Dr. Pappas is a junior faculty member in the Medicine Institute at the Cleveland Clinic, with appointments in the Center for Value-Based Care Research and the Department of Hospital Medicine (Cleveland, OH). This career development award aims to provide training in propensity methods, deep learning techniques, and pilot intervention development, ultimately seeking to identify personalized approaches to cardiac stress testing before surgery. Noncardiac surgery carries risk of mortality and morbidity, and cardiac complications account for the largest share of perioperative mortality. Meanwhile, current approaches are expensive and may not effectively reduce cardiac risk. This proposal uses machine learning techniques to define the value of information provided by each different kind of stress test and the expected benefit of different therapeutic interventions through which preoperative risks might be modified. It then seeks to identify the most helpful cardiac stress test, if any. In this career development award, Dr. Pappas proposes three phases of investigation, and in so doing will acquire new skills critical to achieving his goal of becoming an expert in perioperative risk mitigation.
In Aim 1, Dr. Pappas will use propensity matching techniques to evaluate prior associations between preoperative stress testing and improved postoperative mortality, when including rich clinical data not available to previous large studies.
In Aim 2, he will use machine learning techniques to estimate the value of information provided by each modality of stress testing, and the impact on the risk of each event from each intervention.
In Aim 3, Dr. Pappas will pilot an intervention presenting personalized estimates to physicians in the preoperative clinic. In addition to advanced training through formal coursework, this career development award is supported by an extraordinary mentorship team, including internationally-recognized experts in perioperative outcomes research, cardiovascular disease, use of observational healthcare data, and machine learning. The combination of formal training and mentored research outlined in this application is designed to ensure that Dr. Pappas will emerge from this award as an independent investigator and expert in personalized perioperative decision-making.
Many patients die soon after surgery, and many of those deaths are from heart problems. Physicians have tried to reduce those risks in many ways, but our current approach is expensive, leads to unhelpful testing, and probably doesn't do a very good job reducing the risk of death. This proposal uses a form of machine learning and a large dataset to try to find more personalized, more helpful, and less expensive strategies to try to reduce the risks of surgery, and then tests whether we can present personalized estimates to doctors that they could use when thinking about the risks of surgery.