Two clinical validation studies featuring Exer’s Clinical AI platform were presented in the same week, showcasing the growing impact of computer vision in clinical care.
Advancing Tremor Assessment at the Congress of Neurological Surgeons (CNS) Annual Meeting
At the Congress of Neurological Surgeons (CNS) 2025 Annual Meeting, researchers from Duke Health presented results validating Exer’s AI-based motion analysis, showing that the technology used in Exer’s assessments can also objectively measure tremor severity in neurological disorders like Essential Tremor and Parkinson’s Disease.
The study, “AI Computer Vision to Assess Tremor Severity: Validation Against the Clinical Rating Scale for Tremor,” highlights how Exer’s Clinical AI platform aligns closely with established clinical rating scales, paving the way for faster, more consistent tremor evaluation in both research and care settings.
Key Findings
The AI model achieved exceptional inter-rater reliability (ICC = 0.96), matching clinician consistency (ICC = 0.94).
Model-clinician agreement ranged from κ = 0.79–0.88, with 91–99% of ratings falling within one point of clinician scores.
These results demonstrate that validated computer-vision tools can provide objective, consistent tremor ratings aligned with standard clinical methods.
Validating Upper-Extremity Motion at the American Society for Surgery of the Hand (ASSH) Annual Meeting
Meanwhile, at the American Society for Surgery of the Hand (ASSH) 2025 Annual Meeting, independent researchers presented podium findings validating Exer’s 2D AI motion analysis for hand, wrist, and forearm range of motion assessments.
The study, “Validation of a 2D Artificial Intelligence Camera to Assess Hand, Wrist, and Forearm Range of Motion,” compared Exer AI’s iPad-based measurements to traditional manual goniometry, focusing on accuracy, consistency, and time efficiency across multiple joint motions.
Key Findings
Excellent inter-rater reliability (ICC > 0.9) versus manual goniometry for 96% of hand/wrist motions.
Strong concurrent validity (PCC > 0.8) for 96.3% of motions.
78% faster assessments (average 60 seconds vs. 273 seconds; p < 0.05).
Bland-Altman analysis confirmed Exer’s ROM measurements were within ± 5° of manual goniometry across all motions.
Exer AI works seamlessly without the need for sensors or wearables to improve patients’ lives and providers’ decision-making across complex care needs in multiple specialties, including orthopedics, neurology, pain/spine, PM&R, geriatrics, and more.
