Seasonal Flu Hospitalization Prediction: ML Model Results

Methodology & Key Results Summary

We developed a logistic regression model to classify high vs. low seasonal flu hospitalization incidence based on age group and race category. The dataset includes multiple flu seasons, and SMOTE was applied to address class imbalance before training (Rahim 2019, Huyen 2022).

• Age-Based Model: 74.3% accuracy — some misclassification of high-incidence cases.
• Race-Based Model: 76.9% accuracy — better recall for high-incidence cases.
• Confusion Matrix: Most misclassifications were false positives. Race appears more predictive than age.

ML Model Inference

Age: Age 0-17 18-49 50-64 65-74 75-84 Race: Race American Indian/Alaska Native Asian/Pacific Islander Black Hispanic/Latino White

Model Performance Metrics Table

Confusion Matrix: Model Predictions vs. Actual Data

The confusion matrix visualizes model misclassifications, showing how often high and low hospitalization rates were correctly predicted.