In the last few years, tools like ChatGPT have brought artificial intelligence (AI) into the mainstream. While AI begins to shape different aspects of work and life, researchers are already using AI to make exciting advances in epilepsy care.
Using AI to improve epilepsy diagnosis and treatment was one of the key research priorities identified during our UK Epilepsy Priority Setting Partnership (PSP). As a result, it forms part of our strategic goal to advance research into epilepsy, through our ‘Enabling Technologies’ theme.
In this blog, we explore how AI could transform diagnosis and treatment – and showcase Institute-funded projects already driving these breakthroughs forward.
Improving epilepsy diagnosis
One of the key tools for epilepsy diagnosis is electroencephalography (EEG), to measure brain activity over time.
Spotting abnormal patterns in EEGs can be a difficult and time-consuming process. However, spotting subtle patterns is something that AI tools excel at. EEG software enhanced with AI is already supporting neurologists by flagging abnormal patterns that need a closer look.
AI is also supporting the analysis of videos – something that could one day be applied to the diagnosis of epilepsy in children. Dr Edmond Shu Lim Ho is receiving funding from the Institute to develop a tool that can help diagnose epilepsy in infants from videos taken on smartphones by their parents or carers.
As Dr Ho explains, “This study could be transformative for people affected by epilepsy, particularly infants and their families. The AI system could significantly reduce the time to diagnosis, allowing for earlier intervention and improved treatment outcomes.”
Making surgery more effective
Drug resistant epilepsy might be caused by lesions in the brain which, if removed by surgery, could help improve seizure control. The difficulty is that these lesions can be very difficult to spot on MRI scans.
AI could provide an extra pair of (virtual) eyes. Dr Sophie Adler, with support from the Institute, has been using AI to analyse MRI scans to help spot these lesions. Dr Adler and her team showed that their AI tool, called MELD Graph, was able to spot about two-thirds of lesions that neurologists had missed.
As well as spotting opportunities for surgery to improve seizure control, other AI tools could assist with predicting long-term effects of the surgery too.
Look out for Dr Sophie Adler’s article later this month, where she’ll explain more about her Institute-funded work.
Finding clues to help personalise medicine
We need more accurate methods of predicting whether a person’s epilepsy will be resistant (refractory) to anti-seizure medications. When it comes to research to develop such methods, clues might be hidden in existing medical records. However, gathering insights from these records is a challenge, partly because of the sheer number of them, but also because much of the information may be contained in ‘free text’ notes written by doctors.
What if an AI tool could rapidly read and analyse huge quantities of health records, to identify patterns which predict drug-resistant epilepsy? Prof Mark Richardson received funding from Epilepsy Research Institute and Angelini Pharma in 2022 to do just that.
“We have developed systems to extract highly detailed information from electronic health records, maintaining confidentiality and privacy of the information. We can analyse this information using artificial intelligence to track the course of epilepsy in thousands of people and make a much better attempt to identify the factors that predict refractory epilepsy,” said Prof Richardson.
Following this work, Prof Richardson is now collaborating with another team to take the idea further, with prestigious funding from the Medical Research Council. This partnership brings together clinicians, data scientists and AI experts at Swansea and King’s, as well as individuals with lived experience from Epilepsy Research Institute’s Shape Network.
Addressing the hazards of AI
As well as considering the possibilities, we need to be aware of the potential hazards posed by AI.
People with epilepsy are not all treated equally – and the data which AI learns from may contain these biases. We need to make sure that people are treated fairly, and that AI tools don’t amplify existing inequalities.
Health data contains some of the most sensitive information people will ever share. We must make sure that AI tools keep this information secure and private.
Finally, no AI tool is free from mistakes. AI tools must never be used to replace the judgement of human experts, but to support their decision making.
Bringing the future forward
AI is creating many exciting opportunities for epilepsy and other areas of healthcare. Thanks to this rapidly advancing technology, researchers today are making progress that maybe only a decade ago seemed impossible.
That’s why the Epilepsy Research Institute will continue to invest in research that uses AI effectively and responsibly – to bring forward a future where everyone can live seizure-free.
To hear more about how we are advancing the use of AI and other technology for people with epilepsy, sign up to our e-newsletter.
