Diagnostics of prostate cancer using Machine Learning
KEYWORDS: Cancer | Biomarker discovery | Machine Learning & AI
Prostate cancer is the 2nd most commonly diagnosed type of cancer among men worldwide. It is also the 5th leading cause of death due to cancer in this group. The current standard prostate cancer diagnostics rely on a prostate-specific antigen (PSA) blood test, which is unreliable. An incorrect diagnosis results in unnecessary biopsies, with side effects such as bleeding, infection, and urinary retention, and causes additional costs to the healthcare system.
The goal of our customer was to develop a new standard for prostate cancer diagnostics.
We applied AI/ML to analyze a panel of heterogeneous biomarkers that resulted from a new analysis of semen samples. The ensemble learning included Deep Neural Network, Gradient boosted based on decision trees etc. As a result, we provided an integrated approach to classifying different forms of prostate cancer.
Our customer was able to develop an automated diagnostics tool for prostate cancer (biopsy vs no biopsy) and assessment of the different forms of this cancer.