KG-LIME: predicting individualized risk of adverse drug events for multiple sclerosis disease-modifying therapy
Authors: Jason Patterson, Nicholas Tatonetti
Presenter
Statement of Purpose
Multiple sclerosis (MS) is a debilitating chronic inflammatory disease affecting the central nervous system, with significant implications for millions worldwide, particularly young adults. Current treatments for MS, known as disease-modifying therapies (DMTs), are aimed at reducing relapse rates and delaying disease progression. However, these therapies, especially those with high efficacy, often come with severe adverse effects, including immunosuppression and increased susceptibility to infections. The need for effective pharmacovigilance systems to monitor and predict these adverse effects is crucial, yet existing methods, such as spontaneous reporting systems and clinical trials, face limitations such as reporting biases and insufficient contextual details.
This work advances the field by integrating real-world evidence from electronic health records (EHRs) to enhance pharmacovigilance for MS DMTs. We construct an adverse drug event (ADE) biomedical knowledge graph (KG) and develop a deep graph convolutional network (GCN) model to predict ADEs by leveraging sequences of EHR data. This approach addresses current limitations by incorporating time-awareness in the model and using a KG for dynamic feature selection and causal explanation. Our modified local interpretable model-agnostic explanations (LIME) framework, knowledge graph-LIME (KG-LIME), further enhances the interpretability of predictions, offering insights into causality and improving the reliability of ADE prediction. This work represents a significant step toward more accurate and individualized pharmacovigilance in MS treatment.
Learning Outcomes
- Leverage Biomedical Knowledge Graphs: Participants will learn how to construct and use biomedical knowledge graphs to enhance feature selection and explanation in predictive modeling for drug safety.
- Evaluate and Address Limitations in Current Pharmacovigilance Systems: Participants will be equipped to critically assess and address the limitations of traditional pharmacovigilance methods, such as spontaneous reporting systems and clinical trials, by integrating real-world EHR data.
Format
35-minute presentation by article author(s) considering salient features of the published study and its potential impact on practice
25-minute discussion of questions submitted by listeners via the webinar tools and moderated by JAMIA Student Editorial Board members.
CME Credit
The American Medical Informatics Association is accredited by the Accreditation Council for Continuing Medical Education (ACCME) to provide continuing medical education for physicians.
The American Medical Informatics Association designates this live activity for a maximum of 1.0 AMA PRA Category 1™ credits. Physicians should claim only the credit commensurate with the extent of their participation in the activity.
CNE Credit
The American Medical Informatics Association is accredited as a provider of nursing continuing professional development by the American Nurses Credentialing Center's Commission on Accreditation.
- Approved Contact Hours: 1.0 total
- Nurse Planner: Jenna Thate, PhD, RN, CNE