Use of Conversation and Acoustic Signals in Measuring Depression Severity
Use of Conversation and Acoustic Signals in Measuring Depression Severity
The purpose of this study is to test a new monitoring technology that uses the sound of a depressed person's speech to assess the severity of depression symptoms. The Vocal Social Signals Platform (VSSP) is software that analyzes the non-verbal characteristics of a person's speech. This study will test this software to see if it could be a useful measurement tool for assessing depression symptoms. Participation in this study requires coming to the research headquarters twice over a three-month period. The first visit is to determine eligibility. Throughout the study, participants will be connected to a telephone system five times, on which they will answer questions about their depression symptoms. After answering questions, their voice will be recorded using a structured speech sample that the participant will read out loud. The participant will also give an unstructured speech sample, which will involve describing a typical day or the last movie s/he saw. The voice samples will be analyzed and compared to the results of the depression symptom questionnaires.
We will conduct a 24-month study that will include the test-retest reliability, criterion validity and sensitivity to change measurements of a novel monitoring technology that might provide an objective measure to monitor depression symptoms. There will be no randomization to control and experimental groups as this is a reliability/validity study of a measurement tool. In order to evaluate this technological measure for its psychometric properties, we will recruit 60 subjects recently diagnosed with depression. We will recruit only recently diagnosed patients to maximize the likelihood of the person improving their depressive severity during our study. Even though subjects will be referred by their clinicians as depressed patients we will use CES-D or Centers for Epidemiological Studies Depression Scale to to confirm depressive symptoms in these patients. CES-D and not PHQ-9 will be the screener. We will also recruit 20 non-depressed subjects (who will also be screened by the CES-D). This is done to validate the ability of the VSSP to discriminate between non-depressed individuals and individuals presenting with depressive symptomatology. The subjects with depression will be recruited from the Boston Medical Center's (BMC) Psychiatry Clinic as well as BMC's two primary care practices. The non-depressed individuals will be recruited by posting ads in Boston Metropolitan area daily newspapers. We will administer the combined TLC-VSSP to these subjects 5 times over a three month period. This means that subjects will receive TLC-PHQ-9 during the 5-time encounters with the study. The time interval between the five evaluation sessions will be 1, 3, 4, and 4 weeks respectively. The first two data collection points will include test-retest reliability of TLC-PHQ9 and VSSP. The third, fourth and fifth data collection points will include follow-up assessments to detect change. In addition, the first and last data collections will include an evaluation of the criterion validity of VSS through an administration of the Inventory of Depressive Symptomatology (IDS) by a human professional (Validity). The first and last data collections will also be used to assess the ability of VSSP, TLC-PHQ9 and their combination to detect change in the severity of depression (as measured by the IDS). The first and last evaluations will be conducted in the research headquarters while the intermediate evaluations will occur over the telephone. The IDS interviews will be tape-recorded and conducted by a researcher. Another researcher will listen to a sample of the tapes throughout the study to assess the interviewer¿s fidelity to the IDS clinician administration protocol for each of the 30 items and score the questionnaire independently of the interviewer. In addition, subjects will provide both ¿structured¿ and ¿unstructured¿ voice samples in random order at each assessment point. Baseline Visit 2 Visit 3 Visit 4 Visit 5 (telephone) (telephone) (telephone) Week 0 Week 1 Week 4 Week 8 Week 12 We designed the study with 5 data collection points in order to observe change in symptom severity levels at intermediate points before the final assessment at visit 5. We expect the majority of patients to be in some type of treatment, either receiving an antidepressant medication regimen or some form of therapy or both (we will not be concerned with the specific type of treatment). It is reported that approximately 60% of patients on antidepressant medications respond to treatment. Thus, we anticipate certain improvement in depressive symptoms among a substantial proportion of subjects who are taking antidepressants. We are also aware that some of the subjects who are being treated by antidepressants may experience side-effects. We estimate that approximately 15% of patients on antidepressants might discontinue treatment due to significant side-effects and lack of efficacy. In addition, some subjects may not be completely adherent to their medication and/or therapy regimens for a variety of reasons. These factors will make it possible to explore changes in symptom severity in two different directions of improvement and deterioration. We will collect information on these events (being on treatment, stopping treatment, non-adherence or under adherence to mediation taking, missing therapy sessions, etc.) by providing the subjects with a diary and asking them to write all treatment related events in that diary. The diary will include specific questions about possible events in treatment regimens.
Inclusion Criteria:
Exclusion Criteria:
Not having access to a telephone at home or on the person (cell phone) or at work.
Planning to leave Boston during the study period.
Having a long history of smoking, which is defined as 1 pack per day for 10 years or more.
Meeting the criteria for alcohol abuse.
Having one of the following mental illnesses based on attached instruments: Social Anxiety, Bipolar Disorder (BD) or Borderline Personality Disorder (BPD).
Having a life threatening and/or chronic health condition such as organic brain disease, e.g., Alzheimers' Disease and other dementias, cancer, HIV or other autoimmune diseases, diabetes, heart disease, renal failure, etc.
Having a physical condition, such as one of the following, that could affect the acoustic characteristics of speech or a condition that can alter vocal production by affecting the mouth, larynx, trachea, lungs, thorax, nose, nasal passages and sinuses, ears and hearing: