Quantifying Symptom Causality in Clinical Decision Making: An Exploration Using CausaLM
Read the original on arXiv Machine Learning →The paper investigates the causal impact of the symptom "chest pain" on medical diagnosis predictions, moving beyond traditional correlation-based methods. Using the CausaLM framework, the authors generate counterfactual text representations that effectively remove the symptom, allowing for a principled estimation of its causal effect on disease predictions. They apply Textual Representation-based Average Treatment Effect (TReATE) to quantify how the presence or absence of chest pain shapes diagnostic outcomes, and compare these results to correlation-based baselines such as CONEXP, providing deeper insight into the decision-making behavior of clinical NLP models.
Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv Machine Learning.