A Logistic Regression Model for Intraoperative Hypotension Prediction

This paper presents a study on using a logistic regression model for the prediction of intraoperative hypotension, a common but critical situation occurring during surgeries where the patient's blood pressure drops significantly.

This paper presents a study on using a logistic regression model for the prediction of intraoperative hypotension, a common but critical situation occurring during surgeries where the patient's blood pressure drops significantly. The model was developed using the Python scikit-learn library and was trained on 80% of the available data, while the remaining 20% was used for testing and evaluation. After preprocessing the data, the logistic regression model was trained and then predicted the occurrence of hypotension. The model demonstrated high accuracy, with 97.92% for predictions one minute ahead, but a slight decrease in accuracy was observed as the prediction time increased. When personalized for this type of surgery, the model's performance improved for only a few surgeries, otherwise not highlighting the efficacy of an individualized approach. However, the model faced limitations in predicting hypotensive events beyond a few minutes in advance, necessitating future exploration of more advanced machine learning models. While the logistic regression model's performance was promising, other machine learning models like gradient boosting and random forest algorithms also show high potential. The research underscores the potential of machine learning models in enhancing patient safety and optimizing surgery outcomes through accurate prediction of intraoperative hypotension.This paper presents a study on using a logistic regression model for the prediction of intraoperative hypotension, a common but critical situation occurring during surgeries where the patient's blood pressure drops significantly. The model demonstrated high accuracy, with 97.92% for predictions one minute ahead, but a slight decrease in accuracy was observed as the prediction time increased. When personalized for type of surgery, the model's performance improved for only certain surgeries, not showing a significant improvement overall. While the logistic regression model's performance was promising, other machine learning models like gradient boosting and random forest algorithms also show high potential.