Prediction of in-Hospital Mortality for Icu Patients with Heart Failure

EasyChair Preprint 13899

Abstract

Heart failure, a terminal stage of cardiac conditions, affects millions of people worldwide. It greatly reduces life quality and is associated with high mortality rates. Despite extensive research, the statistical connection between heart failure and mortality rates for ICU patients, remains underexplored, indicating the need for improved prediction models.

This study used data from MIMIC-III and identified 1,177 patients over 18 years old using ICD-9 codes. Preprocessing consisted of handling missing data, deleting duplicates, treating skewness, and oversampling to alleviate data imbalance. 18 features were selected within a LightGBM model by checking the Variance Inflation Factor (VIF) values, LASSO Regression, and univariate analysis. The final output of the LASSO Logistic Regression model had the highest test AUC-ROC [0.8766(0.8065 - 0.9429)] and accuracy [0.7291] compared with the other baseline models, including Logistic Regression, Random Forest, LightGBM, Support Vector Machine (SVM), Decision Tree. All models demonstrated good calibration with relatively low Brier scores, highlighting their reliability in predicting in-hospital mortality.

When compared with the previous best-achieved results in both literature and baseline models, great advancements have been made in terms of predicting the death of heart failure ICU patients. Three major achievements have led to these breakthroughs: new and successful experience and literature search, which could utilize for the selection of key features, application of the method of a preprocessing approach to missing values via the system of imputation strategies, and feature selection. In the end, fasting the Grid-Search was the key to a near-perfect predictive model. These methods brought substantial positive changes, thus highlighting the possibility of increased predictive accuracy of in-hospital mortality in ICU patients with heart failure.

Keyphrases: MIMIC-III, heart failure, in-hospital mortality, machine learning

@booklet{EasyChair:13899,
  author    = {Jiahong Zhang and Hexin Li and Negin Ashrafi and Zhijiang Yu and Greg Placencia and Maryam Pishgar},
  title     = {Prediction of in-Hospital Mortality for Icu Patients with Heart Failure},
  howpublished = {EasyChair Preprint 13899},
  year      = {EasyChair, 2024}}