Chances of treatment success were improved if patients received the model-predicted treatment. Using the model's prediction system may enable personalized, evidence-based epilepsy care, accelerating the match between patients and their ideal therapy, thereby delivering significantly better health outcomes for patients and providing health-care savings by applying resources more efficiently. Our goal will be to strengthen the predictive power of the model by integrating diverse data sets and potentially moving to prospective data collection.

Objective: To determine whether information in medical and pharmacy claims data can predict, at the time of prescribing the first antiepileptic drug (AED), which patients with epilepsy will become resistant to AEDs. Method: We analyzed longitudinal claims data from 1,376,756 patients with epilepsy from 2006 to 2015. Of these, 582,258 satisfied all inclusion criteria; 49,916 were ultimately AED resistant, operationally defined as a patient with claims filed for at least 4 distinct AEDs. We constructed 1,270 candidate predictors (“features”) reflecting demographics, comorbidities, medications, procedures, epilepsy status, and payer status to characterize the cohort. On the training dataset (528,640 patients) we performed ANOVA F-value tests to select predictive features and trained several prediction algorithms, including logistic regression, support vector machines (SVM), and random forests. A model with only age and gender was used as a benchmark model. Results: On a held-out test set (53,618 patients), the best model achieves an area under the receiver operating characteristic (ROC) curve (AUC) [95% CI] of 0.753 [0.747, 0.759], compared to 0.664 [0.658, 0.671] for the benchmark model. Moreover, predicted probabilities for drug resistance closely match the observed frequencies. Compared to waiting for 2 AED failures, our model predicts drug resistance on average 2.25 years earlier. Conclusion: Predictive models created from large claims data using machine learning methods can accurately predict which patients with epilepsy will prove drug resistant at the time of prescribing the first AED. The ability to predict refractoriness may help patients consider alternative therapies earlier in the course of their epilepsy.

Accurate and individualized prediction of response to therapies is central to precision medicine. However, due to the generally complex and multifaceted nature of clinical drug response, realizing this vision is highly challenging, requiring integrating different data types from the same individual into one prediction model. We use the anti-epileptic drug brivaracetam as a case study and combine a hybrid data-/knowledge-driven feature extraction with machine learning to systematically integrate clinical and genetic data from a clinical discovery dataset (n = 235 patients). As such, we construct a model that successfully predicts clinical drug response (AUC = 0.76) and show that even with limited sample size, integrating high-dimensional genetics data with clinical data can inform drug response prediction. After a further validation on data collected from an independently conducted clinical study (AUC = 0.75), we extensively explore our model to gain insights into the determinants of drug response, and identify various clinical and genetic characteristics predisposing to poor response. Finally, we assess the potential impact of our model on clinical trial design and demonstrate that, by enriching for probable responders, significant reductions in clinical study sizes may be achieved. To our knowledge, our model represents the first retrospectively validated machine learning model linking drug mechanism of action and the genetic, clinical and demographic background in epilepsy patients to clinical drug response. Hence, it provides a blueprint for how machine learning-based multimodal data integration can act as a driver in achieving the goals of precision medicine in fields such as neurology.

No abstract

No abstract