<img src="https://ws.zoominfo.com/pixel/eCRg8hzy2qp0kNUl6E1g" width="1" height="1" style="display: none;">

Machine Learning

Our proprietary machine learning platform builds on recent advances in the artificial intelligence field to address the unique needs of clinical diagnostics.


The practice of medicine revolves around information-based decision making. In the fields of cancer biomarker and drug development research, this information is becoming increasingly rich with the advent of new technologies. Scientists’ ability to make sense of this wealth of healthcare data and distill it down to simple hypotheses to inform clinical decisions will inevitably become more and more difficult.

Artificial intelligence is defined as the theory and development of computer systems able to perform tasks normally requiring human intelligence, such as visual perception, speech recognition, decision-making, and translation between languages1. Machine learning is a subset of artificial intelligence, where the machine has the ability to acquire its own knowledge by extracting patterns from data2. The application of machine learning in healthcare has the potential to uncover clinically-relevant patterns and meaning in our healthcare data; helping us better decipher complex diseases like cancer.

We use the Diagnostic Cortex® platform, which is based on modern machine learning techniques to design tests that are reproducible, robust, and answer critical clinical questions for our partners. Based on data found in the circulating proteome and genome, we design tests that support treatment decisions including patient selection for immunotherapies, novel therapy combinations, or alternative treatment pathways.

"The accurate prediction of a disease outcome is one of the most interesting and challenging tasks for physicians. As a result, machine learning methods have become a popular tool for medical researchers. These techniques can discover and identify patterns and relationships between them, from complex datasets, while they are able to effectively predict future outcomes of a cancer type.”

- Kourou, K. "Machine Learning Applications in Cancer Prognosis and Prediction". CSBJ, 2015.

Machine learning has revolutionized many industries with voice and image recognition, search optimization, e-commerce recommendations, self-driving cars, and many more applications. At Biodesix, we are transforming the healthcare industry by using machine learning to design molecular diagnostic tests that answer meaningful clinical questions.

Our platform has been used to design 46 candidate tests to date.

Both traditional and modern machine learning approaches need large datasets to generate useful tests. Unfortunately, biomarker datasets are small and each patient or instance has many attributes or biomarkers. This challenge tends to lead to data overfitting and is called the p>N problem. Our proprietary Diagnostic Cortex® platform takes in to consideration the p>N problem, minimizing overfitting leading to a higher likelihood of a test generalizing or validating in a separate dataset.



Key Benefits of Our Approach

  • Can be used with small clinical trial sample sets (N>40)
  • Minimizes risk of overfitting to improve the chance that tests will validate in independent cohorts
  • Fits the clinical demand for meaningful tests that drive treatment decisions

Learn more about the Diagnostic Cortex

Get in touch

Contact us

Data library

Discover more