Donate to Science & Enterprise

S&E on Mastodon

S&E on LinkedIn

S&E on Flipboard

Please share Science & Enterprise

NSF Funding Data Mining for Depression Diagnosis

Heng Huang

Heng Huang (University of Texas – Arlington)

12 August 2016. A computer scientist at University of Texas in Arlington is developing data mining techniques and algorithms to help clinicians analyze unstructured notes accumulated from patients in therapy for depression. National Science Foundation awarded $500,000 to UT-Arlington’s Heng Huang for the three-year project.

Depression is a common mood disorder, but can become a serious and disabling condition, when it persists for long periods or interferes with day-to-day living. Centers for Disease Control and Prevention says depression affects about  7.6 percent of the U.S. population age 12 and over in any 2-week period. In addition, CDC notes that major depressive disorder, also known as clinical depression, resulted in 8 million visits to U.S. doctor’s offices or hospitals, in 2009-2010.

For many people with depression, cognitive behavioral therapy where a therapist asks questions of the the individual is an effective treatment. Therapists need to keep records of these conversations, which become the basis for uncovering patterns in the person’s language indicating the direction of the disease. In addition, clinicians will ask the individuals to keep a journal of their thoughts between therapy sessions.

But the mass of these journals and session notes, often handwritten and sometimes recorded, can become unwieldy and difficult to analyze. “Each patient will write dozens of thoughts each day, and it can be very difficult to get through them all,” says Huang in a university statement. “Nurses are not trained to analyze these journals, so it is up to the doctor or therapist to read and analyze the information contained in them. Using data mining, it becomes much easier to analyze the thoughts and apply them to a treatment plan.”

Huang and colleagues plan to apply techniques from artificial intelligence and machine learning to tackle these records. The UT-Arlington team expects to develop a learning model that categorizes logical thinking errors in the depression thought records. These are patterns of though, such as unsupported personalization or all-or-nothing thinking that reinforce negative beliefs and deepen a depressed mood.

The model will be refined as new records are collected to uncover patterns of progression for the disease as well as identify coping activities by the individuals. From the model, the team will devise algorithms that can be applied to large-scale and therapeutic analysis. In addition, the project is expected to develop tools for training new therapists to treat people with depression.

Computational techniques can be applied as well to uncovering genomic associations with depression. As reported earlier this month in Science & Enterprise, a team from Massachusetts General Hospital and the personal genetics company 23andMe reported on data from commercial genetic tests that identified 15 regions in the genome and 17 specific variations associated with depression among people of European descent.

Read more:

*     *     *

1 comment to NSF Funding Data Mining for Depression Diagnosis