In Brief | Neurology | Artificial Intelligence
Revolutionising Epilepsy Diagnosis: How Machine Learning Enhances EEG Accuracy
Time to read: 02:46
Time to listen: 05:55
Originally Published: 25 January 2025
Source: Annals of Neurology
Type of article: In Brief
MedED Catalogue Reference: MNIB011
Category: Neurology
Cross Reference: Artifical Intelligence
Keywords: AI, LLM, Epilepsy
EpiScalp a software analytics tool, significantly enhanced epilepsy diagnosis by using advanced network-based EEG analysis, significantly reducing misdiagnosis and improving diagnostic accuracy from a single routine EEG.
This article is a review of recent studies originally published in the Annals of Neurology, 25 January 2025. This article does not represent the original research, nor is it intended to replace the original research. Access the full Disclaimer Information.
Watch the 60 Second Summary
(Please let us know if you like this format by clicking here)
Listen to Full Article
EpiScalp, a groundbreaking tool developed by researchers at Johns Hopkins University, could reduce epilepsy misdiagnoses by up to 70% by identifying hidden markers in routine EEGs.
Epilepsy, which affects approximately 60 million people globally, remains challenging to diagnose. While 8%-10% of individuals will experience a seizure in their lifetime, only 2%-3% will develop epilepsy. Misdiagnosis occurs in nearly 30% of cases, often leading to unnecessary treatments, side effects, and reduced quality of life.
Scalp EEG is central to epilepsy diagnosis, focusing on visual analysis of interictal epileptiform discharges (IEDs) to assess seizure risk. However, IEDs are sporadic, and EEG sensitivity ranges from 29%-55%, contributing to diagnostic uncertainty. While repeat EEGs improve sensitivity to 92%, they place a considerable financial and logistical burden on patients and healthcare systems.
To address this, the research team developed EpiScalp, a machine-learning model trained on dynamic network models to identify hidden seizure markers from a single routine EEG. The tool utilizes spectral and network-derived EEG features to differentiate between epilepsy and non-epilepsy.
Study Methodology
A logistic regression model was used to train the AI. Building on previous research, the team examined two novel EEG metrics—fragility and source-link relationships—derived from both ictal and interictal intracranial EEGs.
The dataset was split, with 90% (198 patients) used to train the model and the remaining 10% (20 patients) reserved for testing.
Findings
EpiScalp demonstrated exceptional diagnostic accuracy, significantly improving epilepsy identification compared to traditional EEG interpretation.
The tool reduced misclassification rates from 54% to 17%, correctly ruling out 96% of false-positive epilepsy diagnoses, thus preventing unnecessary treatments and anxiety.
In the testing cohort, EpiScalp accurately predicted diagnoses in 80% of cases. It achieved an area under the curve (AUC) of 0.940, with an overall accuracy of 90.4%, classifying nine out of ten cases correctly.
Its high specificity (96.3%) ensured accurate identification of non-epileptic conditions, while its sensitivity (83.5%) detected most true epilepsy cases. EpiScalp's probability threshold—values below 0.32 indicated a 92% likelihood of non-epilepsy, while values above 0.61 suggested high certainty of epilepsy—provided clinicians with a reliable tool for diagnosis.
For decades, epilepsy diagnosis has depended on repeated EEGs and time-consuming monitoring, often resulting in patient frustration and misdiagnoses. EpiScalp offers a data-driven, objective approach to analyzing EEGs, detecting subtle network disruptions characteristic of epilepsy that may not be visible through traditional methods. This innovation represents a paradigm shift, offering clinicians an AI-assisted model that enhances diagnostic confidence from a single EEG recording.
Conclusion
By improving specificity and sensitivity, EpiScalp reduces unnecessary anti-epileptic treatments and accelerates appropriate care for patients with true epilepsy. This breakthrough could transform clinical practice, minimizing diagnostic uncertainty and improving patient outcomes.
Importance of this study for South Africa
In low- and middle-income countries, people with epilepsy (PWE) face higher risks of poor health outcomes and premature death due to economic challenges and limited healthcare access. A significant global treatment gap exists, with rural areas in Sub-Saharan Africa experiencing up to 69% of the gap, linked to economic instability. Tools which can increase diagnosis, without significant cost, could impact positively on patient outcomes and reduced disease burden in these areas.1
References
1.Makhado L, Maphula A, Ngomba RT, Musekwa OP, Makhado TG, Nemathaga M, Rammela M, Munyadziwa M, Striano P. Epilepsy in rural South Africa: Patient experiences and healthcare challenges. Epilepsia Open. 2024 Aug;9(4):1565-1574. doi: 10.1002/epi4.12999. Epub 2024 Jun 17. PMID: 38884148; PMCID: PMC11296125.
Back to top
Disclaimer
This article is compiled from several resources researched and compiled by the contributor. It is in no way presented as an original work. Every effort has been made to attribute quotes and content correctly. Where possible all information has been independently verified. The Medical Education Network bears no responsibility for any inaccuracies which may occur from the use of third-party sources. If you have any queries regarding this article contact us
Fact-checking Policy
The Medical Education Network makes every effort to review and fact-check the articles used as source material in our summaries and original material. We have strict guidelines in relation to the publications we use as our source data, favouring peer-reviewed research wherever possible. Every effort is made to ensure that the information contained here is an accurate reflection of the original material. Should you find inaccuracies, out of date content or have any additional issues with our articles, please make use of the contact us form to notify us.
