Developing novel tools to accurately identify dementia subtypes: application of wearable based gait analysis and machine learning approaches
Principal Investigator: Rana Zia Ur Rehman
Many different subtypes of dementia have similarities in clinical presentation, but a more accurate diagnosis is important to allow individuals to understand and manage their condition, as well as making plans for care. To allow this accurate diagnosis, easy-to-use and inexpensive screening tools for differential dementia diagnosis are required.
Gait has potential as a marker of dementia subtype, with early evidence suggesting Alzheimer’s Disease (AD) and Lewy body dementia (LBD) have unique signatures of gait impairment. Additionally, gait may be a hallmark of cognitive decline in Parkinson’s Disease (PD), identifying those most likely to progress to dementia.
Advanced computational analysis using data-driven approaches such as machine learning (ML), including deep neural networks paired with wearable-based gait analysis, can automatically identify early phases of PD, and may be applicable to identification of dementia subtype.
This project aims to develop novel ML tools to support clinicians in making an accurate identification and diagnosis of dementia subtype.