Cancer Researchers Mine Big Data To Individualize Treatment
CancerLinQ prototype from American Society of Clinical Oncology aggregates data from disparate EHRs to find the most effective therapy for a specific patient.
7 Big Data Solutions Try To Reshape Healthcare
(click image for larger view and for slideshow)
The American Society of Clinical Oncology (ASCO) is harnessing big data to build and mine a database of cancer treatment records to help physicians determine the best treatments for particular kinds of patients. A just-completed prototype of the program aggregates data on more than 100,000 breast cancer patients from 27 oncology practices, many of which use different electronic health record (EHR) systems.
What ASCO is trying to do in its CancerLinQ project is create a "learning health system" that will collect and analyze cancer care data from millions of patient visits. By combining that with expert guidelines and other evidence, ASCO hopes to deliver "real-time, personalized guidance and quality feedback for physicians," according to a press release.
ASCO's prototype links together several open source and proprietary IT applications to do the following:
-- Accept cancer care data directly from any EHR system, as well as other sources such as lab data, genomic profiles and physician notes.
-- Provide clinical decision support to help physicians care for patients with breast cancer, based on automated versions of ASCO's breast cancer guidelines.
-- Enable researchers to explore the database to identify real-world trends and associations.
-- Provide accelerated feedback on physician performance against 10 quality measures from ASCO's Quality Oncology Practice Initiative.
According to ASCO, CancerLinQ can exponentially increase the amount of knowledge about patients' responses to treatment. "Today we know very little about the experiences of most people with cancer because their information is locked away in unconnected servers and paper files," said Sandra M. Swain, MD, president of ASCO, in the news release. "Only the three percent of patients who participate in clinical trials are able to contribute to advances in treatment. CancerLinQ will transform cancer care by unlocking that wealth of information and enabling every patient to be a cancer knowledge donor."
In a webcast of the CancerLinQ press conference, Swain elaborated that, when the program is further developed, it will be able to provide point-of-care guidance to physicians, using the best available evidence coupled with the experience of millions of cancer patients. "It's like having the whole medical community available for a second opinion. It redefines what a second opinion is."
Dr. Clifford Hudis, president-elect of ASCO and chief of the breast cancer medicine service at Memorial Sloan Kettering Cancer Institute, gave a demonstration of the prototype at the news conference. He noted that the researchers had to figure out how to aggregate and normalize data from disparate electronic health records such as Varian, Epic, Altos and Impac. (All of these products except Epic are designed specifically for oncology.)
"Our goal was to show that it was possible to create new ways for researchers to explore correlations between different cancer data that might be used to inform real-world decisions as well as generating hypotheses for further research," he said. He added, "This can also allow physicians to share information with other members of their clinical team."
Hudis cited a hypothetical breast cancer patient who'd had a lumpectomy and was taking a hormone-blocking drug because she had a hormone-sensitive form of cancer. Then he showed how an application called Galileo Cosmos could be used to select patients who fit the profile and were taking similar medications. Using a standard Kaplan-Meier plotter, he was able to demonstrate within seconds that the real-world patients in the database who'd kept taking the drug had gone longer without a recurrence of cancer.
Using another application called PopHealth, Hudis demonstrated how to find out whether all of the patients receiving hormone blockers were getting the right therapy for their disease profile and whether anybody who should have been taking one of those drugs wasn't.
"This shows that the estrogen receptor data came in from various records and the tamoxifen and aromatase inhibitor data came in from various records, and the system was able to figure out whether the prescriptions were written and whether the patient fit the category where it should be done. That means it's possible to do this, given the right parameters," he said.
Swain said ASCO plans to broaden the use of CancerLinQ to all types of cancer, and she noted that the treatment of other kinds of diseases could also benefit from a similar use of big data techniques.
Big data has already made a fairly good-sized beachhead in oncology. The University of Texas M.D. Anderson Center, for example, has launched a "moon shots" program to find cures for eight types of cancer, using a big data approach to incorporate genomic sequencing and computational algorithms. Los Angeles-based NantHealth is using supercomputers to sequence genomes and analyze the data so oncologists can tailor treatments to their patients. And it was recently announced that IBM's Watson supercomputer would help Memorial Sloan Kettering physicians diagnose and treat cancer.
We welcome your comments on this topic on our social media channels, or [contact us directly] with questions about the site.