2023-10-08 14:00:00
A special bracelet might warn people suffering from epilepsy of an approaching seizure. Researchers from Upper Austria are currently working on the requirements for this and are using the latest methods of artificial intelligence and dogs’ keen sense of smell.
A sudden seizure: This is what many people suffering from epilepsy fear most. The fear of suddenly becoming unconscious in the middle of the street limits her life, day following day.
A research team from Linz is now working on making life easier for those affected with the help of artificial intelligence (AI): They want to find out whether the body sends out warning signals before one of the dreaded attacks occurs. “If we succeed in establishing an early warning system, those affected might take timely measures, for example taking medication or finding a safe environment,” says project leader Wolfgang Narzt from the Institute for Information Systems/Software Engineering at Johannes Kepler University (JKU). “Although you can’t prevent the seizure itself, you can reduce the risk of injuries that might result from the unconscious person falling. In some forms of epilepsy it also happens that sufferers cannot control their movements and, for example, get their hands into hot water while cooking, whereby they only notice the burns they have suffered when they come to.
Alarm in normal everyday life
Currently, predictions of epileptic seizures are only possible to a limited extent in a clinical setting: the patients have to stay in the hospital and be connected to medical devices at all times. “We, on the other hand, want to find a solution to warn those affected of an impending attack in good time in everyday life and in their usual living environment,” explains Narzt. Therefore, a study is starting these days in collaboration with the Department of Neurology 1 at the University Hospital Linz and the Institute for Machine Learning at the JKU. The company Five Square, which specializes in artificial intelligence, is an industry partner. Doctor: “The test subjects are equipped with a network of wearable sensors. We use this to record biological parameters such as the heartbeat, the oxygen saturation of the blood, the blood flow in the brain, the skin resistance, muscle contractions and much more. Presumably not all will be relevant in relation to epileptic seizures. Our aim is to find out which factors are actually decisive.”
The researchers are inspired by the animal world: specially trained dogs are able to predict epilepsy attacks better than is currently possible with scientific methods. They bark when a seizure is imminent. It is believed that the four-legged friends use their fine noses to detect changes in the body odor of those affected. “We are the first research team in the world to try to prove this by, among other things, recording the sweat produced by those affected using sensors,” says Narzt. This comprehensive monitoring generates a huge amount of data. Heart rate variability alone is measured and recorded at intervals of milliseconds. This is where AI comes into play. “It is fed with data from all sensors and looks for patterns in the biological factors that appear before epileptic seizures. If these patterns then occur in everyday life, those affected know that an attack is approaching.”
Application is a thing of the future
The researchers are using so-called transformer technology, the latest AI models, and expect more meaningful results than those achieved with previous – even less comprehensive – studies. The sick might ultimately be warned by devices to be developed, such as medical bracelets or cuffs, which are equipped with the necessary sensors and give a signal as soon as they detect constellations in the body’s functions that are relevant to impending attacks.
From the current perspective, however, this is still in the future. “After the one-year study has been completed, an international project is initially planned to scientifically validate the results,” explains Narzt.
In numbers
50,000 people in Austria suffer from various forms of epilepsy, the most common chronic disease of the nervous system worldwide. For many it occurs for the first time in childhood: statistically speaking, one in around 1,700 children becomes ill. Numerous new cases also occur from the age of 60 onwards.
250 affected are currently taking part in the study at the Linz University Hospital and the Johannes Kepler University, which aims to identify physical warning signals (biomarkers) of impending seizures.
1696776672
#Early #warning #system #epileptic #seizures #DiePresse.com
