The Michael J. Fox Foundation (MJFF) and the research arm of IBM have developed an AI model that can group typical Parkinson’s disease (PD) symptom patterns. This AI model can accurately identify the progression of these symptoms in a patient, regardless of whether or not they are taking medications to mask those symptoms.
This means that in the future, doctors will be capable of utilizing AI to predict how their patients’ diseases will proceed, allowing them to manage their symptoms better. This was one of the primary goals the two organizations had set out to achieve from the start, according to the discovery results published in The Lancet Digital Health.
The human motor system uses a sequence of distinct movements to accomplish bodily activities, e.g., arm swinging when walking, running, or jogging. These movements and the transitions between them produce activity patterns that may be monitored and examined for indications of Parkinson’s disease. Researchers study the physical measurements obtained from Parkinson’s patients, which differ from those taken from non-patients, and the development in those differences over time indicates disease progression. But it was unknown why some people with Parkinson’s will have their disease turn more severe than others until the recent breakthrough by Big Blue.
Since July 2018, IBM Research and MJFF have been collaborating to see how machine learning may be used to assist physicians in better understanding the underlying biology of Parkinson’s disease, especially given how it proceeds so differently from person to person. This machine-learning algorithm evaluated data from patients over seven years and identified trends in their symptoms that were connected to neurodegeneration. Based on this insight, the team created an AI computer model that might help doctors anticipate how a patient’s condition would evolve, aiding them in giving the correct treatments at the right time or deciding who would benefit the most from a clinical trial. The researchers identified eight distinct states in Parkinson’s disease, each containing both motor and non-motor symptoms, and discovered that the disease might shift between them in no particular sequence over time. One of these states included severe cognitive impairment.
IBM stated that its aim is to use AI to help with patient management and clinical trial design. “These goals are important because, despite Parkinson’s prevalence, patients experience a unique variety of motor and non-motor symptoms,” the company added.
The Michael J. Fox Foundation’s Parkinson’s Progression Markers Initiative provided the data for this research. It is a clinical study originally started in 2010 in cooperation with more than 30 biotech, pharmaceutical, non-profit, and private firms. The study, which included over 1,400 patients from throughout the world, has gathered years of patient data from health records, wearable devices, and cellphones, as well as sequencing genomes and analyzing specimens obtained during their condition. According to IBM, this is the largest and most robust volume of longitudinal Parkinson’s patient data to date.
The findings were compared to those of a control group of 610 Parkinson’s disease patients from the National Institute of Neurological Disorders and Stroke Parkinson’s Disease Biomarker Program (PDBP). This aided in the validation of the AI model that IBM researchers had been working on since mid-2018.