A Bayesian tool for epilepsy diagnosis in the resource-poor world: development and early validation

Area of Science:

• Neurology
• Medical Informatics
• Public Health

Background:

• The epilepsy treatment gap in resource-poor countries is significant, necessitating support for healthcare workers.
• Accurate diagnosis of epileptic seizures is crucial for effective epilepsy management.

Purpose of the Study:

• To develop a tool for estimating the probability of an episode being epileptic using Bayesian analysis of patient history.
• To aid non-specialist health workers in diagnosing epilepsy in resource-limited settings.

Main Methods:

• Collected data on episode characteristics from patients referred to epilepsy camps in Nepal.
• Calculated likelihood ratios (LRs) for each answer compared to clinical diagnosis.
• Developed and validated a diagnostic tool based on the most informative LRs.

Main Results:

• A tool was created using 11 informative questions and Bayesian principles.
• The tool demonstrated strong discriminatory power, with post-test probabilities for epilepsy ranging from 0.88 to 1.
• Post-test probabilities for non-epilepsy ranged from 0.07 to 0.42.

Conclusions:

• A Bayesian-based clinical tool can effectively estimate the probability of epileptic seizures.
• This tool has the potential to empower health workers in diagnosing epilepsy.
• Further validation in diverse populations and conversion to a mobile application are recommended.