Accurate and Interpretable Machine Learning for Transparent Pricing of Health Insurance Plans

• DOI: 10.1609/aaai.v35i17.17776
• Publication Date: 2022-09-08T20:27:09Z
• Authors: Rohun Kshirsagar; Li-Yen Hsu; Charles H. Greenberg; Matthew McClelland; Anushadevi Mohan; Wideet Shende; Nicolas P. Tilmans; Min Guo; Ankit Chheda; Meredith Trotter; Shonket Ray; Miguel Alvarado
• Keywords: Underwriting; Interpretability

Health insurance companies cover half of the United States population through commercial employer-sponsored health plans and pay 1.2 trillion US dollars every year to medical expenses for their members. The actuary underwriter roles at a company serve assess which risks take on how price those ensure profitability organization. While Bayesian hierarchical models are current standard in industry estimate risk, interest machine learning as way improve upon these existing methods is increasing. Lumiata, healthcare analytics company, ran study with large States. We evaluated ability predict per member month cost employer groups next renewal period, especially who will less than 95\% what an actuarial model predicts (groups "concession opportunities"). developed sequence two models, individual patient-level employer-group-level model, annual allowed amount groups, based 14 million patients. Our performed 20\% better carrier's pricing identified 84\% concession opportunities. This demonstrates application system compute accurate fair products analyzes explainable can exceed models' predictive accuracy while maintaining interpretability.