Clinical Decision Support Using a Partially Instantiated Probabilistic Graphical Model Optimized through Quantum Annealing: Proof-of-Concept of a Computational Method Using a Clinical Data Set

Abstract

AbstractAn approach is described to building a clinical decision support tool which leverages a partially instantiated graphical model optimized through quantum annealing. Potential advantages of such a strategy include the practical, and potentially real-time, use of multidimensional patient data to make a host of intuitively understandable predictions and recommendations in complex cases which are informed by a data-driven probabilistic model. Preliminary proof-of-concept of the general approach is demonstrated using a large well-established anonymized patient data set, revealing the predictive capability of a specific model. Ideas for future research are discussed.