Skip to main content

Using Causal AI to Improve Diagnosis of Rare Epilepsies

Project identifier: SDE_TVS_Proj_143

Project title: Using Causal AI to Improve Diagnosis of Rare Epilepsies

Lay summary:

This project aims to improve care for people with Lennox-Gastaut Syndrome (LGS), a rare and severe form of epilepsy that begins in childhood but often goes undiagnosed or misdiagnosed into adulthood. If left undiagnosed or misdiagnosed, the condition often worsens over time, leading to increasing seizure burden, cognitive decline, behavioural issues, and greater functional impairment. 

Many patients with LGS require a distinct treatment regimen that differs significantly from therapies used for other forms of epilepsy. Some commonly used anti-seizure medications may be ineffective or even exacerbate seizures in LGS. Many children with LGS are not referred early enough to specialists who can offer the most effective treatments. In adults, evolving symptoms and loss of classic diagnostic features lead to frequent misclassification as chronic epilepsy, delaying access to appropriate therapies. 

This project will use causal artificial intelligence (AI) – a cutting-edge approach that goes beyond extrapolating correlations in patient data to leveraging the knowledge of causal interactions between patient variables (derived from literature, clinician knowledge and from statistical constraints in the dataset) to identify missed or delayed diagnoses and improve treatment matching in both children and adults. 

Public benefit statement: 

This study will help clinicians make earlier and more accurate diagnoses of LGS, leading to faster access to specialist care and appropriate treatments like Fintepla. 

In children, this may reduce seizures, hospitalisations, and developmental delays. In adults, recognising missed LGS diagnoses could enable targeted therapy adjustments, reduce inappropriate medication use, and improve quality of life.

For the NHS, earlier diagnosis means more effective use of specialist resources, fewer emergency admissions, and cost savings through better targeted treatment.

Organisation: Allos AI Ltd

Date of signed agreement: 02/04/2026

Health Research Classification System (HRCS) category: Neurological

Current project status: Live – data in use

Multiple-SDE project: No

Lead SDE: N/A

Participating SDEs/organisations: TVS SDE

Further reading: