Using Quantum Insights to Detect Disease Biomarkers

The heart of Quantum Insights' prediction process is an algorithm developed at the Stanford SLAC National Accelerator Laboratory as an extension of Dr. Weinstein’s research into quantum mechanics, Dynamic Quantum Clustering (DQC). DQC works by exploiting variations in the density of data to identify subsets of the data that exhibit significant multivariable correlations.  DQC’s approach to density mapping provides important analytical improvements for delivering precision medicine.  It is unsupervised, unbiased, and can build a functional map with as few as 50 observations.  These characteristics make DQC well-suited to the analysis of low volume, high-dimensional data that is common in biology (e.g RNA, SNP, metabolomic data, ATACseq, proteomics, etc.) and it regularly identifies structure that is invisible to other clustering methods. The DQC algorithm is described in 3 patents and several papers. At a high level, DQC represents data as multi-dimensional quantum objects (e.g. a tumor sequenced into 30,000 gene expressions is represented as a 30,000-dimensional Gaussian). Objects that are similar are located near each other. A map of the potential function is constructed where the areas of greatest density are pits and valleys.  The algorithm moves objects “downhill” along the potential function towards areas of greater density, forming clusters and exposing correlations that would be otherwise invisible. The result of a DQC analysis is an animation where each frame is an iteration through the algorithm and the distance between points is a measure of their correlation.  Unlike other algorithms, clusters can be spherical in shape or multi-dimensional, expressing parameter-izable relationships between variables.  Once clusters have been identified, we can feature select down to a smaller number of dimensions (if needed) and output the parameters to a prediction API that can be used classify new data.