EEG Signal Processing as a Predictor of Antidepressant Response
试验速览
- 阶段
- 不适用
- 入组人数
- 150
- 试验地点
- 1
- 主要终点
- Machine learning
研究概览
简要总结
Current methods of choosing treatment for major depressive disorder (MDD) are inefficient. The Strategic Treatment to Achieve Remission of Depression (STAR*D) Trial revealed that only about 1/3 of patients treated with antidepressant drugs will go into remission with the first medication chosen. We hypothesize that pattern recognition software using Machine Learning methods can accurately predict response to a variety of antidepressant medications (ADM) or cognitive behavior therapy (CBT) after training using pre-treatment demographic, clinical, laboratory or electroencephalographic (EEG) data. These algorithms might assist the clinician to chose, for any given patient, an antidepressant treatment option with greater probability of favourable response than is achievable using current best practise methods.
详细描述
Objective of this study:
To improve antidepressant treatment efficacy by determining,in advance, a given subject's probability of response to a range of antidepressant treatments. The study is intended to to further train and test, in a larger sample of depressed subjects, a digital system that has been shown to be an accurate predictor of antidepressant response in pilot studies. The accuracy of the trained predictive model based on machine learning methodology is the primary outcome we are interested in studying.
Subjects:
males and females age 18-70 years of age.
Inclusion Criteria:
研究设计
- 研究类型
- Observational
- 观察模型
- Case Only
- 时间视角
- Prospective
入排标准
- 年龄范围
- 18 Years 至 70 Years（Adult, Older Adult）
- 性别
- All
- 接受健康志愿者
- 否
入选标准
- •Clients with Major Depression
- •Males and Females ages 18 - 70
排除标准
- •Clients who have known neurological problems
- •Clients with a history of severe head injury
- •Clients with strong thoughts of suicide
- •Clients who have had ECT or Cognitive Behavior Therapy within 6 months
- •Females who are sexually active and are not on adequate birth control
研究组 & 干预措施
Drug 1
Venlafaxine
干预措施: Venlafaxine (Drug)
Drug 2
Bupropion
干预措施: bupropion (Drug)
Drug 3
Escitalopram
干预措施: escitalopram (Drug)
Drug 4
Duloxetine
干预措施: Duloxetine (Drug)
Psychotherapy
Cognitive behaviour therapy
干预措施: Psychotherapy (Other)
结局指标
主要结局
Machine learning
时间窗: 6 weeks with medication, or 12 weeks with CBT
The accuracy of the trained predictive model based on machine learning methodology is the primary outcome we are interested in studying. The primary outcome measure, i.e. model performance accuracy, is tested using the jack-knifed "leave N out" nested cross validation method with response being determined using the MDRS scale.
次要结局
- Machine learning(6 weeks with medication, 12 weeks with CBT)
研究者
Gary Hasey
Associate Professor, Department of Psychiatry and Behavioral Neurosciences
St. Joseph's Healthcare Hamilton