A new article in the journal Physiological Measurement describes the efforts of Intelligent Critical Care (IC3) directors and collaborators to develop the MySurgeryRisk platform, an artificial intelligence and machine learning model designed to predict the likelihood of postoperative complications.
Written by IC3 co-directors Azra Bihorac and Parisa Rashidi; associate directors for research Tyler Loftus and Tezcan Ozrazgat-Baslanti; assistant director for AI research Benjamin Shickel; and IC3 collaborators Jeremy Balch, Patrick Tighe, and Philip Efron, “Building an automated, machine learning-enabled platform for predicting post-operative complications” discusses the group’s foundational work in predictive medical AI and future considerations for implementing MySurgeryRisk.
Previous work on machine learning in clinical medicine, such as Dr. Tighe’s work with machine learning classifiers for predictions of acute pain service consultations (2012) and Feng et al.’s risk assessment model architecture (2017), laid the groundwork for MySurgeryRisk. Other IC3 projects, particularly I2CU and DeepSOFA, enhanced MySurgeryRisk‘s framework and increased awareness of the need for explainable models. Now, IC3 is looking to ensure the application is externally validated and continually updated as implementation efforts begin.
Read the Physiological Measurement article here.