Psychiatric drug response predicted by scientists

Psychiatrists and engineers have teamed up to help predict how people with schizophrenia will respond to a medication that can produce serious side-effects.

Psychiatrists and engineers have teamed up to help predict how people with schizophrenia will respond to a medication that can produce serious side-effects.

Psychiatrists say clozapine is an effective treatment for chronic medication-resistant schizophrenia, but it can produce side-effects such as seizures, cardiac arrhythmias or bone marrow suppression and blood problems that require weekly to monthly blood tests to monitor.

Now researchers at McMaster University in Hamilton have used machine learning to "train" a computer to predict whether a patient will respond to the drug based on the brainwave patterns and responses recorded on an EEG device readily available at hospitals and laboratories. 

"Now what we can do is predict beforehand whether the person is going to respond, so we only expose the patient to the risk if there's a very good chance the treatment will be effective," said study author James Reilly, a professor of electrical and computer engineering at McMaster.

Reilly and his psychiatrist and engineering colleagues at the university were able to correctly predict whether 23 middle-aged people diagnosed with schizophrenia would respond to the drug with an accuracy of about 89 per cent, according to the study published in the current online issue of the journal Clinical Neurophysiology.

They found the same rate in further testing on a second group of 14 patients.

If the findings are also found in a larger population then this new approach to EEG analysis may help doctors predict the clinical response to clozapine in treatment resistant schizophrenia, the study's authors concluded.

Antidepressant tests

Neuroscientists have long used EEGs to try to understand the brain. An EEG uses electrodes placed on the scalp to measure electrical impulses produced by brain cells. Until now, researchers were limited in the type of useful information they could extract from EEG signals.

But Reilly's PhD student, Ahmad Khodayari-Rostamabad, thought of applying the power of machine learning — a scientific method that can extract information from large data sets — to better diagnose mental illness. 

A second part of the research applied the same EEG technique for depression.

There is a long list of antidepressant medications in a variety of classes that work differently, but psychiatrists rely on a trial-and-error process to test which drug will work for each patient.

People with depression may be debilitated for a long time before the best medication is found, but the new technique has the potential to shorten the process, Reilly said.

EEG is inexpensive and non-invasive compared with techniques like X-rays or CT scans. The readings take about 30 minutes of a patient's time, said study co-author and psychiatrist Dr. Duncan MacCrimmon.

Traditionally, EEGs are used to monitor for epilepsy, and to diagnose coma, brain death and some brain disorders.

Anyone in the Hamilton area who is interested in taking part in the study can call St. Joseph's at 905-522-1155, ext. 36629.

The research was funded by The Magstim Company, a Wales-based developer and manufacturer of medical and research devices.