Hi everyone! I am an associate professor in the Department of rehabilitation and movement sciences at Rutgers University. I am the director of the digital health lab within the School of Health Professions. I am originally from Canada where I earned my PhD in neuroscience at McGill University. My lab focuses on developing, testing, and validating technology-based methods, such as telehealth, wearable sensing, and AI, to address clinically relevant questions related to chronic neurological conditions, including Parkinson’s disease. I am currently working with several datasets, including the PPMI dataset, and would love to work with others in the DCoP on improving data harmonization and reproducible data analyses. Looking forward to talking to you!
Hello Jean! Nice to meet you!
I’m Daniel, and neurologist specialized in movement disorders and early career data scientist. My PhD. project is related to the application of unsupervised machine learning to identify biological subtypes of PD. My project uses data from both PPMI and PDBP. I have also previously published studies related to telemedicine usage.
I have great interest in wearable sensors and AI and their health applications. I find it vital that professionals with different expertise and background join forces to make new and relevant discoveries. I think that much of the relevant clinical questions posed by us, clinicians, can be better explored with the kind of technologies you deal with.
Just a quick question - have you looked at PPMI’s wearable sensor’s data? I acknowledge its existence but haven’t touched it. I would be greatly interested on how these data relates to our clinical objective measurements from the MDS-UPDRS III. Overall, I think that our clinical scale is prone to some biases that this kind of objective data overcomes. Have you pursued this analysis or do you know of articles that did so? I think this is a great topic to study.
Be welcome!
Hi Daniel! Nice to meet you. I agree that objective measures of symptoms, complications, and other clinically important metrics are needed, and wearable sensor data combined with AI and ML methods could provide this valuable information.
So, I am working with the Verily data within the PPMI database as well as other wearable data collected by myself and collaborators over the past 15 years. There are significant technical challenges that we are still tackling. For instance, the identification of tremor is relatively simple and accurate even in the home setting. Similarly, determining the presence and severity of dyskinesia when a patient is at rest is also simple but becomes much more difficult during ADLs even using complex AI models. There are still several groups developing models to identify the various symptoms of PD and their severity from wearable sensor data both in clinic and home settings. While my lab still works on this, our focus is now on using these data to predict disease progression, response to treatment (e.g., drug, DBS, PT, etc.), and to use these sensors as early warning systems for patients and their care team.
I would be more than happy to discuss this further with you! BTW, colleagues and I have just started a study in Canada that will actually collect more data than the Verily study so there should be opportunities to collaborate with like-minded researcher on this rich dataset.
Best,
Jean-Francois