At KansasCOM, researchers use AI to reveal unseen links between health, environment, and culture, advancing equity and improving patient outcomes.
At Kansas Health Science University’s Kansas College of Osteopathic Medicine (KansasCOM), students learn from faculty at the forefront of research and technology. Among them are two researchers who are applying artificial intelligence (AI) to uncover hidden patterns in health data. They demonstrate how AI can help advance care for all communities, including those hardest to measure.
In the clinic, AI is already transforming health care by reducing errors and improving diagnostics in fields like radiology and pathology. AI models are supporting clinical decision making; just as Netflix learns viewing habits and recommends the next TV show fans are likely to binge, AI in medicine can analyze layers of patient data, such as surgeries, life events, even environmental factors, to suggest a best course of care.
For public health researchers, too, AI can reveal patterns and predictions that would be difficult to spot on their own. And it can retain patient privacy through generating de-identified, sharable datasets. Yet the power of AI also raises a critical question: Who is represented in the data?
Revealing Hidden Patterns From Diverse Data Sources
According to data scientist and AI research engineer Jamie Fairclough, PhD, doctors and medical researchers can leverage AI for its unique strength: revealing the unseen determinants of health by layering different types of data into a clearer picture.
Dr. Fairclough, Engineering Faculty and Director of Mental Health Evaluation and Assessment at Dartmouth College, became an adjunct professor at KansasCOM in 2025 and has worked with the Florida Department of Health and other state health departments.
She explained the power of harnessing data for clinicians: a patient repeatedly treated for a chronic medical condition might have an underlying cause that isn’t immediately obvious. By using AI-powered tools, doctors may be able to draw on public health data to uncover environmental factors contributing to the condition, such as lack of air conditioning or presence of mold in the patient’s home.
While Dr. Fairclough points to environmental determinants of health, Saajan Bhakta, PhD, Associate Dean of Research at KansasCOM, shows how AI can also illuminate cultural and behavioral factors.
For his research on attention-deficit hyperactivity disorder (ADHD) within South Asian populations, Dr. Bhakta and a team of KansasCOM students explored patterns in interview and survey data, using AI tools to support their analysis.
After manually coding the interview transcripts, student researchers used qualitative data analysis software with AI to provide a “second set of eyes” and surface connections between the data sources. It helped to identify recurring themes in the data.
“By using advanced analytical tools, we were able to identify meaningful patterns … correlations between parenting styles, cultural perceptions of ADHD, and reported access to care,” Dr. Bhakta says.
Expanding Research With De-Identified Data
To practice interpreting complex patterns safely, researchers often turn to a growing tool in AI-powered research: de-identified, or synthetic, datasets.
With tools powered by generative AI, data scientists can create synthetic datasets that remove identifying patient information. Dr. Fairclough says, “Synthetic datasets allow institutions to collaborate with each other. For example, if a hospital in Wichita wants to collaborate with a medical center in Boston, instead of having access to personal health records, they may generate a synthetic dataset with a company that has developed a secure, HIPAA-compliant data generation platform that preserves the statistical validity of the original dataset while protecting patient privacy.”
She emphasized that shared access to a dataset presents a valuable opportunity for collaboration, allowing researchers to build models and conduct analyses using common, uniform data. Dr. Bhakta says synthetic datasets can also be used for practice.
“Synthetic datasets offer a safe environment for students to learn the full research process, from data cleaning and modeling to visualization and interpretation, without any privacy concerns,” he says. “Students can experiment, test hypotheses, and refine their analytical skills. This hands-on practice helps build confidence and prepares them for working with sensitive real-world datasets.”
Synthetic data can be highly effective for certain research projects, such as market analysis, but it may fall short in capturing the nuanced complexity required for health studies. Findings must be validated against real-world data to ensure accuracy.
Collecting Data From Underserved Populations
While AI’s ability to reveal hidden patterns and expand datasets is powerful, both researchers emphasized that these tools are only as valuable as the data they are built on. And too often, medically underserved populations are missing from that data.
As Dr. Fairclough explained, “funding may not go to those communities because there’s insufficient data to support it.” Without evidence, health departments struggle to justify interventions.
She noted that many state health agencies still rely on outdated methods like landline phone surveys for data collection. Her own grandmother, skeptical of unfamiliar phone numbers, would never be counted, she says. That blind spot leaves entire communities invisible to the system.
By applying machine learning techniques and innovative thinking, she argues, researchers can begin finding and even filling these gaps to build evidence for equitable funding and programs.
Dr. Bhakta says that gaps in data collection from underserved communities are driven by distance and systemic barriers such as language, culture, and distrust of institutions. He pointed to ways AI could make data collection more inclusive, from digitizing mailed forms for those without internet access, to deploying conversational AI to conduct phone interviews that feel personal and approachable.
AI outputs are only as good as their inputs, so if models are trained primarily on data from well-served populations, they risk deepening health inequities. Researchers must engage communities, build trust, and ensure underrepresented voices are reflected in the data.
KansasCOM’s Role
These insights translate directly into education at KansasCOM. Students learn not only how to use AI tools for research, but also to ask deeper questions about equity and representation in health data.
In the medical field where technology is evolving rapidly, KansasCOM’s role is clear: to equip future physicians and researchers to use AI responsibly, turning hidden patterns into solutions that benefit every community.