An “Evolutionary Mathematical” engine handles billions of variables, integrating many modalities (all ‘omics, demographics, clinical data, labs, medications, vitals, EEG, EKG, sensors, and so on.)

Enables variables to interact in ways not possible with other analytic AI methods, modeling all physical processes (including biology, behavior, and environment. Finds linear and non-linear relationships between previously unknown variables.

Models are human understandable and consistently reproduce with exceptional accuracy.

Populations of Turing Machine Algorithms are randomly created. With no a priori feature selection, the algorithms find connections in the data.

Models must survive against each other during evolutionary “fitness training”, based on accuracy and desired sensitivity/specificity balance.

In every generation, the top 50% of algorithms survive, based on how accurately they predict the patients’ outcomes. The others are eliminated.

Algorithms that survive also have “children” made of part of each parent. The children are added to the population, and the cycle repeats.

The process repeats many times in parallel populations, optimizing ways to combine patient data and math to produce a model of the underlying biology and behavior.

The final model is simpler than AI, machine learning, or statistical models.

Like nature’s evolution, this approach is powerful, efficient, and flexible, but far faster.

NeuBio procures clinical circumstance and outcome data from NIH and other sources.

NeuBio’s data selection criteria prioritizes diversity across: sample types (i.e. blood, brain tissue), geography, demographics etc, ensuring heterogeneity.

NeuBio employs an agnostic quantitative approach to analyzing clinical data to identify the set of biomarkers that illuminate biological, environmental, and behavioral factors involved in driving disease causality and progression.

Biomarkers identified in discovery are prospectively validated in-silico against additional clinical datasets to ensure accuracy & reproducibility.

An analysis is conducted to identify a subset of biomarkers as potential drug targets.

This proprietary methodology has been validated by pointing to known targets & treatments currently in development.