In this study, the researchers explored the use of language markers to diagnose mental disorders, a complex process involving genetic, environmental, and psychological factors. While previous research has primarily relied on the Linguistic Inquiry and Word Count (LIWC) program, the team employed a range of Natural Language Processing (NLP) techniques to analyze Dutch psychiatric interview transcriptions. Their primary objective was to predict whether a patient had been diagnosed with a mental disorder, and if so, the specific type of disorder.
Using LIWC in combination with the random forest classification algorithm, the team achieved an accuracy of 0.952 and a Cohen's kappa of 0.889 in predicting whether a person had a mental disorder or not. Additionally, SpaCy in combination with random forest performed best in predicting the specific mental disorder a patient had been diagnosed with, with an accuracy of 0.429 and a Cohen's kappa of 0.304. These findings could contribute to the development of more accurate and efficient methods for diagnosing mental disorders using language markers.