POTTER-ICU: An Artificial-Intelligence Interpretable Tool To Predict Need For ICU Admission After Emergency Surgery
Anthony Gebran1,4, Annita Vapsi2, Lydia R. Maurer1,4, Mohamad El Moheb1,4, Leon Naar1,4, Sumiran S. Thakur2, Dania Daye3,4, George C. Velmahos1, Dimitris Bertsimas2, Haytham M.A. Kaafarani1,4
1Division of Trauma, Emergency Surgery, & Surgical Critical Care, Department of Surgery, Massachusetts General Hospital, Boston, MA; 2Massachusetts Institute of Technology, Cambridge, MA; 3Division of Interventional Radiology, Massachusetts General Hospital, Boston, MA; 4Center for Outcomes & Patient Safety in Surgery (COMPASS), Massachusetts General Hospital, Boston, MA
Background: Delays in admitting high-risk emergency surgery (ES) patients to the ICU result in worse outcomes and increased healthcare costs. We aimed to use interpretable artificial intelligence (AI) technology to create a pre-operative predictor for post-operative ICU need in ES patients.
Methods: An interpretable AI technology called Optimal Classification Trees (OCTs) was leveraged in an 80:20 train:test split of adult ES patients in the 2007-2017 ACS-NSQIP database. Demographics, comorbidities, and laboratory values were used to develop and validate OCT algorithms to predict the need for post-operative ICU admission. The latter was defined as post-operative death or the development of one or more post-operative complication warranting critical care (e.g. unplanned intubation, ventilator requirement ?48 h, cardiac arrest requiring CPR, and septic shock). C-statistics were used to measure performance.
Results: A total of 464,861 patients were included. The mean age was 55 years, 48% were male, and 11% developed severe post-operative complications warranting critical care. Comprehensive OCT algorithms were derived [Figure 1], and the Predictive OpTimal Trees in Emergency Surgery Risk ICU (POTTER-ICU) application was created. POTTER-ICU had excellent discrimination for predicting the need for ICU admission (c-statistics: 0.88).
Conclusion: We have thus developed POTTER-ICU as an accurate, AI-based tool for predicting severe complications warranting ICU admission after ES. POTTER-ICU can prove useful to appropriately triage ICU patients and to decrease failure to rescue in ES patients.
Back to 2021 Program