New research by IBM researchers and Michael J. Fox Foundation for Parkinson’s Research (MJFF) published in Lancet Digital Health details an AI model that predicts the progression of Parkinson’s disease. The model groups typical symptom patterns and examines longitudinal patient data to predict how quickly symptoms will progress over time, focusing on timing and severity.
In an effort to develop a more accurate AI model, researchers used de-identified datasets from Parkinson’s Progression Markers Initiative (PPMI). The dataset served as input for machine learning approach which enabled discovery of complex symptom and progression patterns.
This AI research relies on up to seven years of patient data, the most ever used in a Parkinson’s disease study. The researchers also limited their assumptions about progression pathways and found that patients often went through many different phases before experiencing typical symptoms.
Researchers used unique modeling techniques to gain more insights into disease states and progression pathways. It turns out that the state of a patient can vary in several factors, including their ability to perform activities of daily living; issues around slowness of movement and postural instability. Non-motor symptoms such as depression or anxiety are also important to consider when determining how patients will function day-to-day.
The researchers studied the progression of Parkinson’s disease and found many different ways it can progress. They also noted how accurate AI predictions were, even for those cases they had never seen before, because their model was trained on a single dataset. For instance, when given an advanced case to predict outcomes including dementia – which would be rare in most datasets-the algorithm still predicted accurately!