- Robert F. Dickerson (email@example.com)
- Bethany Teachman, Professor of Psychology
- Ann Taylor, Professor of Nursing
- Karen Rose, Professor of Nursing
- John A. Stankovic, Professor of Computer Science (firstname.lastname@example.org)
- This work appears in MIT Technology Review January/February issue!
Summary: Depression is a major health issue. Depression is often unrecognized and untreated. It also leads to many other medical problems because of reduced social interactions, less personal hygiene, increased alcohol use, and ignoring medication for current medical conditions. The main goal of this proposed work is to complete the implementation of a real-time depression monitoring product for the home. This product will run 24/7 and can detect the signs of depression early (in real-time) as well as monitor those already diagnosed with depression. It is multi-modal to increase accuracy and provide caregivers with accurate information to aide in their care giving and diagnosis. The same product can also be used to provide information about the effectiveness of any treatment. The end result will be improved quality of life and possible improvement of other medical conditions and problems caused by or related to the depression. The product produced is a cohesive set of integrated wireless sensors, a touch screen station, and associated software that delivers the above capabilities. Once the implementation is complete, it is expected that this product can be transferred to industry for use in homes and assisted living facilities.
Acoustic Sensing: Acoustic features of voice such as pitch, utterance duration, and amplitude have been used in pre and post-treatment studies to help detect signs depression. However, these solutions are typically done in controlled environments under the oversight of speech pathology experts analyzing the patient’s voice at a fixed distance from a microphone. Our challenge is to implement these solutions to work at real-time in natural home settings. This requires a number of novel extensions to known results that include filters, vocal discrimination, and real-time segmentation. We will develop patient-specific classifiers using inputs from other modules in the system to recognize the confluence of factors that arise when someone suffers from depression.
Caregiver Displays: We developed user interfaces for caregivers. The caregiver’s screen is shown in the figure. This shows an overview of all attending patients. For each patient, an overview of the current behavior factors, sleeping quality, hygiene, PHQ score, weight, eating, social level, and mood are presented. Each factor is represented on a 5-point scale representing the anomalous nature or danger of that factor. When the caregiver selects the factor, a new view appears with a time-series plot. Annotations appear on the plot indicating if and where a patient has started a new diet plan or medication. Multiple factors that are highly correlated can be brought up and compared for analysis.
The patient interface runs on the touchscreen and/or tablet placed in an accessible room inside the patient’s apartment. This multi-purpose interface serves as a health trainer, social planner, and mood journal. Personal behavioral factors are shown to the patient, providing objective measurements for positive feedback. The social planner manages a record of activities that are occurring in the patient’s assisted living care center, senior center, or other organization, and holds RSVP’s and attendance records for the patient and also monitors the same for friends. Thus, it becomes a valuable input for measuring social involvement. User-tailored recommendations that can be provided improve the social factor. We propose to enhance this interface with the PSQI, additional user feedback and need to also implement the second level of information for care givers and doctors.