Fernando Bação and Gonçalo Almeida, from NOVA IMS, have developed a new AI technique – Counterfactual SMOTE – that significantly improves the detection of rare diseases by reducing false negatives by up to 34%.

Published in the journal Data Science and Management, the method tackles the challenge of data imbalance in critical scenarios. While designed for healthcare, it also holds potential for fraud detection, churn prediction, and fault identification across industries.