Bayesian Hierarchical Rule Modeling for Predicting Medical Conditions

By Tyler H. McCormick, Cynthia Rudin, David Madigan. 2012

46 citations recorded by the source corpus.

We propose a statistical modeling technique, called the Hierarchical Association Rule Model (HARM), that predicts a patient's possible future medical conditions given the patient's current and past history of reported conditions. The core of our technique is a Bayesian hierarchical model for selecting predictive association rules (such as "condition 1 and condition 2 $\rightarrow$ condition 3") from a large set of candidate rules. Because this method "borrows strength" using the conditions of many similar patients, it is able to provide predictions specialized to any given patient, even when little information about the patient's history of conditions is available.

Subfields: Computational Theory and Mathematics, Information Systems, Molecular Biology, 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 Computational Theory and Mathematics

1. Cynthia Rudin
2. Margo I. Seltzer
3. Nicholas Larus-Stone
4. Daniel Alabi
5. Elaine Angelino
6. Benjamin Letham
7. David Madigan
8. Tong Wang

In Information Systems

1. Cynthia Rudin
2. Margo I. Seltzer
3. Benjamin Letham
4. Nicholas Larus-Stone
5. Daniel Alabi
6. Elaine Angelino
7. David Madigan
8. Tyler H. McCormick

In Molecular Biology

1. Cynthia Rudin
2. Benjamin Letham
3. David Madigan
4. Tyler H. McCormick

In Probabilistic & Bayesian Methods

1. Cynthia Rudin
2. Margo I. Seltzer
3. Benjamin Letham
4. Nicholas Larus-Stone
5. Daniel Alabi
6. Elaine Angelino
7. David Madigan
8. Tyler H. McCormick

In Medical & Health AI

1. Cynthia Rudin
2. Margo I. Seltzer
3. Benjamin Letham
4. Nicholas Larus-Stone
5. David Madigan
6. Daniel Alabi
7. Elaine Angelino
8. Tyler H. McCormick

Citations in the corpus

• Interpretable classifiers using rules and Bayesian analysis: Building a better stroke prediction model (2015, background): “…associative classifiers are formed by averaging several rules together, or having the rules each vote on the label and then combining the votes, but the resulting classifier is not generally interpretable [Li, Han and Pei (2001), Yin and Han (2003), Friedman and Popescu (2008), Meinshausen (2010)].”
• Learning Certifiably Optimal Rule Lists (2017, methodology): “Some methods allow for interpretations of single rules, without constructing rule lists (McCormick et al., 2012).”
• Scalable Bayesian Rule Lists (2016, neutral): “…2005; Marchand & Sokolova, 2005; Rudin, Letham, & Madigan, 2013; Goessling & Kang, 2015), associative classification (Vanhoof & Depaire, 2010; Liu et al., 1998; Li, Han, & Pei, 2001; Yin & Han, 2003) and its Bayesian counterparts (McCormick, Rudin, & Madigan, 2012) all fall into this category.”
• Building Interpretable Classifiers with Rules using Bayesian Analysis (2012, methodology): “This work is also related to the Hierarchical Association Rule Model (HARM) presented recently by [30].”
• Machine learning for science and society (2013, background): “…healthcare (Letham et al. 2013b; McCormick et al. 2012), information retrieval (Letham et al. 2013a), interpretable models (Letham et al. 2013b; McCormick et al. 2012; Ustun et al. 2013), robotic space exploration (Castano et al. 2007; Wagstaff and Born-stein 2009; Wagstaff et al. 2013b), and…”
• Learning theory analysis for association rules and sequential event prediction (2013, methodology): “In practice, the choice of K can be determined in several ways: K can be manually determined (for instance by the customer), it can be set using side information as considered by McCormick et al. (2012), or it can be set via cross-validation on an extra hold-out set.”