Predicting Post-Induction Hypotension in Diverse Surgical Populations: A Multiclass Classification Universal Model...

Area of Science:

• Anesthesiology
• Medical Informatics
• Machine Learning in Healthcare

Background:

• Post-induction hypotension (PIH) is a common complication during general anesthesia.
• Existing predictive models often focus on limited patient populations or only predict occurrence, not severity.
• Accurate prediction and severity assessment of PIH are crucial for patient safety and clinical decision-making.

Purpose of the Study:

• To develop and validate a machine learning model for predicting the occurrence and severity of post-induction hypotension (PIH).
• To address a broader patient demographic compared to previous studies.
• To enhance the clinical utility of predictive models for PIH.

Main Methods:

• Utilized electronic medical records from 71,473 adult patients undergoing non-cardiac surgery.
• Extracted features including demographics, pre-induction blood pressure, laboratory results, surgical details, and anesthetics.
• Employed machine learning techniques, including eXtreme Gradient Boosting (XGBoost), and assessed performance using accuracy, F1-scores, and AUC; severity was quantified by the integral of hypotensive periods.

Main Results:

• A multiclass classification model predicting hypotension severity achieved a superior F1-score of 0.664 compared to binary models.
• The XGBoost model demonstrated the best performance with an accuracy of 0.755 and an F1-score of 0.664.
• Models incorporating preoperative blood pressure and demographic data yielded better predictive performance.

Conclusions:

• The XGBoost machine learning technique effectively predicts PIH before surgery.
• A multiclass classification model for hypotension severity improves prediction performance and clinical utility.
• These findings support the integration of advanced machine learning models into clinical practice for better management of PIH.