CancerLinQ Prototype Demonstrates Use of Big Data in Clinical Care 

Get Permission

[The CancerLinQ prototype] was built to confirm that it’s possible to do and to show proof of principle, and it has.

—Sandra M. Swain, MD, FACP

ASCO has unveiled the prototype of a computer system that will allow oncologists, from their desks, to
leverage “big data” to measure the quality of care that their practices provide. The prototype is a major step in the development of CancerLinQ, a system that will eventually allow millions of electronic health records to interact with analytic software and clinical decision-making tools. The prototype was shown at the National Press Club on March 27 in conjunction with a panel discussion on big data in cancer care.

With more than 100,000 patient records entered by several different cancer centers around the country, “we’ve shown that we can compile and analyze data in any format, in real time,” said ASCO President Sandra M. Swain, MD, FACP, who is Medical Director of the Washington Cancer Institute at the MedStar Washington Hospital Center in Washington, DC. The prototype also showed that it was possible to provide decision support in the clinic and give physicians feedback related to quality practice.

Although CancerLinQ is a multiyear effort, this functional prototype is a major step forward. “It was built to confirm that it’s possible to do and to show proof of principle,” Dr. Swain said, “and it has.”

Rapid Learning System

CancerLinQ is a learning health system as described in a series of reports last year from the Institute of Medicine, with best practices seamlessly embedded in the delivery process and new knowledge captured as an integral byproduct of the delivery experience. Also known as rapid learning systems, these knowledge bases are envisioned as big-data systems that will not only assemble and analyze millions of records (with patient privacy protections), but will also grow “smarter” over time, as more records are added and analyzed.

Eventually the system will enable clinical learning from the experience of a wide range of patients, including the 97% who do not participate in cancer clinical trials, said Dr. Swain. The prototype, “in a way confirms our belief that every patient can be … a knowledge donor,” she said.

The system has four core functions, most of which were built with open-source software: real-time data collection, clinical decision support, data mining and visualization, and quality feedback.

Real-time Data Collection

The prototype contains de-identified data from 100,000 patients with breast cancer, and more are being added. The response from clinicians volunteering to participate in the prototype has been high, demonstrating that the project has the support of the community, noted Dr. Swain. Because the system accepts data from all the widely used electronic health record programs, transferring the records involved minimal effort and time, according to Charles Penley, MD, a medical oncologist at Tennessee Oncology, which participated in creating the prototype. Dr. Penley is also Chair of ASCO’s Conquer Cancer Foundation, which helped obtain funding for this early phase of development of CancerLinQ.

Clinical Decision Support

Using ASCO’s expert breast cancer guidelines, the system can provide individualized guidance on care. And when fully implemented, it will provide guidance based on the collective experiences of other patients with breast cancer. “For instance”, said Dr. Penley, “when physicians now are puzzled by a case, they may consult the literature, bring it up at a tumor board, and call colleagues whom they know personally. With CancerLinQ, they will also be able to query thousands of oncologists whose experience, collectively, includes millions of patients.”

Data Mining and Visualization

The system can show trends and associations. For instance a query about the duration of tamoxifen or anastrozole therapy in estrogen receptor–positive breast cancer generates a chart showing that appropriate patients in the prototype database took one of the drugs for up to 10 years. A query about outcomes generates a Kaplan-Meier plot showing that longer duration was associated with better outcomes. This specific example confirms clinical trial results, but it also shows how the system could be used to inform care where no clinical trial data are available, said panel member Clifford A. Hudis, MD, Chief of the Breast Cancer Medicine Service at Memorial Sloan-Kettering Cancer Center, New York, and ASCO’s President-Elect.

Quality Feedback

The prototype can provide immediate feedback on quality of care, using 10 measures from ASCO’s Quality Oncology Practice Initiative (QOPI). That program, now paper-based, allows physicians to measure their performance against established standards of care. CancerLinQ will automate the process and provide feedback in real time.

The prototype is “just the first step in what is certain to be a long process,” said Dr. Hudis. The results will be published over the next year as work continues. “We are still learning from the prototype and will figure out ways to increase our efficiency as we build out toward the full system,” he said. That will be created in modular fashion, like the prototype, and will include all types of cancer. ■

Disclosure: Drs. Swain, Penley, and Hudis reported no potential conflicts of interest.