Predict Parkinson's Fall Risk Using Clinical Scores

Researchers validated two clinical models that estimate next-year fall risk in people with Parkinson's disease.

Updated on Nov. 9, 2026 in Alzheimer’s

Bold flat-color editorial illustration of a single wooden walking cane leaning against a cream field, a navy and cream composition with one red accent. — Two clinical models estimating next-year fall risk in Parkinson's disease held up in external validation, though biomarkers added little beyond clinical scores. AI Illustration. Upload story photo >

Live Poll

Would you schedule a checkup if your doctor could screen you for fall risk?

Researchers developed clinical models predicting next-year fall risk in Parkinson's disease and validated them in external cohorts. Biomarker augmentation modestly improved discrimination in the development cohort but not in external validation.

Why it matters

Falls remain a major concern in Parkinson's disease, so identifying at-risk patients early is a priority for clinicians and researchers. The models offer a framework for risk stratification that could also inform how clinical trials are designed.

The incident fall model achieved an AUROC of 0.704 in the Parkinson's Progression Markers Initiative and 0.630 in University of Pennsylvania cohorts, with incident fall risk rising from 7.1% to 18.5% across risk groups. Biomarker augmentation's benefit has not yet been confirmed outside the development cohort.

The players

University of Pennsylvania

A private Ivy League research university whose cohorts provided the external validation data for the fall risk models.

Parkinson's Progression Markers Initiative

A multicenter observational study that supplied the development cohort used to build the Parkinson's fall prediction models.

The details

The models were built from motor severity, disease stage, cognition and depression measures, with the incident model developed in 773 visit-pairs and validated in 589. The recurrent model, using Hoehn and Yahr stage, depression and processing-speed scores, was developed in 302 visit-pairs and validated in 137, with recurrent fall rates rising from 22.2% to 65.0% across risk groups.

2026: Study published in Annals of Neurology.

Health Landscape

The work builds on the Parkinson's Progression Markers Initiative, a longitudinal dataset used across Parkinson's research. Its external validation in University of Pennsylvania cohorts tests whether models trained on that resource generalize beyond their original population.

For people living with Parkinson's, a validated fall risk score could help clinicians decide which patients need closer monitoring or preventive steps. The findings are early and still require confirmation before they change routine care.

The takeaway

Fall risk in Parkinson's can be estimated from routine clinical measures such as motor score, cognition and mood. Patients and caregivers should raise fall concerns with their care team, since the tools described here are not yet standard practice.