Genetic algorithms for predictive analytics

Summary

We examine the ability of a genetic algorithm (GA) to learn predictive models for classification problems in the healthcare domain. Specifically, we apply a GA to the problem of predicting the likelihood that a physical therapist (PT) will receive annual Medicare payments above or below the industry median based on the PT's practice parameters. The classification function is represented as a weighted sum of the input parameters which produces a value above or below zero indicating above or below median, respectively. We find that this approach, not only achieves 93% accuracy in classification, but also provides information on which variables are most relevant to making a classification.

Participants

• Dr. Annie S. Wu
• Dr. Xinliang Liu
• Dr. Ryan McMahan
• Reamonn Norat
• Esteban Segarra
• Stephen Maldonado
• Blake Oakley

Recent publications

• Reamonn Norat, Annie S. Wu, and Xinliang Liu (2023). Genetic algorithms with self-adaptation for predictive classification of Medicare standardized payments for physical therapists. Expert Systems with Applications, 218, 119529. [pdf] [Info] [bibtex]

• Esteban Segarra Martinez, Stephen V. Maldonado, Annie S. Wu, Ryan P. McMahan, Xinliang Liu, and Blake Oakley (2022). Effects of imputation strategy on genetic algorithms and neural networks on a binary classification problem. In the Proceedings of the Genetic and Evolutionary Computation Conference, pp. 1272-1280. July 9-13, 2022. [pdf] [bibtex]

• Annie S. Wu, Xinliang Liu, and Reamonn Norat (2019). A genetic algorithm approach to predictive modeling of Medicare payments to physical therapists. In the Proceedings of the 32nd International Florida Artificial Intelligence Research Society (FLAIRS) Conference, pp. 311-316. [pdf] [bibtex]