Patient social health networks have grown in number over the last few years, and as a result, the amount of data collected about disease has been augmented in tandem.
For example, since its inception in 2004, PatientsLikeMe, based in Cambridge, Mass., has been trying to speed the process of bringing new therapies to market by connecting its patient community of more than 125,000 and giving context to the millions of data points these patients provide about the more than 1,000 conditions they actively discuss.
PatientsLikeMe on one level is a social health network where patients can meet and share symptoms, treatments and experiences with others "like me"-struggling through a breast cancer diagnoses or living with multiple sclerosis, for example-says Chairman and Co-founder Jamie Heywood.
On another level, it's part of a health analytics infrastructure that can augment more traditional data collection methods.
"We are a health-outcome focused clinical discovery platform," Heywood says. "Effective disease discovery requires patient quantification and stratification, and right now, medicine as a business practice does not do that. Electronic medical records do not really contain outcomes and we provide a way for patients to contribute their outcomes. Our business is ultimately in understanding the meaning of health and contributing to discovery."
Heywood's personal journey started in 1998 when his brother Stephen was diagnosed with amyotrophic lateral sclerosis (ALS). In a quest to find and develop a new therapy to save Stephen, Heywood, an engineer with a degree from the Massachusetts Institute of Technology, created the ALS Therapy Development Institute, a not-for-profit biotechnology firm focused on developing new ALS therapies. While successful in bringing two drugs to clinical trials, Heywood learned the complexities and costs inherent to the process and-all too painfully close to home-the time constraints related to success.
PatientsLikeMe, co-founded with his other brother, Ben, and friend Jeff Cole, also MIT engineers, is a result of those experiences. The company is one of a number of organizations try to use social networks and sophisticated screening and analytics technologies to create patient registries and feed clinical trials.
Leveraging communities
Clem McDonald, Ph.D., director of the National Library of Medicine's Lister Hill Center for Biomedical Communication, oversees the ClinicalTrials.gov database and also sees advantages to the communities that social health networks provide.
"Recruitment of patients to clinical trials is important to the National Institutes of Health," says McDonald. "Anything that could facilitate this would be very, very useful. It would speed the complicated process." McDonald acknowledges that recruiting even 1,000 patients is slow going and typically takes much longer than expected.
"Currently, investigators can advertise in newspapers or on the Web, but in-ways are limited because of the fears around coercion," he says.
The number of legal and privacy hoops that must be jumped through has yielded sophisticated matching systems to find the right fit for patients who are willing to participate in trials. For example, PatientsLikeMe is developing a system to enable patients in its database to be matched against the ClinicalTrials.gov database, a registry of both federally and privately supported clinical trials around the world.
According to Heywood, matches occur in real-time, as engineers download the entire ClinicalTrials.gov database each night. Approximately 75 percent of the PatientsLikeMe population can be matched to a trial; however, geography is a limitation, as only 50 percent of those who match to a trial can match to one within 75 miles of their home.
"Nearly half of our active members in the main disease communities on our site have accessed the trials list, and 60 percent of those have linked through to individual trials," Heywood says. "We do not have data on the patients who actually enroll, but we when we push e-mails about trials, we accelerate the final study."
From a technical point-of-view, matching patients to trials is a difficult process. Trials are often too specific or too broad in their definition of disease, making a match to exact criteria challenging.
"Because the trials are not coded by us," says Heywood, "we have to match both specific and broader medical terms, which is a capability we are currently working on. The coding system at ClinicalTrials.gov is very flexible. It's difficult to define eligibility because you need to do so by looking at who the patients are, and what the data architecture is that will make it all work."
The National Library of Medicine (NLM) provides the keys for mapping trials into their system by computing the medical subject heading (MeSH) codes from the investigator-entered trial descriptions and includes the codes in the machine-readable clinical trial data files. PatientsLikeMe then uses NLM's Unified Medical Language System (UMLS) to map the MeSH codes onto the SNOMED hierarchy, which the health data integrity team has already coded with PatientsLikeMe conditions. They are still searching for "the perfect" MeSH-to-SNOMED mapping, as matches are missed when coding is either more or less specific in SNOMED. The future calls for greater flexibility in searching the ontology, which will improve the precision of the matching process.
A need for speed
Speed is at a premium during the matching process because time is an enemy for getting a therapy to market. The Pharmaceutical Research and Manufacturers Association (PhRMA) states that the drug development process takes anywhere from 10 to 15 years, from pre-discovery through the FDA review process. The completion of related clinical trials can eat up six to seven years of that time frame. And with an average cost for the journey estimated to be between $800 million to $1 billion to manufacturers, which is undoubtedly passed on through the health care system, it would seem that inroads to ethically shorten the process through technological advances would only make sense.
