Big Data, Better Health Tracking

Healthcare Robotics: Patently Incredible Inventions

More often than not, clinical trials fail. They usually don't make it past phase two, the stage at which drugs are first tested on patients. Pharmaceutical companies spend billions of dollars on failed projects, which reach dead ends largely because of problems recruiting and retaining trial participants.

More than 80% of clinical trials in the U.S. fail to meet enrollment timelines and are stopped early because they can't enroll or retain enough patients throughout the trial in order to develop a sufficient data set, according to the National Institutes of Health.

Enter big data, which is quickly emerging as an answer to this high failure rate by streamlining the clinical trial process and matching patients to trials more effectively.

U.S. pharmaceutical giant Merck recently announced a partnership with Maccabi, the second largest health organization in Israel, to use their data to track disease progression and medication effectiveness over a patient’s lifespan. The multi-year agreement is one of several initiatives that show the importance of big data in managing disease treatment.

[ Can Google extend our lives? Read What Google's Calico Means For Healthcare.]

Maccabi was of particular interest to Merck because of its history collecting patient data -- it has been using electronic health records for more than 20 years, painting a longitudinal picture of the evolution of patients' health over lifetimes. Israel's healthcare structure also makes Maccabi a good fit. Maccabi is one of the four Israeli health organizations that were created under the country's 1995 National Health Insurance Law, meaning Maccabi's patient base represents nearly 25 percent of the Israeli population.

Another advantage is Maccabi's Israeli origins, or rather the fact that it's not a U.S. company.

"There's reticence in the United States to give anyone access to any data," said Ira Kalina, a partner at Drinker Biddle and Reath, a U.S. law firm that has a healthcare division. "It would be a PR nightmare."

All of the patient data is de-identified. As a result, Merck and Maccabi didn't need patients' permission to use their data.

"Part of the goal is to understand with greater granularity the natural history of treatment and disease, and how medications benefit certain patients and sub-groups of patients," said Sachin Jain, Merck's chief medical information and innovation officer.

One of Merck's initial uses for Maccabi's data will be looking at the progression of osteoporosis.

"Traditional recruitment and enrollment into clinical trials is unfocused, laborious and expensive," said Lisa Griffin Vincent, president of PatientPoint Outcomes Research Solutions, a healthcare technology company that has partnered with Miami Children's Hospital to develop clinical trial data technology. "Recruitment methods typically include clinic staff poring through paper clinical charts page by page to identify patients that may meet the trial eligibility criteria. Access to big data can transform and accelerate the clinical trials process if leveraged in the right ways." Some of these "right ways" of using data include finding the right patients for the right trial in a timely manner. Aggregated health data can enable researchers to determine which patients are good candidates for particular clinical trials or treatment protocols, said Mark Hulse, CIO at Moffitt Cancer Center. Moffitt's for-profit company M2Gen works with pharmaceutical companies to match patients with clinical trials.

The challenge is the need for a large patient pool because of the number of patients that end up being ineligible for trials due to pre-existing diseases or medications. The traditional process of accruing enough patients can take over a year. Big data saves time and money by providing the tools to more accurately match qualified patients to trials.

Moffitt is in discussions with other health centers to combine data and possibly set up a national research information exchange.

"This certainly isn't something one system can do effectively," Hulse said. "There aren't a sufficient number of patients in total. You need to combine data from many different institutions."

But combining data from multiple institutions presents its own set of challenges, mainly a lack of standard data formats and definitions across many data systems and sets, Griffin Vincent said. For example, when identifying a patient as male or female, one system might define this data field as "sex," while another might define it as "gender." The large amount of unstructured data also presents problems: discrete data fields can be mined much more easily than written notes.

Data that can advance clinical trials includes EHRs, practice management systems, payer databases, pharmacy databases and patient-reported data. The EHRs include all clinical data: diagnoses, procedures, labs, medications and adverse events. Practice management systems include scheduling and visit information, claims submissions and insurance information.

PatientPoint's partnership with Miami Children's has the potential to accelerate the recruitment, enrollment and retention of patients in clinical trials. Miami Children's has more than 200 ongoing clinical trials with multiple pharmaceutical companies as well as government-sponsored research. PatientPoint's technology can integrate diverse data sets to flag patients that have a high probability of qualifying for specific trials. It also uses e-signatures to obtain patient consent as approved by the Institutional Review Board, another time-saving measure.

But access to all of this data can get complicated.

"Getting the right approvals and consents to use the data from different sources for the research is complex but imperative to maintaining ethical and legal rights, and ensuring data privacy and security," Griffin Vincent said.

There is always room for improvement, but the use of big data in clinical trials is starting to transform the way trials are conducted.

"There's a lot of wasted time in clinical trials," Hulse said. "If we can mine the data in the right way, it's good on the patient side because it can get drugs to the market much faster, and it's good on the business side because it saves the pharmaceutical companies money."