Using Deep Learning to Identify Patients with Cognitive Impairment in Electronic Health Records

Dementia is a neurodegenerative disorder that causes cognitive decline and\naffects more than 50 million people worldwide. Dementia is under-diagnosed by\nhealthcare professionals - only one in four people who suffer from dementia are\ndiagnosed. Even when a diagnosis is made, it may not be entered as a structured\nInternational Classification of Diseases (ICD) diagnosis code in a patient's\ncharts. Information relevant to cognitive impairment (CI) is often found within\nelectronic health records (EHR), but manual review of clinician notes by\nexperts is both time consuming and often prone to errors. Automated mining of\nthese notes presents an opportunity to label patients with cognitive impairment\nin EHR data. We developed natural language processing (NLP) tools to identify\npatients with cognitive impairment and demonstrate that linguistic context\nenhances performance for the cognitive impairment classification task. We\nfine-tuned our attention based deep learning model, which can learn from\ncomplex language structures, and substantially improved accuracy (0.93)\nrelative to a baseline NLP model (0.84). Further, we show that deep learning\nNLP can successfully identify dementia patients without dementia-related ICD\ncodes or medications.\n