Logistic regression (LR) is widely used in clinical prediction because it is simple to deploy and easy to interpret. Nevertheless, being a linear model, LR has limited expressive capability and often unsatisfactory performance. Generalized additive models (GAMs) extend the linear model with transformations of input features, though feature interaction is not allowed for all GAM variants. In this paper, we propose a factored generalized additive model (F-GAM) to preserve the model interpretability for targeted features while allowing a rich model for interaction with features fixed within the individual. We evaluate F-GAM on two tasks, postoperative acute kidney injury and acute respiratory failure from a single center database and find superior model performance of F-GAM in terms of AUPRC and AUROC compared to several other GAM implementations, random forests, support vector machine, and a deep neural network. We find that the model interpretability is good with results with high face validity.
Learning Objective: After participating in this session, the learner should be better able to:
Define what it means for a prediction model to be accountable and actionable.
Identify machine learning techniques that yield accountable and actionable models.
Understand the architecture and characteristics of this novel Factored Generalized Additive Model (F-GAM) technique
Zhicheng Cui, Washington Univ in St Louis
Bradley Fritz (Presenter)
Washington University in St Louis
Christopher King, Washington University in St Louis
Michael Avidan, Washington University in St Louis
Yixin Chen, Washington Univ in St Louis