Photo: Sean Anthony Eddy/Getty Images
At HIMSS25, experts will share the details of PCCI’s Know Thy Patient (KTP) algorithm, which leverages an unsupervised machine learning method called clustering to identify groups of individuals based on similar patterns of healthcare utilization and access, rather than just disease state, according to Yusuf Tamer, PhD, principal data and applied scientist for PCCI.
PCCI is the Parkland Center for Clinical Innovation in Dallas, Texas, a nonprofit healthcare analytics research and development organization.
“Know Thy Patient: AI/ML-Driven Clustering of Diabetes/Hypertension Populations” is being held Thursday, March 6, from 2-3 p.m. in the Venetian | Level 2 | Veronese 2501, at HIMSS25 in Las Vegas.
In Dallas County’s safety-net population, the algorithm identifies clusters of diabetic and hypertensive patients with a combination of social and clinical risk factors.
luster analysis uncovers underlying, actionable risk drivers such as criminal justice involvement and immigration concerns, according to Tamer.
“Identifying underlying risk factors such as criminal justice involvement and immigration concerns provides a comprehensive understanding of patients’ socio-demographic characteristics and potential drivers of ineffective patterns of health service utilization,” Tamer said. “These underlying factors are difficult to uncover using routine questionnaires for obvious reasons, yet show up as underlying factors in a cluster analysis.”
These social factors also contribute to translation issues, and other culturally driven hesitations on healthcare access, he said. By leveraging this information, Parkland can enhance its population health strategy, ensuring that interventions are more precisely targeted to address the needs of diabetic and hypertensive patients.
For instance, community outreach in specific settings perceived as “safe” or “friendly” might be more effective in engaging some patients than digital health or office-centered approaches.
Additional in-depth analyses identify missed and potential opportunities for care engagement that inform workflow modifications using both traditional, EHR-based standing orders and nontraditional modalities such as telehealth and mobile units.
This approach, applied to diabetic and hypertensive patients in Dallas County, highlights the importance of considering factors beyond just disease category, such as demographics, utilization, payer class, digital engagement and other social determinants of health, Tamer said.
“Clustering patients into cohorts with high degrees of similarity across clinical utilization, personal, and behavioral characteristics allows us to understand the unique medical and social challenges faced by each group and how they impact access to quality care and health outcomes,” Tamer said. “This understanding enables the development of targeted clinical programs that are tailored to the specific needs of these patients, ultimately improving patient outcomes and resource allocation.”
The data sets and analytical approaches are scalable and replicable to other vulnerable populations nationwide.
“Know Thy Patient: AI/ML-Driven Clustering of Diabetes/Hypertension Populations” is being held Thursday, March 6, from 2-3 p.m. in the Venetian | Level 2 | Veronese 2501, at HIMSS25 in Las Vegas. Speakers include Tamer, Dr. Yolande Pengetnze, senior vice president, Clinical Leadership at PCCI; Michael Lane, senior vice president, chief quality and safety officer at Parkland Health; and Teresita Oaks, director, Community Health Programs at Parkland Health.
Email the writer: SMorse@himss.org
