IBM Corp. has open-sourced three artificial intelligence projects focused on cancer research.

In a blog post today, IBM researchers Matteo Manica and Joris Cadow, who work with the company’s Computational Systems Biology group in Zurich, noted that cancer is the second leading cause of death worldwide, with an estimated 18.1 million new cases diagnosed in 2018 and 9.6 million deaths attributed to the disease in the same year.

With so many deaths, developing a cure for cancer has to be one of mankind’s highest priorities, and IBM’s team wants to use its expertise in AI and machine learning to help the cause. The idea with its new AI projects is to try to accelerate understanding of the “leading drivers and molecular mechanisms of these complex diseases.”

“Our goal is to deepen our understanding of cancer to equip industries and academia with the knowledge that could potentially one day help fuel new treatments and therapies,” IBM’s researchers said.

The first of the newly open-sourced tools is called PaccMann, a fun-sounding acronym that stands for what is actually a very serious project called “Prediction of anticancer compound sensitivity with Multi-modal attention-based neural networks.”

PaccMann is meant to aid in the development of cancer-treating drugs, IBM said. It’s an algorithm that’s designed to analyze chemical compounds and predict which ones might have cancer-fighting abilities, so that these can then be tested as possible medicines. The initiative could be helpful because it typically costs millions of dollars to develop a single cancer drug, not too mention years of research.

The second project is called IntERAcT, which stands for “Interaction Network infErence from vectoR representATions of words.”

The idea behind this tool is that it can automatically scan scientific research papers relating to cancer and extract and analyze data from them more quickly to boost the understanding of cancer. That’s important because about 17,000 papers on the subject are published every year, which means it’s impossible for researchers to keep up with everything.

IBM said IntERAcT is being used to extract data from papers on “protein-protein interactions,” which scientists believe could be responsible for the biological process disruptions caused by many cancers.

One notable strength of INtERAcT is its capability to infer interactions in the context of a specific disease, IBM said. Comparing them with normal interactions in healthy tissue could help provide insight into disease mechanisms.

Lastly, IBM has created a project called PIMKL, or “pathway-induced multiple kernel learning,” which is an algorithm that ingests data on our current knowledge of molecular interactions to predict how cancer progresses, and whether patients are likely to experience any relapses following successful treatment. Using this information, doctors might be able to come up with tailored treatment plans for individual patients.

Constellation Research Inc. analyst Holger Mueller said he welcomed IBM’s move to open-source the projects, which should gain much more traction as a result.

“Healthcare and cancer research and diagnostics can see massive upside with the adoption of modern technology,” Mueller said. “These truly next-generation applications can have a lasting impact on disease and mortality rates.”

IBM said the source code for each project has been released to the public domain for other researchers and academics to use.

Photo: ColiNOOB/Pixabay