KELVIN: Computer Vision Based Biomarkers of Motor Dysfunction in Parkinson’s Disease

Diseases of the CNS (Central Nervous System) often affect mobility and behaviour. However, historically it has been extremely difficult to measure mobility and behaviour objectively. Instead, researchers have relied upon subjective and unreliable measures such as patient questionnaires and clinical scores. Not only do these tools yield poor quality data, but they are labour intensive, time consuming and expensive, and negatively impact on clinical trials in CNS disease by making them 30% more expensive and 50% more likely to fail than in other categories (e.g. cardiovascular). This has meant that many drug companies have shunned drug development for CNS disease, despite the fact that CNS disease is responsible for the largest burden of disease of any disease category.

Machine Medicine Technologies (MMT) is developing KELVIN, a video and computer vision analytics platform that is adapting the latest developments in AI to perform precise and objective measurement of mobility and movement in Parkinson's disease (PD). In collaboration with PD specialists at UCL/UCLH, through which they have access to large video datasets of PD patients, they are adapting and extending the latest computer vision and AI techniques to this clinical application. Through prospective video capture of PD patients they are furthermore providing clinical validation of the technology and optimising the video capture.

The technology they are building can be applied to multiple realms beyond PD and constitutes a strong foundation on which they aim to become the definitive medical video and computer vision platform, supporting a raft of other technologies such as medical robotics.