[Paper Review] Machine Learning Techniques for Predicting the Short-Term Outcome of Resective Surgery in Lesional-Drug Resistance Epilepsy

Zahra Jourahmad, Jafar Mehvari Habibabadi+5 more|arXiv (Cornell University)|Feb 10, 2023

Epilepsy research and treatment4 citations

TL;DR

This study develops and evaluates machine learning models to predict short-term surgical outcomes in patients with lesional, drug-resistant epilepsy using noninvasive clinical and demographic data. Using a support vector machine with a linear kernel on ten key features, the model achieved 76.1% accuracy, 96.7% recall in temporal lobe epilepsy, and 79.5% in extratemporal cases, demonstrating strong predictive potential for personalized surgical planning.

ABSTRACT

In this study, we developed and tested machine learning models to predict epilepsy surgical outcome using noninvasive clinical and demographic data from patients. Methods: Seven dif-ferent categorization algorithms were used to analyze the data. The techniques are also evaluated using the Leave-One-Out method. For precise evaluation of the results, the parameters accuracy, precision, recall and, F1-score are calculated. Results: Our findings revealed that a machine learning-based presurgical model of patients' clinical features may accurately predict the outcome of epilepsy surgery in patients with drug-resistant lesional epilepsy. The support vector machine (SVM) with the linear kernel yielded 76.1% in terms of accuracy could predict results in 96.7% of temporal lobe epilepsy (TLE) patients and 79.5% of extratemporal lobe epilepsy (ETLE) cases using ten clinical features. Significance: To predict the outcome of epilepsy surgery, this study recommends the use of a machine learning strategy based on supervised classification and se-lection of feature subsets data mining. Progress in the development of machine learning-based prediction models offers optimism for personalised medicine access.

Motivation & Objective

• To develop a machine learning-based presurgical prediction model for epilepsy surgery outcomes in lesional, drug-resistant epilepsy.
• To evaluate the performance of multiple supervised classification algorithms using noninvasive clinical and demographic features.
• To identify the most effective machine learning technique and feature subset for accurate outcome prediction.
• To support personalized medicine by enabling early, data-driven surgical decision-making.

Proposed method

• Seven different categorization algorithms were trained on a dataset of clinical and demographic features from epilepsy patients.
• The Leave-One-Out cross-validation method was used to evaluate model robustness and generalization.
• Feature selection was applied to identify the most predictive subset of clinical variables.
• Performance was assessed using standard metrics: accuracy, precision, recall, and F1-score.
• A support vector machine (SVM) with a linear kernel was selected as the optimal model based on performance.
• The final model used ten key clinical features to predict surgical outcomes in both temporal and extratemporal lobe epilepsy.

Experimental results

Research questions

• RQ1 Which machine learning algorithm performs best in predicting short-term outcomes after resective surgery in lesional, drug-resistant epilepsy?
• RQ2 How accurately can noninvasive clinical and demographic features predict surgical outcomes using supervised classification?
• RQ3 What is the optimal subset of clinical features that maximizes prediction accuracy for epilepsy surgery outcomes?
• RQ4 Can machine learning models improve personalized surgical decision-making in epilepsy patients?

Key findings

• The support vector machine (SVM) with a linear kernel achieved the highest accuracy of 76.1% in predicting surgical outcomes.
• The model demonstrated 96.7% recall in temporal lobe epilepsy (TLE) patients, indicating strong sensitivity to positive outcomes.
• For extratemporal lobe epilepsy (ETLE), the model achieved 79.5% recall, showing consistent performance across epilepsy types.
• The use of feature subset selection significantly improved model performance and interpretability.
• The study confirms that supervised machine learning on clinical data can reliably predict short-term surgical outcomes.
• The results support the integration of machine learning into clinical workflows for personalized epilepsy surgery planning.