Improved Detection of Mild Cognitive Impairment from Temporal Language Markers Using Subject‐invariant Representation Learning using I‐CONECT data

Using only features extracted from semi-structured conversation, we achieved a reasonalbe AUC. Our study demonstrates the additional benefits of using temporal sequences of language markers to detect Mild Cognitive Impairment. Moreover, applying temporal harmonization helps remove subject-specific components in features and further enhances cognitive detection performance.