Fair Cognitive Impairment Detection Through Unlearning
For researchers and practitioners in clinical AI, this addresses the problem of demographic bias in automated MCI screening, offering a method to improve fairness without sacrificing performance.
The paper tackles fairness in cognitive impairment detection from speech, proposing a multimodal framework with demographic unlearning that reduces performance gaps across subgroups while improving classification accuracy over state-of-the-art baselines on two benchmarks.
Mild Cognitive Impairment (MCI) is a medical condition characterized by a noticeable decline in memory, language, or thinking abilities. MCI detection from spontaneous speech is promising for scalable screening. However, learned models often exploit demographic cues correlated with labels, resulting in a large performance gap across subgroups. We present a multimodal framework that combines (i) cross-model fusion between modalities (speech, text, and image), and (ii) unlearning using gradient reversal that discourages the shared embedding from encoding task-irrelevant demographic attributes. Evaluated on the multilingual benchmarks TAUKADIAL and PREPARE, our method outperforms the state-of-the-art multilingual and multimodal baseline in MCI classification while substantially reducing the performance gap across patient subgroups (sex and language). We further analyze transfer across datasets, showing that demographic unlearning helps learn more robust representations for MCI detection.