ODU and CHKD to Use AI to Map Childhood Asthma Hotspots

Children ages 5-14 in Hampton Roads were hospitalized for asthma at nearly twice the statewide rate in 2024, according to the Virginia Department of Health, prompting local researchers to deploy geographic and artificial intelligence tools.

For Jessica Parrott, the most concerning part is that asthma is “100% treatable.” Parrott serves as the director of the Pediatric Nurse Practitioner Program at Old Dominion University and works as a nurse practitioner with Children’s Hospital of the King’s Daughters Urgent Care.

“It’s really not okay that those kids are having to have so many emergency department visits and then potentially death from it when we can do something about it beforehand,” she said.

Environmental Triggers Drive Childhood Asthma Hospitalizations

Childhood asthma flares stem from multiple environmental triggers, including smog, smoke, pollen, dust, mold, and pet dander. Pinpointing where hospitalizations concentrate serves as the initial step toward addressing the underlying causes, according to Aditya Chakraborty, an assistant professor at ODU’s Joint School of Public Health.

“We plan to use local hospitalization data and use the geographical hotspot analysis to see where childhood asthma hospitalizations are concentrated across Hampton Roads and Western Tidewater, and to explore what might be contributing to those patterns,” Chakraborty said.

The upcoming analysis will apply multiple artificial intelligence models to hospitalization records from Children’s Hospital of the King’s Daughters. Patient privacy will be maintained by anonymizing records to zip codes exclusively, alongside demographic variables, average home ages, income levels, and the availability of neighborhood healthcare providers.

Mapping Hotspots Helps Direct Medical Resources and Early Care

Researchers expect the resulting maps will direct healthcare providers and community stakeholders toward areas requiring targeted medical resources. Chakraborty described the clinical intent for patients with frequent prior episodes.

“Imagine a child who’s already had several asthma-related visits,” Chakraborty said. “If we can develop and validate a tool that helps a primary care team recognize when a child might need closer follow up, the team could reach out sooner.”

Parrott identified patient access barriers as a primary driver behind elevated hospitalization numbers in the region. She pointed to operational hurdles observed through an ODU-operated free clinic network.

“We see that regularly with our free clinic that we run with ODU, because kids, even if they have insurance and they have a provider, they often can’t even get into their provider,” Parrott said.

Expanding Previous Machine Learning Work

The current initiative expands upon a previous research effort published in July by Chakraborty, which examined machine learning models to identify pediatric patients carrying persistent asthma diagnoses based on medication usage timing, symptom dates, and bedroom carpeting.

That earlier work demonstrated that predictive algorithms can enhance early diagnostic accuracy and encourage prompt clinical intervention. The team is currently securing final authorizations and gathering data sets for the expanded regional study.

“Our next step with the CHKD is to bring those methods closer to home,” Chakraborty said.

Project leaders anticipate completing the analysis and releasing findings sometime in 2027.

Disclaimer: The information provided in this article is for educational and scientific communication purposes only and does not constitute medical advice. Always consult with a qualified healthcare provider regarding any medical condition, diagnosis, or treatment plan.