Epilepsy care is moving from reacting to seizures toward anticipating them, with wearable AI for early warning and the first new antiseizure drug mechanism in years heading to the FDA.

Key Takeaways

• From reacting to predicting: Wearable devices and AI are being built to warn people minutes before a seizure, not just record it afterwards. Some are already CE-marked in Europe; most are still in research.

• A new drug mechanism: In September 2026, Xenon Pharmaceuticals filed with the FDA for azetukalner, a once-daily add-on that opens KV7 potassium channels. If approved, it would be the first drug of its kind for epilepsy.

• Beyond pills: Gene therapies, responsive neurostimulation and deep brain stimulation are widening options for people whose seizures do not respond to medication. Epilepsy affects about 50 million people worldwide, and roughly one in three does not become seizure-free on drugs.

Catching Seizures Earlier: AI, Wearables and EEG

The gold standard for diagnosing seizures is still electroencephalography (EEG), which records the brain's electrical activity. It is accurate but hard to use for long-term monitoring at home. Wearables aim to fill that gap by tracking signals such as heart rate, movement and skin response, and using machine learning to flag a seizure.

A 2026 review in Frontiers in Neurology synthesised 23 studies of wearable devices and AI algorithms for automated seizure detection. It found that newer work is moving from detecting seizures to anticipating them, while noting that reliable physiological markers and testing across diverse populations are still needed before these tools become routine care. An earlier scoping review of 67 studies showed the imbalance: about 81% of models focused on detection and about 21% on prediction.

Two developments show where this is heading:

• Earpiece for seizure forecasting: Neuraxpharm and the Spanish start-up mjn-neuro have developed an AI-powered earpiece that records brain activity and sends risk alerts to a mobile app. It holds a CE mark (Class IIa) and UKCA approval, is intended for people with drug-resistant epilepsy, and was scheduled to launch in Europe and the UK in the second half of 2026.

• Headset from Scotland: Researchers at Glasgow Caledonian University reported an AI-enabled headset that combines brainwave and heart data to predict a seizure minutes before it happens. It is a research prototype, not a product you can buy.

New Treatments: Drugs, Gene Therapy and Neuromodulation

The FDA has already approved more than 30 antiseizure medications, yet only a handful of distinct mechanisms are available. That is why 2026's pipeline matters: it adds new mechanisms and targets the cause in specific syndromes. A 2026 European Academy of Neurology update groups the advances into three categories: new small molecules, gene-targeted therapies and devices.

New small molecule: azetukalner. In the phase 3 X-TOLE2 trial, 380 adults with highly treatment-resistant focal epilepsy, previously treated with five antiseizure medications, received add-on azetukalner at 15 mg or 25 mg, or placebo, for 12 weeks. Seizure frequency fell significantly more than with placebo. It is taken once daily without dose titration. Xenon submitted its New Drug Application on September 17, 2026. It remains investigational and is not FDA-approved.

Gene-targeted therapies. More than 90% of people with Dravet syndrome have an SCN1A mutation. Zorevunersen (Stoke Therapeutics and Biogen) is designed to boost production of the missing protein from the healthy copy of that gene, with a data readout expected in the third quarter of 2027. A one-time gene therapy, ETX101, targets the same gene and has shown seizure reductions in early trials. For drug-resistant mesial temporal lobe epilepsy, uniQure has dosed the first patient with AMT-260 in an early-stage trial.

Neuromodulation. Responsive neurostimulation (RNS) and deep brain stimulation (DBS) are established options for drug-resistant epilepsy. RNS listens for abnormal brain activity and delivers stimulation in response, while DBS delivers stimulation on a set schedule. Both can help people who are not candidates for surgery or did not respond to medication.

What Patients and Families Can Do Now

• Ask about specialist referral. If two appropriately chosen medications have not controlled seizures, ask your doctor about an evaluation at an epilepsy centre. Surgery, devices and trials are options only a specialist can assess.

• Keep a seizure diary. Record dates, times, triggers, sleep and missed doses. Good records help doctors judge whether a treatment is working and are useful if you consider a wearable.

• Treat wearables as support, not a safety guarantee. Check whether a device is approved or cleared for your region and seizure type, and keep following your care plan. Prediction tools are still maturing.

• Know seizure first aid. Stay with the person, time the seizure, cushion their head, turn them onto their side once the shaking stops, and put nothing in their mouth. Call emergency services if it lasts more than 5 minutes, repeats without recovery, or the person is injured or not breathing normally.

• Be sceptical of miracle claims. New drugs and gene therapies in the news are mostly investigational. Do not stop or change medication on your own, and ask your doctor about legitimate clinical trials.

This article is for general information and is not medical advice. Please consult a qualified neurologist about your own care.

Sources

• Frontiers in Neurology: AI in wearable seizure detection devices (2026)

• JMIR: Wearable AI for epilepsy, scoping review

• EMJ: Epileptic seizure detection device to launch in Europe

• WAM: Scottish scientists create AI-enabled headset

• NeurologyLive: Xenon submits NDA for azetukalner

• Practical Neurology: X-TOLE2 phase 3 results

• PharmaVoice: Epilepsy R&D, gene therapy and seizure drugs

• EAN: Updates in epilepsy treatment 2026