Controlling for Unobserved Confounding with Large Language Model Classification of Patient Smoking Status
This addresses the issue of unobserved confounding for researchers in medical causal inference, but it is incremental as it extends prior methods to more complex classifiers and categorical variables.
The paper tackled the problem of unobserved confounding in causal inference from observational medical data by using a large language model to predict patient smoking status from clinical notes, then applying measurement error correction to estimate the causal effect of transthoracic echocardiography on mortality in the MIMIC dataset, achieving an unbiased estimate.
Causal understanding is a fundamental goal of evidence-based medicine. When randomization is impossible, causal inference methods allow the estimation of treatment effects from retrospective analysis of observational data. However, such analyses rely on a number of assumptions, often including that of no unobserved confounding. In many practical settings, this assumption is violated when important variables are not explicitly measured in the clinical record. Prior work has proposed to address unobserved confounding with machine learning by imputing unobserved variables and then correcting for the classifier's mismeasurement. When such a classifier can be trained and the necessary assumptions are met, this method can recover an unbiased estimate of a causal effect. However, such work has been limited to synthetic data, simple classifiers, and binary variables. This paper extends this methodology by using a large language model trained on clinical notes to predict patients' smoking status, which would otherwise be an unobserved confounder. We then apply a measurement error correction on the categorical predicted smoking status to estimate the causal effect of transthoracic echocardiography on mortality in the MIMIC dataset.