Despite having several FDA-approved therapies and clear treatment guidelines, the approach to treat seizures for Dravet syndrome can often be a repeating cycle of ‘trial-and-error’ as patients are placed on anti-seizure medications and then families must wait to assess the impact on seizure control over time. This method often takes months to determine whether a therapy is having a positive impact, and, if it is not, then the entire process must start over again as that medication is weaned and another medication is started. A biomarker that could help to guide medication choice or give earlier insight about the effectiveness of a medication would transform this entire cycle. DSF just announced funding for a project that aims to solve this issue using artificial-intelligence-driven analysis of behavior.
In 2023, a ground-breaking research study was published describing how artificial intelligence and machine learning could be used to analyze videos of mice to find “hidden behavioral fingerprints” that could provide insights about the health of the mice and how well different interventions worked to control seizures in models of epilepsy. They used motion sequencing (called “MoSeq”) which combines machine learning with assessment of 3D imaging, allowing for automated review of video recordings of mouse models of epilepsy and categorize behavioral patterns. These behavioral patterns could then be compared between mice with epilepsy (including a SCN1B model of Dravet syndrome) and healthy mice, where differences emerged (that would not have been observable by a human reviewer alone) that could distinguish between the groups of mice. They could detect these differences without including segments of video that included actual seizure activity, focusing on behaviors during normal activity. The data was so powerful, that they could determine differences related to the sex of the animal, seizure and epilepsy types, and importantly, after treatment with antiseizure medications they could determine information about the dose and effectiveness of the therapy without tracking seizures or mortality.
The same year this study was published, Dr. Ted Odlaug spoke at DSF’s annual Research Roundtable about his granddaughter, Anna, and the challenges her family has faced with finding effective treatments, describing the trial-and-error cycle they found themselves in. Some of the researchers in attendance felt particularly inspired to determine if the MoSeq approach could work in humans just as in mice, to provide an objective biomarker for insights about seizure burden, treatment response, and other aspects of patient health. Now with funding from DSF, researchers from Stanford University and University of Michigan are working together to test this approach in children with Dravet syndrome over the next two years. If successful, this could shorten the timelines to find the most effective treatment regimens for patients, and it could also be used in clinical studies as a faster outcome measure to determine if a new therapy might be working.
Now, the Dravet patient-family community can help push this research forward!
In order to “teach” the MoSeq program, the research team will need to have it analyze videos of children (2-12 years old) with Dravet syndrome going about their normal activities (e.g., playing or eating). Ideal videos for analysis should include primarily only the child with Dravet syndrome, have good lighting, and minimal background objects or patterns (such as a solid color wall).
Once they are able to “teach” the MoSeq program, they hope to do a larger scale study in the future to determine whether MoSeq can predict if a child with DS will respond to an antiseizure medication based on how the medication affects their movements and behavior.
If you are interested in contributing a video of your child to help teach the MoSeq program or if you have any questions, please contact: moseq-study@umich.edu
The original study that this work is based upon:
Gschwind, T., Zeine, A., Raikov, I., Markowitz, J.E., Gillis, W.F., Felong, S., Isom, L.L., Datta, S.R., Soltesz, I., 2023. Hidden behavioral fingerprints in epilepsy. Neuron. https://doi.org/10.1016/j.neuron.2023.02.003
To find more information about this study and other studies currently enrolling patients with Dravet syndrome visit DravetClinicalTrials.org
