Researchers in Germany have developed a program that quickly identifies which of the four subtypes of colorectal cancer the tumor belongs to. The program immediately recommends the most effective treatment.
Researchers at the Berlin Institute for Medical Systems Biology, reported by Science Daily, have decided to bring in artificial intelligence to treat colon cancer. The special platform is called Multi-omics Autoencoder Integration (maui).
In the context of mutations and gene expression, there are many types of colon tumors. The strategy and effectiveness of treatment depend on how accurately it is possible to diagnose a subspecies of a tumor. Unfortunately, analyzing the differences between tumors is quite difficult.
Unlike other traditional machine learning systems, the new algorithm of German scientists does not require the introduction of marked data and looks for patterns on its own.
The program reveals patterns in the genetic material of the tumor, which makes it possible to determine one of the four types of colorectal cancer.
In addition, thanks to the program, it turned out that one of the subtypes of tumors should be divided into two subspecies.
The program is not only accurate, but also operational. Thanks to it, diagnostics are faster than using other similar machine learning algorithms.
In particular, it took just three minutes instead of twenty to process 100 templates.
- say scientists.
Doctors can use maui as a “search engine” to quickly understand what type of tumor a patient’s sample is.
The algorithm not only classifies the tumor, but also offers optimal treatment methods.
In the future, the program can be taught to work with other types of cancer.