OPTIMI: Early Prediction and Prevention of Depression

Institute for Response-Genetics, Departement of Psychiatry (KPPP)

Psychiatric Hospital, University of Zurich


Monitoring the Time Course of Recovery by Voice Analysis

Recovery under Antidepressants

Investigations into antidepressant drug response are generally based on observer ratings and self-ratings. Therefore, the question arises as to the extent to which the respective results primarily reflect the expectations of raters and patients. To address this question, we conducted a study of 43 hospitalized depressive patients and assessed the time course of recovery at 3-day intervals throughout the first two weeks of treatment plus a final assessment at hospital release. Our sample included 17 mild cases (HAMD-17 baseline score < 22), 13 moderately depressed patients (HAMD-17 baseline score 22-27), and 13 severely depressed patients (HAMD-17 baseline score > 27). The patientsí psychopathology was assessed on the basis of the HAMD rating scale immediately after the speech recordings. All patients were treated with antidepressants.

Onset of Improvement

The onset of improvement (20% sustained baseline score reduction) occurred in the great majority of patients (79.1%) within the first 12 days of study, virtually independent of the severity of depression at baseline. Early improvement was highly predictive of later outcome as 67.6% of the patients showing improvement within the first 12 days were responders (50% sustained baseline score reduction) by the end of the observation period. Inversely, 92% of the responders at the end of the observation period exhibited an onset of improvement within the first 12 days. Early improvement could not be attributed to a few HAMD items, because score reductions included all items throughout all stages of treatment.

Speaking Behavior and Voice Sound Characteristics

The analysis of the patientsí speaking behavior and voice sound characteristics yielded, in 62.8% of the cases, an essentially parallel development over time for the HAMD scores on the one hand, and acoustic variables on the other (see Figures). The time course of improvement thus appeared to have a strong biological component and was unlikely to be attributable to the expectations of doctors and patients. Specifically, the Figures demonstrate that (1) the time course of recovery varies from patient to patient and may include deterioration, recovery can even be incomplete; and (2) speech parameters follow the psychopathology score changes very closely in either direction. A configuration of 15 speech parameters enabled discrimination between patients and controls at an overall performance of 75% correctly classified subjects.

voice analysis
Time course of a patient's recovery from depression as reflected by HAMD-17 scores (green square points) assessed at two-day intervals over an observation period of two weeks, plus a final assessment at the time of discharge from hospital ("day63"). The corresponding change over time of the speech parameter "F0-amplitude" is also shown (red circle points) in order to demonstrate the close relationship between the two courses of development (all but one of the chosen patients responded to therapy).

Everis, Spain
ETH, Switzerland
UZH, Switzerland
Freiburg, Germany
MA Systems, UK
Bristol, UK
Xiwrite, Italy
Ultrasis, UK
Jaume, Spain
Valencia, Spain
Lanzhou, China


EU-Grant (FP7):

[ Mail to Webmaster ] k454910@bli.uzh.ch