Screening for posttraumatic stress disorder using verbal features in self narratives: a text mining approach.

Much evidence has shown that people's physical and mental health can be predicted by the words they use. However, such verbal information is seldom used in the screening and diagnosis process probably because the procedure to handle these words is rather difficult with traditional quantitative methods. The first challenge would be to extract robust information from diversified expression patterns, the second to transform unstructured text into a structuralized dataset. The present study developed a new textual assessment method to screen the posttraumatic stress disorder (PTSD) patients using lexical features in the self narratives with text mining techniques. Using 300 self narratives collected online, we extracted highly discriminative keywords with the Chi-square algorithm and constructed a textual assessment model to classify individuals with the presence or absence of PTSD. This resulted in a high agreement between computer and psychiatrists' diagnoses for PTSD and revealed some expressive characteristics in the writings of PTSD patients. Although the results of text analysis are not completely analogous to the results of structured interviews in PTSD diagnosis, the application of text mining is a promising addition to assessing PTSD in clinical and research settings.