Optimizing Therapy in Epilepsy Using Seizure Forecasts Via EEG and Wearables

For most people with epilepsy, seizures do not appear to happen on a predictable schedule, but recent research shows that seizure risk may actually rise and fall in patterns over time, similar to other body rhythms. Right now, most people with drug-resistant epilepsy (epilepsy that has not responded well to medications) take antiseizure medication every day, whether or not their seizure risk is high that day, which means they deal with side effects every day even though seizures happen only sometimes. This study tests using a small implanted sensor and a wearable device to predict when a person's seizure risk is elevated, and then giving a fast-acting medication in advance, before a seizure happens. The goal is to find out whether this approach is safe and workable in everyday life. This study will enroll adults with drug-resistant epilepsy that involves the temporal lobe, at Mayo Clinic in Rochester, Minnesota, Jacksonville, Florida, and Phoenix, Arizona. Each participant has a small device called the UNEEG SubQ placed just under the scalp, in a minor surgical procedure. This device does not go into the brain itself. It sits under the skin and continuously records the brain's electrical activity. Participants also wear an smart watch to track heart rate and movement, and they keep a daily seizure diary on a smartphone. The study takes place over about 12 months for each participant, in three parts. Part 1, the first three months: The device and watch collect data in the background. No warnings or predictions are shown to the participant during this time. This information is used to build a personalized computer model that learns each person's individual patterns of seizure risk. For four weeks during this period, participants take a scheduled dose of VALTOCO (a nasal spray form of the medication diazepam) once a week, which helps the research team see how the medication affects the recorded brain activity. Part 2, the next three months: The personalized computer model starts making daily predictions about seizure risk, but these predictions are reviewed only by the research team and are not shown to the participant yet. This lets the team check how accurate the predictions are before relying on them. As in Part 1, participants take one scheduled weekly dose of VALTOCO during a four week period. Part 3, the final six months: Starting at a mid-study visit, participants begin receiving the daily predictions on their smartphone. On days when the model predicts a higher chance of seizures, the participant (or a caregiver) receives an alert and is asked to use VALTOCO nasal spray at that time, before a seizure occurs, rather than waiting for one to happen. To help keep the medication use safe, participants cannot use VALTOCO more than 10 times in a month, and the system will not allow the medication to be used on two days in a row. At the end of the study, the implanted device is removed in a simple outpatient procedure and all study equipment is returned. The main goal is to find out whether this forecast-guided approach to using rescue medication is safe and practical to use in daily life. As additional goals, the study will also look at whether participants' quality of life and the impact epilepsy has on their daily life improve during the study, using standard questionnaires, and whether the rate of seizures during the final six month part of the study is lower than each participant's own personal baseline rate measured earlier in the study. This study is conducted with authorization from the U.S. Food and Drug Administration under Investigational New Drug application number 170383, which allows the research team to study VALTOCO for this new, investigational use. The study is supported by a grant from the National Institutes of Health (UG3NS123066).