Predicting cognitive impairments with a mobile application

Abstract

Assessment of cognitive impairments is of social and clinical importance for vulnerable populations, such as elderly, athletes and soldiers, who are prone to falling victim to cognitive impairments. This paper presents ongoing work for developing an application that predicts the neurological state of users with the state-of-the art performance through analyzing the structural complexity of users’ utterances. We present a novel method that estimates the neurological state of users with Pearson correlation of 0.66 with respect to the Mini-mental state exam score. Unlike previous work, our method does not depend on assumptions of relating linguistics representations to human language-processing capabilities, but discovers the discriminative patterns automatically.