Memory is often affected after a traumatic brain injury, and it can be long-term or even permanent. Artificial intelligence-guided electrical brain stimulation can successfully treat memory loss in patients with moderate-to-severe traumatic brain injury, a new study finds.
Traumatic brain injury (TBI) is a type of brain injury, usually caused by a hard blow or shaking to the head or body. It is often caused by motor vehicle accidents, contact sports, firearms, and falls. The severity of a TBI depends on the nature of the attack on the brain and the extent of inflammation and damage to neurons and blood vessels.
Immediately after a traumatic brain injury, a person may experience post-traumatic amnesia (PTA), where they feel confused, disoriented, and have no memory of what happened after the injury. While PTA is short-lived and can last from minutes to months, moderate-to-severe TBI can have lasting effects on memory, especially episodic memory, the ability to recall and mentally re-experience specific scenes from the past.
In a new study, Penn researchers investigated how providing artificial intelligence-guided electrical brain stimulation to patients with moderate-to-severe TBI could improve their memory.
“These patients are often relatively young and healthy, but they face decades of impaired memory and executive function,” said Michael Kahana, lead author of the study.
The current research builds on Study in 2017 Researchers at the university have found that applying electrical stimulation improves memory function when a machine learning model predicts that the memory is about to decline. The 2017 study used open-loop stimulation, meaning the AI applied it without considering the state of the brain.
2018The researchers then tested their AI-guided model on 25 epilepsy patients, monitoring their brain activity in real time and stimulating the left side of the brain only when memory decline was expected; that is, closed-loop stimulation. They found a 15% improvement in memory and academic performance.
For the current study, the researchers recruited eight patients with refractory epilepsy, defined as epilepsy that cannot be controlled with medication, and a history of moderate-to-severe TBI, who were being monitored for seizures using implanted intracranial electrodes. Each participant took a delayed recall test, asking them to recall a series of consecutive occurrences of nouns. Recall tasks were performed with or without closed-loop brain stimulation.
During treatment without stimulation, the researchers identified individualized biomarkers of successful memory encoding that they could use later in the course of treatment to control closed-loop brain stimulation. During stimulation, a machine learning algorithm decoded memory success rates in real time and triggered stimulation when participants’ memory performance fell below the predicted average.
When the researchers examined recall rates for the closed-loop stimuli, they found that the participants recalled the items more reliably than when no stimulus was used. On average, closed-loop stimulation improved memory by 19%.
The researchers say their study demonstrates a proof-of-concept for using closed-loop brain stimulation to treat memory loss in TBI patients.
“By demonstrating the efficacy of the treatment in patients with a history of moderate to severe TBI and documented memory impairment, we hope that our findings will accelerate the development of technologies for patients with acquired brain injury that could restore some of their lost memory. They come into play when they’re trying to reconstruct life after an injury,” the researchers said.
The study was published in the journal brain stimulation.
source: University of Pennsylvania