Psychiatry Transcript Annotation: Process Study and Improvements

Background: The demand for psychiatry is increasing each year. Limited research has been performed to improve psychiatrist work experience and reduce daily workload using computational methods. There currently no validated tool or procedure the mental health transcript annotation process generating “gold-standard” data. purpose of this paper was determine transcripts how it can be improved acquire more reliable results considering human factors elements. Method: Three expert clinicians were recruited in study evaluate transcripts. asked fully annotate two An additional five subjects randomly (aged between 20-40) pilot study, which divided into phases, phase 1 (annotation without training) 2 with Kappa statistics used measure inter-rater reliability accuracy subjects. Results: 0.26 (CI 0.19 0.33) 0.49 0.42 0.57), respectively. In testing mean higher training (k= 0.35 0.052 0.625)) than 0.29 0.128 0.451)). After training, percentage among significantly A (p=0.04) B (p=0.10). Conclusion: This focused on understanding transcripts, will applied machine learning models. Through exploratory found appropriate categorical labels that should included annotation, importance Contributions case help psychiatric researchers implementing recommended data collection develop a accurate artificial intelligence model fully- semi-automated annotation.