Product key features
Indication agnostic, unbiased machine learning
Any data type (biological, clinical, digital) from any technology
Handles unbalanced and sparse datasets
Supports time-series, longitudinal data
Human-interpretable and robust models
Patented filtering and selection methods
Handles missing values (with or without imputation)
Data validation for real world deployment
Providing value at different phases
In-silico biomarker discovery and biomarker ranking
The goal is to identify biomarkers (e.g. secreted proteins in the blood or urine) to be used for diagnostics, patient stratification, follow-up or treatment response of certain diseases. SimplicityBio uses its proprietary software platform to rank potential biomarkers from the most relevant to the less relevant.
Real application example: identification and ranking of secreted biomarkers in the blood or urine to be used for diagnostics, follow-up and treatment response of a certain medical condition. These biomarkers needed to be novel, for patent purposes.
Model creation (biomarkers + classification algorithm)
SimplicityBio’s software platform could have as input omics data from healthy patients and from disease patients. Based on this input our platform is able to learn which biomarker models can precisely classify a new patient in one of the groups of patients used as example (e.g. healthy or with a disease). The created model (i.e. biomarkers and the associated rules) is used to classify new patient’s, with certain disease, or subtypes of a disease.
Real application example: identify, from approximately 13000 genes, 270 biomarkers and develop the models able to discriminate 9 different medical conditions.
Improvement of an existent model (sensitivity, specificity and robustness)
Molecular diagnostic tests must reach certain sensitivity and specificity thresholds set by regulatory bodies in order to reach the market (e.g. CE IVD mark, FDA approval). Our technology is used to help companies to improve their models in terms of sensitivity, specificity and robustness.
Real application example: decrease the number of biomarkers for the diagnostic of a certain medical condition from more than 150 to less than 30 biomarkers, while at the same time increasing sensitivity and specificity.