Non-parametric recurrent events analysis with BART and an application to the hospital admissions of patients with diabetes

• Published in: Biostatistics (Oxford, England) Vol. 21; no. 1; pp. 69 - 85
• Main Authors: Sparapani, Rodney A, Rein, Lisa E, Tarima, Sergey S, Jackson, Tourette A, Meurer, John R
• Format: Journal Article
• Language: English
• Published: England Oxford University Press 01.01.2020
• Subjects: Adult; Aged; Aged, 80 and over; Biostatistics - methods; Computer Simulation; Diabetes Mellitus - therapy; Female; Humans; Male; Middle Aged; Models, Statistical; Outcome and Process Assessment, Health Care - methods; Patient Admission - statistics & numerical data; Young Adult; Bayesian Additive Regression Trees; Electronic health records (EHR); Non-proportional; Cumulative intensity; Machine learning; Variable selection
• ISSN: 1465-4644; 1468-4357; 1468-4357
• DOI: 10.1093/biostatistics/kxy032

Much of survival analysis is concerned with absorbing events, i.e., subjects can only experience a single event such as mortality. This article is focused on non-absorbing or recurrent events, i.e., subjects are capable of experiencing multiple events. Recurrent events have been studied by many; however, most rely on the restrictive assumptions of linearity and proportionality. We propose a new method for analyzing recurrent events with Bayesian Additive Regression Trees (BART) avoiding such restrictive assumptions. We explore this new method via a motivating example of hospital admissions for diabetes patients and simulated data sets.