Skip to main content

STORIES

Dr. Keith Feldman Receives Frontiers Grant to Develop Computational Framework to Improve Subject Matching in Retrospective Study Designs

STORIES

Dr. Keith Feldman Receives Frontiers Grant to Develop Computational Framework to Improve Subject Matching in Retrospective Study Designs

Headshot of Keith Feldman, PhD
Keith Feldman, PhD
Assistant Professor of Pediatrics, University of Missouri-Kansas City School of Medicine
Full Biography

Keith Feldman, PhD, Doctoral Research Faculty, was recently awarded a one-year, $23,597 Frontiers BERD Trailblazer Award from the University of Kansas Medical Center Research Institute.

The project, titled “Design of a computational framework to improve the quality of subject matching in case-control study designs,” seeks to utilize state-of-the-art data representations to identify optimal control subject(s) for a given case through a broader set of clinical and temporal criteria.

“Although randomized control trials represent the gold standard in evaluating clinical hypotheses, lengthy recruitment timelines and challenges in generalizability have prompted researchers to adopt alternative, large-scale observational study designs,” Dr. Feldman noted.

“Matched case-control (MCC) studies have become particularly common, pairing subjects who exhibit an outcome of interest (cases) with similar individuals who do not (controls). This pairing offers an effective means to study associations between exposures and outcomes, particularly for rare diseases or conditions with long latency periods. Unfortunately, current matching strategies are highly manual and focus on simplistic measures of demographics or disease state. These methods are also subject to significant bias and may overlook confounding factors or temporal relationships within subject histories.”

The study team, which includes Vince Staggs, PhD, is currently working to develop an automated framework to select matched controls in a faster and more balanced manner. Their approach focuses on utilizing the entirety of a subject’s medical history to better capture similarity in previous exposures.