Medical Biases and Lexical Diagnoses: A Comparative Lexical Analysis of Psychological and Psychiatric Patient Records

Abstract

As levels of mental health related diagnoses are on the rise across the United States, the nationwide discourse on mental health is only becoming increasingly more pressing. Medical professionals conduct psychiatric and psychological evaluations to best diagnose and treat their patients, and their comments are mostly concentrated in the text of patient records. Due to the increasing urgency to investigate mental health as depression rates soar across the country, it is crucial to examine the lexical evidence of diagnoses most commonly associated with particular racial, gender, marital, and age demographics. It is also critical to understand how medical professional bias or judgment presents itself in patient records. To better understand the mental health crisis from a data-driven perspective, I conducted a sentiment analysis of anonymized psychological and psychiatric patient records to understand which mental illnesses different demographics are most commonly afflicted with, and then identified positive and negative biases associated with these demographics. Ultimately, this investigation aims to illuminate the vulnerability of certain demographics to particular mental illnesses and to ensure that medical transcriptions remain as unbiased as possible.