When compared to population-based models, a machine learning-based model improved prediction of mortality risk for cardiac surgery patients. The team used electronic health record data and machine learning methods to demonstrate how institutions can build their own mortality prediction models. The models were tested against STS population-based models for five different surgeries, and were found to have higher accuracy, precision, and recall for each of them.
You are here:
Researchers Show That a Machine Learning Model Can Improve Mortality Risk Prediction for Cardiac Surgery Patients
Submitted by: Cameron Lind
Source: EurekAlert
Source URL: https://www.eurekalert.org/news-releases/989057
Keywords:
Author(s): Mount Sinai Health System News Release
1 Comment
More needs to be done . Individual institution based models are not the answer . Developing a model requires extensive resources and despite its accuracy the exact mechanism how ML came to conclusion remains elusive . As a PhD student evaluating AI and ML in my current project it’s too early to rely completely despite everything adds up . I strongly recommend STS and large bodies to combine resources to develop a uniform model like STS PROM , which can adjust to local demographics and needs . Nevertheless it’s great to walk into the future with Dr. Cameron at the frontier . Kudos .