Machine learning model can predict which anticonvulsants work best

August 29, 2022

2 minute read


Hakeem does not report any relevant financial information. Please see the study for relevant financial information from all other authors. Chiang does not report any relevant financial information and Rao reports personal expenses from NeuroPace.

We have not been able to process your request. Please try again later. If you continue to have this problem, please contact [email protected]

A deep learning model has the potential to predict an individual’s response to anticonvulsants and may inform treatment choice in patients with newly diagnosed epilepsy, researchers reported in JAMA Neurology.

Haris Hakeem, MDfrom the Department of Neuroscience at Monash University in Melbourne, Australia, and colleagues developed and validated a machine learning model using readily available clinical information to predict the treatment success of the first prescribed anticonvulsant drug given to patients.

Source: Adobe Stock.

According to the authors, it is currently not possible to predict which anticonvulsant drug will work best, and since most are tested through a trial-and-error approach, patients may experience several ineffective treatments before the right drug is found.

Hakeem and colleagues used clinical data from patients treated with anticonvulsant drugs from 1982 to 2020, all of whom were followed for at least 1 year or until the first anticonvulsant drug failed. A total of 2,404 adults with epilepsy treated at five specialist centers in Scotland, Malaysia, Australia and China were considered for inclusion, with the final cohort comprising 1,798 adults (mean age, 34 years; 54.5% women).

To train and test the model, the researchers used 16 clinical factors and information from seven anticonvulsant drugs. Treatment success was defined as complete absence of seizures during the first year of treatment while taking the first drug.

The researchers included five independent cohorts and initially pooled all cohorts for model training and testing. The model was trained a second time using the largest cohort and validated externally on the other four cohorts. The researchers compared the performance of their deep learning model, known as the transformer model, with other machine learning models.

The transformer model that was trained with the entire pooled cohort had an area under the receiver operating characteristic curve (AUROC) of 0.65 (95% CI, 0.63-0.67) and a weighted balance accuracy of 0.62 (95% CI, 0.60-0.64) on the test set. The AUROCs of the model trained using the largest cohort ranged from 0.52 to 0.60, with a weighted balanced precision of 0.51 to 0.62 in the external validation cohorts. The transformer model that was developed using the pooled cohort performed better than two of the other five models tested.

“Although its current performance is modest, our model serves as a foundation for further improving clinical applicability and comparison with expert consensus decision-making algorithms,” the authors wrote. “Because our model can predict an individual’s response to a range of [anti-seizure medications]it can potentially be used as a clinical decision support tool to inform drug selection.

In a related editorial, Sharon Chiang, MD, Ph.D.and Vikram R. Rao, MD, PhD, both of the Weill Institute for Neurosciences at the University of California, San Francisco, noted that the study provides a “useful assessment”, but still question whether the model “beyond the intuition of experienced clinicians”.

“The fact that two-thirds of people with epilepsy currently achieve seizure control with pharmacotherapy using only clinician experience suggests that further improvements will be needed before [machine-learning] personalized prediction methods of [antiseizure medication] response are considered ready for prime-time clinical practice,” they wrote.


Chiang S, et al. JAMA Neurol. 2022; doi:10.1001/jamaneurol.2022.2441.

About Michael Bill

Check Also

Maltese Fibromyalgia and M/E Activist Shows How Much Medication She Needs to Function

Ruth Debono, Malta’s fibromyalgia leader and M/E campaigner, showed people how much medicine she needs …