The New York Genome Center is the latest healthcare organization to take advantage of the cognitive learning capabilities of the IBM Watson supercomputer.
During the past year, Memorial Sloan-Kettering Cancer Center has been feeding medical literature on cancer over three decades to Watson, teaching the computer how cancer has been traditionally treated without the benefits of whole genome sequencing. With the benefit of that knowledge, New York Genome Center and IBM are developing a new Watson prototype that will use its data analytics, natural language processing and retrieval technologies to deliver more personalized treatment plans to cancer patients based on their individual genetic makeup.
For 150 years, cancer has been categorized by where it shows up in the body, says Steve Harvey, global technology and analytics leader in IBM’s Global Technology Services unit. With the new prototype--built during the past year--New York Genome Center will be able to take biopsies of normal and cancerous cells and compare differences to understand what is causing cancer rather than it just showing up.
When comparing cancer and non-cancerous cells, the amount of differences can range from about 20,000 up to 1 million, Harvey explains. There are “driver” mutations causing cancer and “passenger” mutations showing the cancer. But only one to a handful of these mutations are the causal factors and researchers need to understand what drugs are best to use in treatment. The goal is to use Watson to reduce the current timeframe of about three weeks to be able to select the best drugs to a few minutes.
Also significant is the fact that in a handful of cases, literature has shown that a drug for a specific type of cancer also is beneficial in treating another type of cancer, Harvey says. Consequently, Watson will search to find more cases and determine how often they appear.
New York Genome Center will conduct a small clinical study with 20 to 25 patients and give them standard treatments, but at the same time sequence their entire genome to seek the drivers of causing and showing cancer, then search literature through Watson for appropriate drugs. If an individual patient is not responding adequately to the drugs, then clinicians can seek out other treatments and review additional information that Watson delivers.
This story originally ran on the Health Data Management website.