Abstract: In one implementation, a computer-implemented method includes accessing patient-derived blood data; identifying biomarker pair interactions based on signal processing of the patient-derived blood data; generating a data model that characterizes one or more aspects of human immune system interactions based on reverse engineering of the human immune system interactions using the identified biomarker pair interactions; evaluating suitability of the data model, wherein the evaluating includes: statistically testing the data model in characterizing correlations between biomarker pairs; decomposing the data model to determine (i) model order accuracy, (ii) whether biomarker nodes act in together or opposite each other, and (iii) normalized data with dynamic effects of homeostasis removed; and quantifying degrees to which the data model accurately correlates patient treatments to patient outcomes; and providing the data model for use in treatment-based decision making based on the evaluating.

Abstract: In one implementation, a computer-implemented method includes accessing patient-derived blood data; identifying biomarker pair interactions based on signal processing of the patient-derived blood data; generating a data model that characterizes one or more aspects of human immune system interactions based on reverse engineering of the human immune system interactions using the identified biomarker pair interactions; evaluating suitability of the data model, wherein the evaluating includes: statistically testing the data model in characterizing correlations between biomarker pairs; decomposing the data model to determine (i) model order accuracy, (ii) whether biomarker nodes act in together or opposite each other, and (iii) normalized data with dynamic effects of and providing the data model for use in treatment-based decision making based on the evaluating.