Technology giant IBM has just released a group of artificial intelligence (AI) projects that are working on curing cancer. The projects have been released to the open source community, which means they can be worked on publicly and improved upon collaboratively. All those working on the project share their improvements with the rest of the public. The full scope of these projects will be explained by IBM at a conference in Switzerland later this month.
The Goal of the Projects
Cancer caused nearly ten million deaths last year, with almost 20,000 new cases being diagnosed. Cancer development can be influenced by both genetics and environmental factors like smoking and diet. While we understand risk factors about the disease, and many forms of it are now treatable, there is still a lot of research to be done.
IBM’s Biology Group, based in Switzerland, are hoping that AI can help the research community understand the molecular structure of cancer as well as tumor composition. By finding out more about the molecular level of these complex diseases, the researchers hope to help develop more effective therapies and treatments.
Project One: PaccMann
The first project IBM released is called PaccMann. This project is said to examine the neural network in the body to predict how each person might react to anti-cancer compounds and treatments. It works by analyzing the genes in the body and the chemical compounds in the treatment and figuring out which compounds are most likely to kill cancer strains.
One of the biggest problems with cancer treatment is that it is highly expensive. IBM hopes that the PaccMann algorithm can identify the anti cancer compounds earlier and thus reduce the cost of treatment because it will reduce the cost of drug development. Financial restraints are often what causes delays in innovative treatment types, and this project helps combat this problem.
Project Two: INtERAcT
The second project is called INtERAcT. This one works a bit differently. Its function is to automatically scan scientific papers and gather information in one place to better help our understanding of cancer. Just under 20,000 papers on cancer research are published every year, so it is very difficult to keep up with every aspect of cancer research. INtERAcT tries to make the research process easier by automatically compiling all the relevant data. It is currently being used to increase our understanding of protein-protein interactions, which many researchers think is key to finding cancer cures.
Project Three: PIMKL
The last project to be introduced is called pathway-induced multiple kernel learning, or PIMKL for short. This AI technology analyzes sets of data about the molecular interactions of cancer to predict how cancer will progress and the chance of relapse in a particular patient. It works by finding molecular pathways affected by cancer and categorizing each patient. This means that treatment plans could become individualized and hopefully more effective as a result.
The code for all three projects has been released through either the company’s website or the IBM cloud. All projects are open source and available in the public domain. By making the technology public, IBM is hoping to maximize its potential to help patients.