Machine Learning: Helping Determine How a Drug Affects the Brain

News   Nov 16, 2017 | Original story from University College London

 
Machine Learning: Helping Determine How a Drug Affects the Brain

Gaze recovery high-dimensional classifier weights. Represented as 3D cubic glyphs varying in colour and scale are the weights of a transductive linear support vector machine classifier trained to relate the high-dimensional pattern of damage to gaze outcome, achieving k-fold cross-validation performance of 78.33% (SE = 1.70%) sensitivity and 82.78% (SE = 0.56%) specificity for distinguishing between patients who recovered from a leftward deviation of gaze and those who did not. Positive weights (dark blue to cyan) favour recovery, negative weights (dark red to yellow) persistence of symptoms. Though hemispheric asymmetry is prominent, note the distribution of weights is highly complex, as one would expect from the complexity of the functional and lesional architectures that generate the critical pattern. Credit: https://doi.org/10.1093/brain/awx288

 
 
 

RELATED ARTICLES

Blood Sample Detection Method for Multiple Sclerosis Discovered

News

A method for quickly detecting signs of multiple sclerosis has been developed by a University of Huddersfield research team. The process will avoid the invasive, painful process of collecting fluid from the brain and spine

READ MORE

Regularly Playing Action Video Games Improves Cognition

News

An international team of psychologists has assembled data from the last fifteen years to quantify how action video games impact cognition.

READ MORE

Some Immune Cells May Help Tumors Instead of Destroying Them

News

New data shows that neutrophils promote tumor progression and can actually hamper the work of immunotherapy in lung cancer.

READ MORE

 

Comments | 0 ADD COMMENT

Like what you just read? You can find similar content on the communities below.

Drug Discovery Neuroscience

To personalize the content you see on Technology Networks homepage, Log In or Subscribe for Free

LOGIN SUBSCRIBE FOR FREE