Can AI help diagnose autoimmune encephalitis?

Researchers in Germany and Switzerland have developed a ground-breaking artificial intelligence (AI) system that can predict whether patients with suspected autoimmune encephalitis will test positive for disease-specific antibodies.

The AI system, details of which were published in Scientific Reports in March this year, could dramatically speed up diagnosis of this serious brain condition.

Diagnosing autoimmune encephalitis

Autoimmune encephalitis occurs when the body’s immune system mistakenly attacks the brain, causing inflammation that leads to seizures, memory problems, and altered mental states.

Early treatment can significantly improve outcomes, but this requires rapid diagnosis. Currently, however, diagnosis relies on lengthy antibody testing that can cause delay.

Speeding up diagnosis

The new AI system analyses subtle patterns in magnetic resonance imaging (MRI) brain scans that are largely invisible to the human eye, focusing specifically on the hippocampus – a brain region frequently affected in autoimmune encephalitis.

The system achieved remarkable accuracy rates of 89% in predicting antibody status, potentially allowing doctors to begin treatment days or weeks before laboratory results are available.

The scientists tested their AI model on 98 patients, whereupon it was found to reliably distinguish between those who tested positive versus negative for autoimmune antibodies.

Potential global impact

This non-invasive approach could be particularly valuable in resource-limited settings where specialised laboratory testing may not be readily available.

The system represents a new frontier in precision medicine, where AI can assist doctors in making faster, yet more accurate, diagnoses based on standard medical imaging that’s already routinely performed.

A promising future

While larger studies are needed to validate these results, this research demonstrates how AI can transform diagnosis in a neurological setting.

As techniques developed from AI continue to emerge, we look forward a brighter future where conditions affecting the brain can be detected and treated faster and more accurately.

To view the full paper, click here.