Interpretable classifiers using rules and Bayesian analysis: Building a better stroke prediction model

By Benjamin Letham, Cynthia Rudin, Tyler H. McCormick, David Madigan. 2015

807 citations recorded by the source corpus.

We aim to produce predictive models that are not only accurate, but are also interpretable to human experts. Our models are decision lists, which consist of a series of if...then... statements (e.g., if high blood pressure, then stroke) that discretize a high-dimensional, multivariate feature space into a series of simple, readily interpretable decision statements. We introduce a generative model called Bayesian Rule Lists that yields a posterior distribution over possible decision lists. It employs a novel prior structure to encourage sparsity. Our experiments show that Bayesian Rule Lists has predictive accuracy on par with the current top algorithms for prediction in machine learning. Our method is motivated by recent developments in personalized medicine, and can be used to produce highly accurate and interpretable medical scoring systems. We demonstrate this by producing an alternative to the CHADS$_2$ score, actively used in clinical practice for estimating the risk of stroke in patients that have atrial fibrillation. Our model is as interpretable as CHADS$_2$, but more accurate.

Subfields: Artificial Intelligence, Computational Theory and Mathematics, Safety, Robustness & Interpretability, Probabilistic & Bayesian Methods, Medical & Health AI

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Computational Theory and Mathematics

In Safety, Robustness & Interpretability

In Probabilistic & Bayesian Methods

In Medical & Health AI

Citations in the corpus

Rationalizing Neural Predictions (2016, background): “Developing sparse interpretable models holds considerably interest in the broader research community(Letham et al., 2015; Kim et al., 2015).”
Interpretable Machine Learning: Fundamental Principles and 10 Grand Challenges (2021, neutral): “Sometimes decision lists or decision sets are optimized by sampling (Letham et al., 2015; Yang et al., 2017; Wang and Rudin, 2015a), providing a Bayesian interpretation.”
A Survey on Explainable Artificial Intelligence (XAI): Toward Medical XAI (2019, background): “Likewise, [60] creates a model that provides such statements for stroke prediction.”
Fairwashing: the risk of rationalization (2019, background): “A rule list (Rivest, 1987; Letham et al., 2015; Angelino et al., 2018) d = (dp, δp, q0,K) of length K ≥ 0 is a (K + 1)−tuple consisting of K distinct association rules rk = pk → qk, in which pk ∈ dp is the antecedent of the association rule and qk ∈ δp its corresponding consequent, followed by a def”
Peeking Inside the Black-Box: A Survey on Explainable Artificial Intelligence (XAI) (2018, background): “[66] proposed a model called Bayesian Rule Lists (BRL) based on decision tree, the authors claimed that preliminary interpretable models provide a concise and convincing capabilities to gain domain experts trust.”
Learning Optimal Decision Trees with SAT (2018, background): “…is illustrated by a range of work in the recent past ([Li et al. , 2018; Lakkaraju et al. , 2017; Angelino et al. , 2017; Lakkaraju et al. , 2016; Letham et al. , 2015; Lou et al. , 2012] among others), by a number of recent events dedicated to the topic [IJ-CAI XAI Workshop, 2017; ICML WHI…”
Anchors: High-Precision Model-Agnostic Explanations (2018, background): “A number of approaches construct globally interpretable models, many based on rules (Lakkaraju, Bach, and Leskovec 2016; Letham et al. 2015; Wang and Rudin 2015; Wang et al. 2015).”
Interpreting Deep Classifier by Visual Distillation of Dark Knowledge (2018, background): “In contrast to deep neural networks, an alternative is to restrict the models to belong to a family that is inherently easy to interpret (Caruana et al., 2015; Doshi-Velez et al., 2014; Letham et al., 2015).”
What do we need to build explainable AI systems for the medical domain? (2017, methodology): “These can include Bayesian Rule Lists (Letham et al., 2015), and in order to learn richer, more conceptual and generative models, techniques such as Bayesian Program Learning (Lake et al., 2015), learning models of causal relationships (Maier et al., 2010, 2013; Aalen et al., 2016), and stochastic g”
Interpretable & Explorable Approximations of Black Box Models (2017, methodology): “We benchmark the performance of our framework against the following baselines: 1) Locally interpretable model agnostic explanations (LIME) [8] 2) Interpretable Decision Sets (IDS) [4] 3) Bayesian Decision Lists (BDL) [6].”
Interpretability via Model Extraction (2017, methodology): “In this paper, we take T to be a decision tree, which has been established as highly interpretable [3, 20, 23].”
Methods for interpreting and understanding deep neural networks (2017, methodology): “Furthermore, while some methods aim at reaching a comprehensive functional understanding of the model [32,41,8], we will focus here on interpreting the outputs of a DNN, and explaining individual predictions.”
Model-Agnostic Interpretability of Machine Learning (2016, methodology): “, 2014), a set of rules (Letham et al., 2015) or line graphs (Caruana et al.”
The Bayesian Case Model: A Generative Approach for Case-Based Reasoning and Prototype Classification (2014, neutral): “, [12, 32, 13, 23, 15]); and prototype- or case-based classifiers (e.”
Learning Optimal and Fair Decision Trees for Non-Discriminative Decision-Making (2019, methodology): “Finally our work relates to interpretable ML, including works on decision rules (Wang et al. 2017; Letham et al. 2015), decision sets (Lakkaraju, Bach, and Jure 2016), and generalized additive models (Lou et al. 2013a).”