For some, the eyes are a window into the soul. But for Jayashree Kalpathy-Cramer, PhD, professor of ophthalmology at the University of Colorado School of Medicine, they're a window into human health.
The researcher was granted $300,000 by The Michael J. Fox Foundation this fall to analyze clinical data curated at the Sue Anschutz-Rodgers Eye Center using artificial intelligence (AI) in an effort to identify biomarkers of Parkinson's disease, a progressive disorder that affects the nervous system and causes uncontrollable movements, such as shaking, throughout the body.
"This approach could be impactful because there is a lot you can learn from looking at the eyes," Kalpathy-Cramer, chief of the Division of Artificial Medical Intelligence in Ophthalmology, says of the research she and her team in the CU Department of Ophthalmology will undertake over the next year-and-a-half. "The goal is to predict the disease well before symptoms manifest. This would mean that the patient and the clinician are aware sooner so we can improve future care."
The eyes have it
Kalpathy-Cramer is optimistic about the role of AI in Parkinson's disease detection, especially because AI's implementation to analyze retinal imaging in other diseases -; such as schizophrenia, dementia, and cardiovascular risk factors -; has shown to be fruitful.
The eye is very accessible compared to other parts of the body, like the heart, for example. We can easily take a photo of the eye, and it can tell us a lot about neurological or cardiovascular conditions. It's an easy way to get a sense of the overall health of the patient."
Jayashree Kalpathy-Cramer, PhD, professor of ophthalmology, University of Colorado School of Medicine
Kalpathy-Cramer and her team will also have a lot of data to pull from at the eye center for the research. Working with UCHealth's Health Data Compass, School of Medicine IT, and guidance from Colorado Multiple Institutional Review Board, plus regulatory, compliance, and informatics leaders on campus, her team has created a large retrospective dataset consisting of images and health records for patients seen at the Rocky Mountain Lions Eye Institute over the last decade in a highly secure research repository.
They are currently using AI to better curate the data through the analysis of structure and unstructured records.
"We will have access to eye exams and a lot of other information, including basic demographics, potential diseases a patient might have, and when they were diagnosed," she explains.
Using that data to train AI to look for biomarkers could clue researchers into a lot about Parkinson's disease, like how it presents and potentially how it evolves over time.
A view of the future
Kalpathy-Cramer hopes to expand the scope of work in the future and look at larger and more diverse data.
"If we can incorporate different modalities, we might be able further improve our ability to more accurately predict disease prevalence and incidence," she says. "New developments in machine learning and AI allow us to analyze data at unprecedented scales. Our goal is to utilize the large amounts of retrospective clinical data to better care for our patients in the future."
Work on the research is expected to start Nov. 1 and last 18 months.
"We are extremely grateful to The Michael J. Fox Foundation and their interest in exploring and funding this work," Kalpathy-Cramer says. "It's a phenomenal initiative, and they've funded a lot of groundbreaking work. It has been a wonderful experience to work with them so far."