Closing Case 2 POM Big Data and the Treatment of Cancer aggregate them, and provide them to physicians, who can use the data The Problem to make better decisions about how to treat their patients. In theory, electronic medical records (EMRs) were supposed to The global statistics on cancer are sobering. Every year, 8 million make such data aggregation and integration easier. Unfortunately, people die from cancer, and 14 million people discover they have the those benefits have not totally materialized. In fact, more than 25 per- disease. Approximately $100 billion is spent on cancer drugs globally. cent of U.S. medical records remain in hard-copy format Unfortunately, the majority of cancer treatments are not successful. The Flatiron founders spent more than two years building what Despite years of effort by the medical establishment to persuade they call a data model, which is their strategy to organize clinical doctors and hospitals to embrace electronic medical records (EMRs), information into categories. Working with a team of physicians, they oncology data have remained difficult to access and use. (Oncology decided to focus initially on one type of cancer: colon cancer. Using is the branch of medicine that deals with the study and treatment of published clinical trials, they extracted more than 350 data categories, cancer.) including demographics, geographic location, cancer stages, biologi Data on a single cancer patient can come from multiple sources, cal markers of disease, and responses to therapies. Then they repeated including internists, oncologists, radiologists, surgeons, and labora- the process for other forms of cancer. tory and pathology reports. Even when the data are digitized, they To automate the process of extracting data from medical records, are often in an unstructured format. Rather than being organized in which can be labor intensive, Flatiron used matching algorithms databases, the data are often in multiple, inconsistent formats across targeted at pinpointing values in lab reports. They also utilized nat- different lab reports and records. Making matters worse, much data ural-language processing to enable computers to read documents remain hidden in reports that have been written by hand and scanned, and extract data from them. Such systems must be trained. To accom- in audio recordings, and in low-resolution PDF files printed from fax plish this task, Flatiron hired a team of 60 nurses to enter data on 500 machines. Finally, incompatible systems and strict privacy regula- patients by hand, creating a "training set" that was used to detect tions-for example, the Health Insurance Portability and Accounta- errors in data that had been collected automatically. Data collection bility Act, or HIPAA-that govern personal health information make it errors were then fed back into the system as inputs to help improve the even more difficult for data to be shared across thousands of oncology automated collection process. practices. Only a small fraction of cancer patient treatment data are being The Results collected systematically. Those data are typically collected from ran- Using Flatiron's OncologyCloud, oncologists are able to see the most domized clinical trials, which cover only 4 percent of adult cancer effective therapies for the most patients in similar circumstances. Fur- patients. ther, these physicians are able to evaluate their own treatment out- comes against those of other specialists across the nation and then A Proposed IT Solution quickly correct any deficiencies. The OncologyCloud also highlights Flatiron Health (Flatiron; www.flatiron.com) wants to help doctors cost-effective therapies and wasteful healthcare spending. Finally, the develop better treatment options for cancer. Founded in 2012, Flatiron system helps to match patients with suitable clinical trials, hopefully essentially fights cancer with organized data. With its OncologyCloud, speeding up the development and approval of new medicines. the company is helping oncologists enhance patient care. The com- In 2014, Flatiron acquired Altos Solutions, which developed the pany collects, organizes, and standardizes much of the information for first oncology-specific electronic medical record. This acquisition gave the 96 percent of patients not included in clinical trials and then offers Flatiron a larger installed base and closer contact with physicians. By those data back to physicians in a format that can be analyzed. October 2015 Flatiron systems were being used in 210 cancer centers The two Flatiron founders began their startup by visiting 60 can- that collectively see about 300,000 new patients every year. Further, cer centers, speaking directly with experts, and visiting patients with in 2014 Google invested more than $100 million in Flatiron via Google physicians. Working with oncology experts, they decided that the Ventures, the company's venture capital unit. most pressing need in cancer treatment was to organize the massive In 2015, Flatiron and Guardant Health (www.guardanthealth volumes of clinical data that are scattered in the filing systems of .com) began to collaborate to enable more effective cancer treatment. oncology treatment centers throughout the country. They proposed to Flatiron will provide the structure and all of the clinical trial informa- collect the data-both digital and otherwise-and then organize them, tion for the OncologyCloud. Guardant's commercially available cancer 90 CHAPTER 3 Data and Knowledge Management screening product, Guardant360, will be used to collect data from June 28, 2015; "Varian Medical Systems and Flatiron Health to Develop patients' blood samples in a much more efficient manner than was Next Generation of Cloud-Based Oncology Software," Flatiron Health previously possible. Press Release, May 26, 2015; "Foundation Medicine and Flatiron Health And the bottom line? In 2014, nearly 1.7 million Americans were Collaborate to Develop First In-Class Data Platform to Accelerate Precision newly diagnosed with cancer. If oncologists using the OncologyCloud Medicine for Cancer," Foundation Medicine, December 2, 2014; M. Helft "Can Big Data Cure Cancer?" Fortune, August 11, 2014; S. Baum, "Flatiron could improve the patient survival rate by 5 percent, they would save Health Finds Ideal Match with Duke Cancer Care Research Director," tens of thousands of lives every year. MedCityNews, July 2, 2014; K. Noyes, "Flatiron Health's Bold Proposition to Fight Cancer with Big Data," Fortune, June 12, 2014; N. Taylor, "Buzz: Sources: Compiled from L. Ramsey, "Cancer Treatment Is on the Brink Google Ventures Leading $100M Round in Oncology Big Data Platform," of a Data Revolution," Business Insider, September 22, 2015; N. Versel, FierceBioTech/T, May 5, 2014; www.flatiron.com, www.guardanthealth "GuardantHealth, Flatiron Health to Link Genomics, Analytics for .com, accessed August 26, 2015. Personalized Cancer Care," MedCityNews, August 19, 2015; C. Magee, "GuardantHealth and Flatiron Health Team Up to Cure Cancer with Big Questions Data," TechCrunch, August 19, 2015; "Fighting Cancer with Big Data," 1. Describe the Big Data issues in this case. The Rambus Blog, August 10, 2015; T. Stephens, "California Initiative to Advance Precision Medicine Funds UC Santa Cruz Pediatric Cancer 2. How does Flatiron use Big Data in its attempt to improve cancer Project," University of California at Santa Cruz News Center, August 3, 2015; treatment? B. Marr, "How Big Data Is Transforming the Fight Against Cancer," Forbes