Google is teaming up with academics to create new artificial intelligence (AI) algorithms to better brain stimulation devices used to treat patients with mental illnesses and acute brain injuries like stroke.

The internet giant has teamed up with Mayo Clinic researchers to create a collection of paradigms, or points of view, that make comparing the effects of electrical stimulation on the brain easier.

They created a "base profile curve identification" algorithm, which is a new sort of algorithm.

"Our findings demonstrate that this new sort of algorithm can help us understand which brain areas directly connect with one another," said Kai Miller, a Mayo Clinic neurosurgeon.

"As additional technology becomes available, this sort of algorithm may help us better treat patients with epilepsy, movement disorders such as Parkinson's disease, and psychiatric diseases such as obsessive compulsive disorder and depression," he added.

In a human patient, the novel method was shown with an array of implanted brain surface electrodes.

Before a brain tumour was excised, an electrocorticographic electrode array was implanted to identify seizures and map brain activity in a patient with a brain tumour.

Hundreds to thousands of time points were generated for each electrode contact, allowing the novel method to be analyzed.

The approach allows for easy analysis of single-pulse brain stimulation data and may be used to investigate the broad milieu of connections that make up the connectome, according to the researchers in a paper published in PLOS Computational Biology.

"To date, neurologic data has been probably the most hard and fascinating data to model for AI researchers”, said Klaus-Robert Mueller, a Google Research Brain Team member.

"Our findings suggest that this new sort of algorithm may help us understand which brain regions directly communicate with one another," says Kai Miller, M.D., Ph.D., a Mayo Clinic neurosurgeon and the study's first author. "As additional technology becomes available, this sort of algorithm may help us better treat patients with epilepsy, Parkinson's disease, and mental diseases including obsessive compulsive disorder and depression”.

Klaus-Robert Mueller, Ph.D., study co-author and member of the Google Research Brain Team, adds, "Neurologic data to date is arguably the most demanding and fascinating data to model for AI researchers." Dr. Mueller is the director of the Machine Learning Group at Technical University of Berlin and co-director of the Berlin Institute for the Foundations of Learning and Data.

The authors of the study give a downloadable code package so that others can experiment with the approach. Dora Hermes, Ph.D., a Mayo Clinic biomedical engineer and senior author, adds, "Sharing the created code is a fundamental component of our efforts to promote reproducibility of research”.