In order to create robust signatures from complex biological data SimplicityBio brings a new technology, which is able to create signatures from single or multiple biomarkers (panel of biomarkers). We are able to test billions of hypothesis and biomarkers combinations in order to find the most accurate and robust signature. These biomarkers are selected in our high performance software platform for in-silico biomarker discovery.

This platform consists of a collection of mathematical methods and artificial intelligence techniques of data analysis that we developed and continuously adapted in a modular and highly configurable workflow. The core of this platform is our proprietary software tool FUGE-MM (Fuzzy Engine Modeling Manger) that implements the innovative technique: co-evolutionary fuzzy modeling (Fuzzy CoCo).

Why fuzzy logic in biomarker discovery?

Biomarker discovery usually applies data mining algorithms, such as support vector machines or logistic regression. In most cases, they produce models that are not easily interpretable by biologists and biomedical researchers, given the high number of variables and parameters. Fuzzy logic transforms complex models into simple but accurate rules making interpretation greatly improved. Equations involving several coefficients, interaction terms and constants are difficult to translate into biological relevance, whereas rules such as “if gene A is upregulated and gene B is downregulated, then the probability of disease X is high” becomes more obvious.

Why multiple biomarkers?

The advantages of a multi biomarker approach in comparison to a single biomarker assay are based on the premise that the single-valued index, with its aggregated information from complementary biomarkers, will outperform each of its component biomarkers used individually.

The ability of multivariate models to capture complex patterns in high-dimensional data also means that non-disease related artifacts that happen to confound the samples used to train the models will also be captured.

The inclusion of biomarkers in a diagnostic test requires that they are complementary, and that they collectively outperform a single marker with respect to the test’s intended use.

Robustness and reproducibility

The reproducibility requirement on biomarkers means that the identified group of biomarkers is always expected to exhibit good performance at distinguishing cases from controls across different studies. In our tool we provide methods that are used to find robust subsets of the most discriminative biomarkers, which are expected to produce reproducible results.

Biomarker discovery by using different data sources

Complex data is being generated all the time, from different instruments or sources and no insights are being obtained from it. Our algorithms are able to use several information sources (e.g. omics data, clinical data) in order to use them in a holistic manner for the discovery of robust signatures.

Multi-biomarker multi-model approach

In order to improve the robustness of the results we use an approach called multi-model. We develop combinations of models to produce a final signature. This “modular redundancy” approach allows the correction and/or compensation of a model that is not deemed efficient by other models; thereby it improves considerably the reliability of the final answer. This concept can be summarized as “multi-biomarker multi-model”.