The aim of this project is the preliminary evaluation of the psychometric properties (test-retest reliability, criterion validity and sensitivity to change) of a novel depression monitoring technology that uses acoustic measures of the patient's speech. The Vocal Social Signals Platform (VSSP) is software that performs analysis on the prosodic (non-verbal) characteristics of speech. The acoustic attributes of a person's speech once analyzed are referred to as the Vocal Social Signals (VSS). Since depressed individuals may demonstrate changes in their speech prosody, Vocal Social Signals could present a useful measurement tool for assessing depression symptoms. In the proposed study, we will employ VSSP to analyze the speech of depressed patients who converse with an automated telephone-based conversational system (Telephone-Linked Communications {TLC}). The psychometric properties of the resulting Vocal Social Signals (VSS) will be the subject of our proposed study. In this study, the Vocal Social Signals will be compared to two different reliable and valid instruments. One instrument, (PHQ-9), is a brief, commonly used questionnaire with good reliability and validity. Since the PHQ- 9 is a brief, validated and reliable questionnaire, it has been adapted to different modalities such as over the telephone and in automated formats (e.g., computer telephony and the Web). The second instrument, the Inventory of Depressive Symptomatology (IDS), is administered in a semi-structured interview by a health professional. The IDS is a relatively recent 30-item measure of depression symptom severity that includes all the symptom domains needed to diagnose a DSM-IV major depressive disorder and it also assesses for symptom remission. The IDS has scaling items to detect different levels of symptoms. We will evaluate the test-retest reliability of VSS, its criterion validity and its ability to detect change by comparing it to the IDS and to PHQ-9 among 80 individuals (60 depressed and 20 non-depressed). The VSS will be compared to the PHQ- 9 in 5 data collection points while it will be compared to IDS in two data collection points. During all five data collection points, however, the participants will provide a recorded voice sample as they will converse with the TLC system which will administer the PHQ-9 (TLC-PHQ9). The importance of this study is two-fold: 1) If the VSS alone is demonstrated to be reliable and valid, then it would be possible for researchers and clinicians to utilize the VSSP software to detect depression and level of depression symptom severity by recording a patient's speech. 2) However, if the combination of the VSS and PHQ-9 is better, then an ideal methodology for assessing depression level might be an automated telephonic administration of PHQ-9 combined with a VSS analysis of the patients'speech (e.g., their responses to the PHQ-9 questions). Altogether, the evaluation of VSS in this study offers the potential of making a significant improvement in how depression severity level is measured in research and possibly in clinical practice.
This project is relevant to public health as it has the ability to provide clinicians with an objective measure to easily and unobtrusively assess symptom severity in depressed patients. This in turn will increase and improve the efficiency and efficacy of depression treatment.
