Imagine a world where seizures are no longer misdiagnosed, and patients receive the right treatment from the start. That’s the bold vision driving groundbreaking research at the University of Pittsburgh School of Medicine, where clinician-scientists are harnessing the power of artificial intelligence (AI) to revolutionize seizure care. Published in Epilepsia, their latest findings reveal how AI can act as a crucial ally for clinicians, particularly those outside neurology, by identifying complex cases of functional seizures—a condition often mistaken for epilepsy. But here’s where it gets controversial: while AI doesn’t outperform expert neurologists, it levels the playing field for less specialized clinicians, raising questions about how technology should complement, not replace, human expertise.
Functional seizures, affecting roughly 10% of people with seizures, are a prime example of how the body’s response to stress can manifest in unexpected ways. Unlike epileptic seizures, which stem from abnormal brain electrical activity and respond to medication, functional seizures are rooted in a mix of biological, psychological, and social factors. And this is the part most people miss: these seizures are essentially the body’s ‘panic without panic,’ a hidden stress response that bypasses conscious awareness. As lead researcher Wesley Kerr, M.D., Ph.D., explains, ‘Your body is having a panic attack, but your brain shields you from feeling it—until it erupts as a seizure.’
To test AI’s potential, the team presented 117 anonymized patient cases to 163 reviewers—ranging from non-clinicians to expert epileptologists—and measured their diagnostic accuracy before and after AI assistance. The results? AI improved accuracy for 66% of reviewers, with average and AI-literate users seeing a nearly 20% boost in diagnosing functional seizures. Interestingly, experts saw little improvement, as the AI largely mirrored their existing knowledge. ‘AI won’t make experts better,’ Kerr notes, ‘but it ensures that even clinicians unfamiliar with functional seizures have a tool that prompts them to consider it.’
Looking ahead, Kerr’s team is exploring AI-assisted video analysis of seizures, which could reduce reliance on inpatient monitoring—the current gold standard. Their ultimate goal? An automated system that flags potential functional seizure cases within patient charts, paired with clinical oversight to ensure ethical and compassionate care. This could be a game-changer for pregnant, elderly, or medically complex patients, who often face unnecessary medication risks.
But here’s the provocative question: As AI becomes more integrated into healthcare, how do we balance its potential to improve access and accuracy with the need for human judgment and empathy? Kerr’s work suggests AI isn’t about replacing clinicians but expanding their impact—yet it’s a fine line to tread. What do you think? Is AI a tool for empowerment or a slippery slope toward dehumanized care? Share your thoughts in the comments below.
To dive deeper into this research or explore seizure care options, visit UPMC Epilepsy Center. And if you suspect your seizures might be functional, don’t hesitate to reach out for a consultation. Together, we can close the diagnostic gap and transform lives.
