Friedman Risk Score: A Custom Primary Care Score Predicting Consumption

Poster Number: P32
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Population Health, Machine Learning, Healthcare Quality
Primary Track: Applications

We extracted data from >100,000 University of California San Diego Health primary care patients to predict top healthcare utilization with six multiclass prediction algorithms. XGBoost outperformed other models, achieving the highest area under the receiver operating characteristic curve of >0.80 for patients with very high healthcare utilization (top 5%). Our findings demonstrate the potential of advanced risk score models to optimize healthcare resource allocation and improve care efficiency.

Speaker(s):
Yufei Yu, BS
University of California San Diego

Author(s):
Yufei Yu, BS - University of California San Diego; Alson Mo, BA - UC San Diego Health; Tsung-Ting Kuo, PhD - University of California San Diego; Amy Sitapati, MD - UC San Diego;