Beyond Accuracy: Latent Perturbations for Cognitive-Aware Diagnosis

Diagnosing rare diseases remains a persistent challenge, often hindered by cognitive anchoring: once clinicians settle on a common diagnosis, they often discount alternative explanations, including rare conditions. To address this, we propose a cognitive-aware counterfactual reasoning framework using a Denoising Masked AutoEncoder (DMAE) to simulate what-if diagnostic scenarios that probe clinicians’ initial assumptions. Our model jointly learns (1) the true distribution of diseases and symptoms, and (2) human diagnostic behavior, revealing critical gaps between medically possible and clinically considered diagnoses. By strategically perturbing latent patient representations, it generates contrastive counterfactuals that highlight rare yet plausible diseases that cognitive bias often obscures. Unlike traditional decision-support tools, our system suggests rare diseases not because they are statistically dominant, but because they are systematically under-considered relative to the observed evidence and learned diagnostic behavior. Across four public and three private rare-disease datasets, our approach outperforms standard machine learning classifiers in detecting rare conditions while maintaining strong performance on common diagnoses. Beyond boosting accuracy, the counterfactual evidence encourages hypothesis-driven reasoning and supports clinical learning.

Rare diseases are often difficult to diagnose because their symptoms can closely resemble those of much more common illnesses. In real clinical settings, missing test results and early impressions can further narrow a doctor’s attention, making rare but plausible explanations easier to overlook. This paper presents an AI tool that helps doctors ask better “what-if” questions. It learns patterns in patient records and in typical diagnostic decisions, then points out missing symptoms or tests that could make an overlooked rare disease worth considering. The tool is designed to support, not replace, clinicians by encouraging careful second-opinion reasoning and follow-up testing. Experiments on several rare-disease datasets and expert reviews suggest that the generated examples can highlight plausible rare-disease possibilities using clear medical evidence.