Semantically Enabled Modeling of Major Depressive Disorder (Depression Research)

Printer-friendly version
Research Areas: Data Science, Ontology Engineering Environments
Principal Investigator: Joanne S. Luciano
Co Investigator:
Concepts: eScience, Data Science
Description:
Although antidepressant treatment is readily available, the US government reports that Major Depressive Disorder is the leading cause of disability in the U.S. for ages 15-44. Why? Depression is complex. The diagnosis for Major Depressive Disorder describes the signs of depression, but not the cause. There are several distinct biological, physiological, and psychological systems that, when not functioning properly, could result in depression. Furthermore, clinicians are unable to predict which treatment will be successful, since not all patients respond to all antidepressant treatments. In order to properly treat depression, we need to understand the different types of depression, their underlying mechanisms of action, and how they interact in all three aspects: (1) behavioral (emotional and cognitive processes), (2) functional (chemical pathways from brain region to brain region), and (3) mechanistic (dynamics of the neurotransmitters and their receptors).

In this project, we study the effects of how different antidepressant treatments, including non-pharmacological treatments, affect the underlying brain regions, clinical symptoms, and behaviors. We use mathematical modeling and computer simulation to combine clinical research with neuroscience research. So far, these methods have indicated that different treatments result in different response patterns, i.e. with different treatments, the individual symptoms of depression (i.e. anxiety, mood, sleep disturbance, etc.) improve at different times and in a different order. Our current goals include increasing the number of treatments that we explore, incorporating semantic technologies, and ultimately having clinical impact. Findings like these have the potential for huge clinical impact; they could help reduce patient suffering and suicide by reducing the trial and error of current treatment practices.


Collaborators

Models

There were two differential equation models built, and one back propagation model. The differential equation models are implemented in matlab. We use the second order model to look at the dynamics of depression recovery to determines patterns of recovery for each treatment, i.e. when symptoms improve and the order in which they improve. The differential equation models analyzes the severity of individual symptoms of those patients who responded to treatment during the course of their treatment.


The backpropagation model, which is not currently live on this site, is used for prediction of response level (continuous and categorical) of responders and non-responders to antidepressant treatment.

  • Second Order Model (Use this one)
  • First Order Model
  • was included in the dissertation for comparison. It's not currently being used because the second order model provided a better fit to the datasets.

Antidepressant Treatment

Antidepressant treatment is whatever we have data for that was used to treat depression. The two most common forms of treatment are psychotherapy, as in Cognitive Behavioral Therapy (talk therapy), and a pharmacological agent (e.g. drug, medication, pills, etc.). There are other effective treatments, such as exercise, time with pet, placebo, that are also effective. If you know of sources of data for these other testaments, please contact Joanne Luciano. We would welcome the opportunity to analyze those data.

Brendan's Port Forward Script for Accessing the SEMMDD Virtuoso Endpoint

"C:\Program Files (x86)\PuTTY\Putty.exe" -ssh -C -N -l username -pw password -P 2244 -L 8080:localhost:8890 aquarius.tw.rpi.edu

On Linux/Mac (using OpenSSH)

ssh -N -l username -p 2244 -L 8080:localhost:8890 aquarius.tw.rpi.edu

Then navigate your brower to http://localhost:8080/sparql

Documents

[we probably want to move this elsewhere]

  1. Translational Medicine: Using Systems of Differential Equations to Identify Patterns in Symptom Remission in Response to Treatment and the Underlying Dynamics of their Interactions (2010 AMIA TBI abstract) [Download]
  2. Translational Medicine: Using Systems of Differential Equations to Identify Patterns in Symptom Remission in Response to Treatment and the Underlying Dynamics of their Interactions (2010 AMIA TBI paper [Download]
  3. Translational Medicine: Using Systems of Differential Equations to Identify Patterns in Symptom Remission in Response to Treatment and the Underlying Dynamics of their Interactions (2010 AMIA TBI presentation) [Download]
  4. Luciano_Dissertation_Prospectus_1994.PDF [Download]
  5. Neural Network Modeling of Unipolar Depression: Patterns of Recovery and Prediction of Outcome - Luciano Thesis Dissertation 1996 [Download]
  6. Individualized Treatment for Depressive Disorders: Providing a Biochemical Basis for Treatment Selection - NARSAD Proposal [Download]
  7. Treatment Outcome Prediction for Depressed Patients - NARSAD Proposal- [Download]
  8. A Neural Network Model of Biochemical Changes in Treatment Response of Depressed Patients - A NARSAD Proposal [Download]
  9. 12th EMBO|EMBL Conference on Science & Society talks available online
  10. Semantically Enabled Neural Network Modeling of Major Depressive Disorder - Yuezhang Xiao's Thesis [Download]