Cluster validity index to determine the separation of inferred clusters, which can be readily incorporated with k-means, hierarchical, and mixture-model-based clustering algorithms.
Under review
Bayesian logistic regression with Dirichlet process mixture priors for vaccine safety analysis, improving estimation of association with rare adverse events via data-guided information sharing.
Under review
Manuscript Link
Presentation Link
Standard representation of findings in medical literature by significance and clinical direction, enabling evidence computing tasks.
Accepted to MedInfo 2021
Medical evidence Dependency (MD)–informed attention, an extension of Multi-head self-attention that incorporates medical domain knowledge into a neural model.
Published in JAMIA