![An illustration of a brain's circuitry hovers over a tablet device.](/sites/default/files/2022-11/machine-learning_0.png)
David E. Hamilton, MD, presented the poster “Merging Machine Learning and Patient Preference: A Contemporary, Comprehensive, Patient-Centered Tool for Risk Prediction Prior to Percutaneous Coronary Intervention” during the American Heart Association Scientific Sessions 2022 held November 5th – 7th in Chicago.
Dr. Hamilton and his colleagues hypothesized that using machine learning models would allow predictions of common post-PCI complications using pre-procedural factors.
The team engaged a panel of patients who underwent PCI procedures at BMC2 hospitals to rank the importance of common PCI complications. A separate group of 66 adults underwent a semiquantitative survey to assess a preferred list of outcomes and model display.
Using data, the team identified percentages of:
- acute kidney injury
- new need for dialysis
- major bleeding
- transfusion
- overall rate of mortality
They then designed an XGBoost machine learning model that accurately predicts individualized post-PCI outcomes.
Patients, including BMC2 PCI Patient Advisory Council members, provided feedback to help create a patient-centered tool. The tool displays risks to patients and providers, allowing them access to enhanced risk prediction before PCI. This tool could help inform shared decision-making discussions and treatment selection.
The study co-authors are Jeremy Albright, PhD; Milan Seth, MS; Devraj Sukul, MD, MSc; and Hitinder S. Gurm, MD. You can view the poster on bmc2.org.