A to Z Challenge: D is for Depression (& Machine Learning)

mlI knew that I wanted to write about depression today for the A to Z Challenge, but I didn’t want to basically rewrite yesterday’s post. Nor did I want to post a bunch of statistics about depression. Most of us know, not only the statistics, but people near and dear to us who suffer from depression. I’m also sure a good number of us suffer from the disease. We know the general stuff. So what would I write about today about depression? Google to the rescue. After a basic search I found a fascinating link between my husband’s current computer obsession and my depression.

The Hubby is a total computer nerd. He is a software engineer, and has spent most of his career writing software (those search results Google brings up: he helped write that software). His current obsession is machine learning. What is machine learning you ask? I think WhatIs.com has the best definition in normal words that normal people will understand: “Machine learning is a type of artificial intelligence (AI) that provides computers with the ability to learn without being explicitly programmed. Machine learning focuses on the development of computer programs that can change when exposed to new data.”

What does that have to do with depression? Psychologists are now using machine learning hoping to come up with more accurate, medically based ways to diagnose depression and treat it more effectively. Last month Science Daily published the results of Dr. David Schnyer’s research at The University of Texas in Austin. Dr. Schnyer and his team are “using Magnetic Resonance Imaging (MRI) brain scans, genomics data and other relevant factors, to provide accurate predictions of risk for those with depression and anxiety.”

In other words they are using A LOT of data–so much data that it’s impossible for human beings to correlate the information in a reasonable amount of time. That is where machine learning comes in. Dr. Schnyer is receiving help from the Texas Advanced Computing Center to feed information into their supercomputer, Stampede:

The type of machine learning that Schnyer and his team tested is called Support Vector Machine Learning. The researchers provided a set of training examples, each marked as belonging to either healthy individuals or those who have been diagnosed with depression. Schnyer and his team labelled features in their data that were meaningful, and these examples were used to train the system. A computer then scanned the data, found subtle connections between disparate parts, and built a model that assigns new examples to one category or the other.

Schnyer and his team then went on to scan the brains of 52 people with depression and 45 healthy people in a control group. Stampede was able to diagnose those with depression with a 75% accuracy rate.

The thing that struck me the most about this study is that the researchers discovered that depression does not effect just one area of the brain (one of the reason past studies have failed to come up with a way to diagnose depression using brains scans). Depression affects multiple regions of the brain and how the brain works.

In Scientific American I came across another study. Dr. Conor Liston of Weill Cornell Medicine does not like how people with vast, multiple symptoms are grouped together under the umbrella diagnosis: depression. He is using machine learning to diagnose depression into four subsets that will make treatment, both with and without drugs, more effective.

I’m looking forward to keeping an eye out for more studies. This is a new frontier for depression diagnosis and treatment, and I’m interested in seeing if they will be able to replicate findings in different studies, and if they do what that means for being able to more accurately diagnose the disease. I’m very interested in what this will mean for future treatments. Right now treatment for depression is a hit or miss approach to see what works for each individual. In the future brain scans and machine learning might help, not only to diagnosis depression and its subsets, but suggest treatments that will be much more effective for those of us with depression.

Advertisements

15 thoughts on “A to Z Challenge: D is for Depression (& Machine Learning)

  1. Very informative, I must say. I have studied the basics of Machine Learning, especially Support Vector Machines (SVM), as a part of my post-graduate coursework in computational biology. It involves some amount of programming with statistics. Of course, the predictions depend largely on the extent of data for training the model as well as the descriptors (in this case – the features used for differentiating a particular subset of depression from another). The test data (symptoms of patients who haven’t been diagnosed) is used for testing the accuracy of the SVM model.

    The application of ML and SVM in clinical diagnostics will surely boost further treatment. But here’s the big question, how much intelligence can a computer possess?

    Great read. Hope you feel better.

    Love and light,
    Anjali

    Like

    1. Oh I’m married to a coder too, but he never stops learning new things. One of the reasons I married him (that and free computer problem solving. :)) Mental health overall is neglected, which is one of the reasons I talk about my depression. There’s still so much stigma and shame surrounding it that needs to be done away with, so it can be treated as the medical condition it is.

      Like

  2. Off late I am reading and seeing so many thing about depression that sometimes it scares me
    Maybe 10-20 years back this wasn’t the case, people were happy. Atleast they pretended to be.
    Its must be alarming to operate the machines and find out that the person they are using it on is depressed.
    Intresting link though

    Like

    1. I’m glad it is. I get tired of mental illness by the red-headed stepchild in the attic. It is a valid medical diagnosis and needs to be treated that way. That’s why these studies make me so happy: depression is being taken seriously as a medical condition.

      Like

  3. Didn’t know any of this. It’s so interesting. Depression is a very sneaky illness. Sometimes people suffer from it and they don’t even realise it. These new technique may help in that sense too.

    Like

  4. This is fantastic. I’ve had friends with mental health issues, including depression, and, when they got their medication, it didn’t always work. It would be great to use technology to create more effective treatments for these challenging conditions.

    Like

    1. I’m excited about the prospects of psychiatrists being able to tailor treatment to the patient’s depression. I’m not on medication right now, but that doesn’t mean I won’t go back on at some point in the future (especially when my hormones start screwing with me in a few years when I start menopause). It would be nice if my doctor would be able to pinpoint the right kinds of anti-depressants to put me on instead of the hit and miss approach they have to take now.

      Like

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s