| dc.description.abstract | Diagnosing mental disorders is complex due to the genetic, environ-mental and psychological contributors and the individual risk factors.Language markers for mental disorders can help diagnose a person.The differences in the usage of language between groups are called language markers. Research thus far on language markers and the associated mental disorders has been done mainly with the Linguistic Inquiry and Word Count program (LIWC). We found that research thus far mainly focused on LIWC. So, first task was to investigate several traditional and deep learning models and spaCy, fastText and RobBERT were chosen. Next, the prediction performances of LIWC, spaCy, fastText and RobBERT were compared. The best performing technique to find out if a person has a mental disorder is LIWC in combination with the classification algorithm random forest which reached an accuracy-score of 0.952 and a Cohen's kappa of 0.889. spaCy in combination with random forest predicted best which mental disorder a person has with an accuracy-score of 0.429 and a Cohen's kappa of 0.304. Furthermore, several language markers were found. With these markers, the LIWC-decision tree and an interview transcription, there could be determined if a person has a mental disorder or not. | |
