Mining audio cues from PTSD interview recordings
Mining Audio Cues from PTSD Interview Recordings (MACPI) is a web based prototype system. This innovative tool is designed and developed to assist clinicians in conducting diagnostic analysis of PTSD interviews. It enables automatic transcription and phonetic indexing of the interview recordings so that the interview recordings can be accessed and analyzed with search and retrieval for systematic, quantitative, and correlational analysis. The interviews can be conducted in the office or via telephone. The recordings can also be analyzed anywhere and at any time repeatedly after the interview over the web with security-controlled access to data and the tool. Through natural language processing and machine learning algorithms, the system automatically detects and flags keywords and expressions that may identify individuals with PTSD. The prototype system will also detect and extract key prosodic features of speech (e.g., speech rate estimation, pause pattern, and variations of speech patterns). These non-lexical audio cues, combined with the lexical items in the interview, may assist clinicians in identifying certain mental or psychological states or processes and perhaps even confirm PTSD diagnosis. The MACPI prototype system has great potential to provide a systematic and objective means by which clinicians can measure and compare patients’ speech content, as well as prosodic characteristics accompanying the lexical content during PTSD interviews. This innovative tool will improve the efficiency, consistency, and accuracy of PTSD diagnosis with repeatable analysis and evidence. It will also give clinicians another method by which they can systematically monitor the treatment progress of their PTSD patients.