KENNER – LSUS computer science student Devesh Sarda won first place in the oral division of the Southern Biomedical Engineering Conference for his research into Alzheimer’s disease.

Sarda constructed an artificial intelligence model that can predict the severity of Alzheimer’s disease in a patient by ingesting information such as MRIs and electronic health records from open-source databases like the Alzheimer’s Disease Neuroimaging Initiative (ADNI) and the Open Access Series of Imaging Studies (OASIS).

“We were looking at this multimodal data and are able to predict which stage of the disease the patient is in,” said Sarda, who added that all information in public databases isn’t tied to a specific patient but is anonymous. “We know that doctors and radiologists who are looking at these scans are overwhelmed because of the volume of patients they are dealing with in an America with an aging population.

“AI isn’t going to make a diagnosis for a doctor, but it can be a helpful tool in assisting in the diagnosis and treatment process. Can an accurate AI tool help a doctor get to a diagnosis faster?”

The graduate student said his model, which is part of his thesis research, detects current stages of Alzheimer’s disease correctly about 87 percent of the time.

The Shreveport native has worked in LSUS’s Artificial Intelligence and Machine Learning Lab for nearly four years, starting as an undergraduate student under the direction of computer science faculty member Dr. Subhajit Chakrabarty.

Sarda’s exposure to expansive datasets combined with his “fascination of the brain” has led him into research with spinal stenosis and now Alzheimer’s.

“Alzheimer’s in an area where there’s a lot of data, and one of the first papers I read when I started here was about Alzheimer’s,” Sarda said. “I’ve gained some expertise with MRIs specifically dealing with the brain and the entirety of the central nervous system, and these are things that impact people’s personal lives.

“If we can diagnose and treat Alzheimer’s faster, catch it before it becomes severe, then people will still retain the memory of their loved ones. So that’s a constant driving force as well.”

Chakrabarty, who is leading a number of student-involved research projects involving AI and medicine, said Sarda will publish in a journal as part of his first-place reward.

“Devesh presented extremely well, and he was able to answer a lot of questions that came up,” Chakrabarty said. “There were about 200 presentations from students all across the Southeast.

“Experiences like this benefit all the students in our lab as they conduct research that’s really more on a doctoral level instead of just a graduate or undergraduate level.”

Sarda isn’t stopping with his model just being able to predict current stages of Alzheimer’s.

He is progressing the work in an attempt to more accurately predict future disease.

“There can be visual biomarkers and sometimes even hidden biomarkers, stuff we can’t see but that AI models might because they can consistently deconstruct and reconstruct an image to only keep aspects that they want,” Sarda said. “Every patient is unique, but if we’re looking at a series of images that shows disease progression, is there a way to arrive at the same result (diagnosis/treatment) but sooner?

“AI can be a tool at doctors’ disposal to help them make their decision sooner and ultimately help the patient faster.”