As a senior in high school, Julia Winn was diagnosed with clinical depression. For two years, she took medication, sought therapy and saw 10 different doctors, but no one could figure out a cause. Finally, after seeing a specialist, she learned that the culprit was her hormonal birth control pill, which has a well-documented relationship with depression.
Using her medical history and experience with the first birth control pill, the specialist recommended a more appropriate alternative and, since then, Winn said she’s been fine. But, having figured out the best medication for herself, the 23-year-old Harvard graduate is building a startup that leverages social data and natural language processing to help other women discover what could work better for them.
“The larger goal is to change the way doctors think about patients and change the way patients think about themselves,” she told me at the Health 2.0 conference this week in San Francisco. “It’s not one size fits all.”
Social data uncovers insights for a range of drugs
Part of TechStars Boston’s fall class, MyBetterFit aggregates comments from health forums and then parses out relevant information about treatments and side effects to create patient profiles against which users can compare their experiences.
Winn and her team are still building out the site, but the goal is for women to be able to answer a few questions about their current experiences with birth control, and ultimately their medical history, and then receive social-data-driven recommendations. A Las Vegas biotech startup, called Lucine Biotechnology, which my colleague Derrick Harris has written about, is somewhat similarly attempting to improve women’s healthcare by drawing connections between social comments online and information gathered through salivary hormone monitoring. And Alliance Health Networks, a Salt Lake City, Utah-based company recently covered by our colleague Stacey Higginbotham, is using machine learning models that have been taught using public medical research to uncover insights in online healthcare discussions.
So far, MyBetterFit has analyzed 100,000 comments from just two forums, eHealth and Topix, and already, Winn said, they’ve observed that for each drug there are about half a dozen cohorts of patients who respond similarly.
Down the road, Winn said, the plan is to expand the platform to patients taking less-understood medication for other conditions, such as autoimmune disorders like psoriasis, depression, ADHD and menopause.
Treato, a company based in Israel, similarly applies big data analytics to social data pulled from comments on thousands of blogs and health forums. But instead of giving patients an opportunity to personalize recommendations, it algorithmically calculates how many patients respond positively to a certain drug relative to others, as well as surfaces patients’ top concerns.
To date, the Reed Elsevier-backed company (see disclosure below) says it’s analyzed more than one billion online patient discussions and, earlier this month, it rolled out a new insights platform for pharmaceutical companies. Similar to social listening platforms that let marketers observe how consumers are talking about their brands on Twitter and Facebook, Treato’s tool lets drug companies monitor patient experiences with their medications.
As genomic research develops, social data could be a complement
Even though Treato doesn’t personalize results, it still gives patients an important and quick window into the range of side effects caused by a specific drug.
As I wrote about earlier this week, sophisticated genomics research is working to personalize drug prescriptions and enable physicians to recommend medication most appropriate for a patient’s genetic profile.
As that technology develops, Winn said, phenotypic data collected by her startup will still be able to provide a valuable complement, as well as provide insight into reactions that aren’t related to genetic factors.
“We’re excited to be providing a service that can help everybody, right now,” Winn said. “But it would be natural in the future to pair our phenotype findings with genetic data to get an even more complete patient profile.
Disclosure: Reed Elsevier is an investor in GigaOM through its Reed Ventures arm.
