AI can detect epileptic behavior better than the human eye
Epilepsy is a neurological disorder that affects millions of people worldwide, yet diagnosing and treating the condition remains a complex and subjective process. While high-resolution imaging has provided healthcare professionals and researchers with a deeper understanding of the brain circuit malfunctions seen in people with epilepsy, little is known about how epilepsy affects behavior.
Now, a new study used state-of-the-art artificial intelligence (AI) on mice to identify epilepsy-related behavior often missed by the human eye. The team led by Stanford researchers studied mice with acquired and genetic epilepsies. They found that machine analysis was better able to distinguish epileptic vs. non-epileptic mice than trained human observers, reports New Atlas.
Traditional methods of epilepsy diagnosis and treatment assessment involve the use of continuous video-electroencephalogram (EEG) monitoring over several days or weeks. While this approach can be helpful, it is also a relatively blunt tool, given the complexity and diversity of the condition and the fact that some seizures may not appear on EEG. Moreover, it is labor-intensive and subjective, requiring trained professionals to analyze hours of video-EEG recordings and relies on their ability to notice often slight behavioral changes.
How does AI analyze the behavior of epileptic mice?
Researchers used an AI technology called MoSeq (Motion Sequencing) to analyze the behavior of epileptic mice and identify behavioral "fingerprints" that may go unnoticed by humans. MoSeq is a machine-learning technology that trains an unsupervised machine to identify repeated patterns of behavior. After identifying the behaviors, MoSeq offers visualization tools and statistical tests to help scientists understand those behaviors and compare them to various experimental conditions.
By using MoSeq to analyze 3D videos of freely moving mice, the researchers could locate, track and quantify the behavior of the mice. They found that the technology could better distinguish between epileptic and non-epileptic mice, outperforming trained human observers. Furthermore, it required only one hour of video recording and did not need a seizure to occur before offering its analysis, unlike traditional methods. The researchers were able to differentiate between patterns of behavior in the mice after they were given one of three anti-epileptic medications.
The successful use of machine-learning technology demonstrates its potential for diagnosing epilepsy and testing the efficacy of anti-epileptic medications in humans. This technology offers a faster, less labor-intensive, less costly, and more objective approach to diagnosis and treatment assessment.
The study was published in the journal Neuron.
Epilepsy is a major disorder affecting millions of people. Although modern electrophysiological and imaging approaches provide high-resolution access to the multi-scale brain circuit malfunctions in epilepsy, our understanding of how behavior changes with epilepsy has remained rudimentary. As a result, screening for new therapies for children and adults with devastating epilepsies still relies on the inherently subjective, semi-quantitative assessment of a handful of pre-selected behavioral signs of epilepsy in animal models. Here, we use machine learning-assisted 3D video analysis to reveal hidden behavioral phenotypes in mice with acquired and genetic epilepsies and track their alterations during post-insult epileptogenesis and in response to anti-epileptic drugs. These results show the persistent reconfiguration of behavioral fingerprints in epilepsy and indicate that they can be employed for rapid, automated anti-epileptic drug testing at scale.
A simulated moonwalk in Arizona allowed engineers to test a wearable for future Artemis astronauts.