Research from Old Dominion University may offer public health officials a more timely way to spot spikes in deadly diseases before they become full-blown crises.
The framework, developed by Aditya Chakraborty, assistant professor in public health, analyzed weekly mortality data to flag unusual increases in 14 major diseases such as heart disease, diabetes, and COVID.
The model also measures how multiple risk factors interact, such as obesity, smoking and limited access to care, offering a broader picture of analyzing public health threats.
Chakraborty said responding quickly is a pressing need in Hampton Roads, where mortality rates for heart disease and diabetes remain higher than the state average, according to data from the National Institutes of Health.
In Virginia Beach, Virginia’s largest city, the average annual count of heart disease deaths climbed to nearly 800 from 2019 to 2023, ranking second-highest among Virginia localities.
In Norfolk and Chesapeake, the diabetes death rate was 10% higher than the statewide average from 2019 to 2023.
The model flags when mortality counts exceed expected levels, calculating the probability of those spikes.
“If the mortality is exceeding, that means something is going wrong,” Chakraborty said. “And we need to inform that to the public health professionals, so they should be more alert.”
Chakraborty said the model not only forecasts the mortality rate with more than 90% accuracy but also projects survival timelines, estimating how long it might take for a death or other critical health event to occur under certain conditions.
That, Chakraborty explained, allows researchers to test techniques that highlight which risk factors should be reduced first — and by how much — to extend survival or prevent deaths.
“If we have all this information related to behavioral risk factors, we might want to find out the optimum levels at which the time to death is minimized to improve patient outcomes,” Chakraborty said.
The next step for Chakraborty is to apply the model to cancer mortality data provided by the Virginia Department of Health, focusing on the five most common cancers in Hampton Roads: prostate, lung, colorectal, breast and pancreatic.
By combining statewide mortality records with local geospatial codes, he hopes to give policymakers a neighborhood-level view of where deaths are concentrated and how risks are evolving.
At Sentara, one of the largest health systems in Virginia, professionals are training and using AI models for early detection of sepsis and heart disease.
A sepsis model already in use scans 66 patient variables every 15 minutes and can warn doctors up to eight hours before symptoms appear.
Another AI tool in development analyzes routine chest images to detect hidden risks of heart disease in patients who show no outward signs.
“It would give us the opportunity to detect something that could be definitely life-threatening before it really has any symptoms,” said Joe Evans, Sentara’s chief health information officer. “So it has a great impact on patients' lives.”
The Virginia Department of Health declined WHRO’s request for an interview.
