Data-Driven Neurology: Potential AI Patient Care Innovations

A session at the 2025 NeuroNet Pro Annual Summit discussed the value of collective clinical data and its impact on neurology, featuring a panel discussion of leading neuroscience physicians on leveraging data for clinical excellence and patient access through artificial intelligence (AI).

The session was sponsored by NeuroDiscovery AI, a comprehensive longitudinal neurology dataset compiled through partnerships with over 470 leading providers across the US. The dataset contains multimodal data extracted from millions of patients to derive meaningful insights.

“The goal has always been about personalized medicine,” said Howard Friedman, an economist and data scientist at Columbia University and session moderator. “And as we have richer data sets, I see the neurology space moving to where oncology went decades ago. As this data allows us to better target our treatments, that's going to give us the opportunity to have more and more focused care for patients.” Ajith Thomas, Chairman of Neurosurgery at Cooper University Healthcare, emphasized that more real-world data is available to advance treatment than ever before from sources such as electronic health record (HER) data, patient-generated data, and administrative data.

“The question is, when you have all this data, how do you make sense of it as a human being?” Dr Thomas said. “Because we all have a cognitive overload issue, and there's only so much we can process. We need the data to be presented to us in a manner that we can actually understand. That's where AI comes in.”

Dr Thomas explained that AI, especially in the form of generative AI and multimodal large language models (LLMs), has revolutionized health care by aiding in clinical decisions, drug discovery, and personalized medicine. He recalled an example from his time at Beth Israel Hospital where electrocardiogram (EKG) data was collected from a cohort of 180 000 patients to predict mortality and risk of arrhythmia. AI processing of the EKG data had an 80% mortality prediction accuracy score.

“At some point in the near future, just from an MRI of the brain, you should be able to predict who has Parkinson’s, who is going to get Alzheimer’s. Because the data is already there; it's just that we don't understand how to interpret that data,” Dr Thomas said.

He also introduced the concept of a “digital twin” created from real patient information, including genetic and behavioral data, laboratory and imaging studies, biomedical signals, and social determinants. A digital twin may be used to simulate outcomes of interventions, assist in biomarker and drug discovery, support clinical trials, and more.

Under the overarching umbrella of AI, Dr Thomas highlighted the importance of agentic AI, which not only supports human decision-making but can operate autonomously.

“Instead of just Googling the best restaurants in Nashville, with agentic AI, you'll be able to say, ‘Hey, find me the best Asian restaurant, make a reservation for 3 people,’ and it will go and figure out all that, and your credit card details, and do it for you,” he said.

From a clinical perspective, Taif Kaissi, MD, CEO of The Neurology Group Inc and NeuroDiscovery AI, focused on opportunities to use data analysis to improve outcomes and financial performances in health care practices. He explained that data can help in negotiating contracts, understanding insurance reports, and making more informed decisions as both a private practice owner and a clinician. To utilize the wealth of untapped data already available to many neurologists, Dr Kaissi recommended partnering with established companies capable of reducing the burden of managing raw data.

“Data is not a gold mine; data is a diamond mine,” Dr Kaissi said. “Look into this rough diamond that you have, figure out a way to partner up with somebody who has already figured out a way for you to utilize it for your own good. And quality [data] is very important.”

Reference

Friedman H, Thomas A, Kaissi T. Leveraging our collective data. Presented at: Neuronet Pro Annual Summit 2025; January 31, 2025; Nashville, TN.