Future of Health IT: Trends and Scenarios

Using crowdsourced librarians to outsmart Google

Thanks to Ian Jones, I found a very interesting article on Ars Technica entitled Using crowdsourced librarians to outsmart Google. It's about a project that aims to "out-Google Google" using the collective brainpower of librarians in crowdsourcing mode. It's based on what the project leaders say is sound science. There is a growing body of research on the credibility of online material, so their claims may be valid, but I haven't had time to vet them yet, so I am making no claims of my own here.

Still, the idea is very interesting. I've been hyperfocused on the problematic findability of quality consumer health information via the commercial search engines. If there is a better way of building an online search engine in terms of returning credible results, I'm all for it, and who knows? This could be it.

The Reference Extract project hopes to turn all of this credibility research into something practical. The team consists of professors like Lankes and the OCLC, an academic technology organization that counts more than 60,000 (!) libraries among its members. The initial work is funded by the MacArthur Foundation. Their first goal is certainly ambitious: use the web sites that librarians suggest most often as the basis of a more credible search engine that can return reliable results.

Reference Extract already believes that it can best Google, even before work has begun in earnest. That's because it has already used librarian-recommended sites to populate a custom Google search engine. Simply doing this produced search results that testers ranked as "more credible" than searches run on the main Google index, even when the testers had no idea that one of the searches was based only on librarian-approved sites.

I don't have time to think about this much at the moment, much less write intelligently about it at any length, being in crunch mode on a couple of projects all at once. Still, I wanted to pass it along. Others with a similar passion out there may find this very useful.

Yahoo’s desperate search times call for open source | Between the Lines | ZDNet.com

Larry Dignan and Tom Steinert-Threlkeld posted a very intriguing story about Yahoo! Search going Open Source, or more accurately, Open APIs. The post in their Between The Lines blog is entitled Yahoo’s desperate search times call for open source.

Yahoo has opened its search for developers to customize in its latest step in what it calls its Open Strategy, a plan to make Yahoo a platform. The effort is dubbed Yahoo Search BOSS, which allows “allows developers and companies to create and launch web-scale search products by utilizing the same infrastructure and technology that powers Yahoo! Search.”

Yahoo! is billing BOSS as the next step in their Open Search ecosystem, and it has some interesting possibilities. It may be, as the post points out, a move made out of desperation, which could bode ill for developers who invest a lot of time in writing to the APIs. What happens if Microsoft does finally make an offer they can't refuse?

I wish I could write more about this right now, but instead I must make an apology. I haven't posted much in the last two weeks (or maybe not at all, I can't recall) due to my being away at NetSci 2008 in Norwich, Norfolk, UK, followed by a brief appearance at work and then a long weekend holiday mini-vacation. Now that I'm back, I'm swamped with catch-up work, so I am having trouble finding posting time. But this news item was worthy of mention, because it bears on my hot-button topic, consumer health search. This was my topic at NetSci 2008, and I will soon be posting my slides and some thoughts and ideas about the current state of and future directions for consumer health search. I'm taking some vacation time this weekend to get caught up on personal things, including blogging.

I wish I had time to play with their APIs! Maybe that, too, will be soon...

More on Microsoft's HealthVault Strategy

Sean Nolan, chief architect of Microsoft HealthVault, commented on my post of a few days ago regarding Microsoft's HealthVault strategy. He gave me a pointer to an entry on his own blog to clarify how they are tracking the the provenance of data, or its "pedigree" as he refers to it (And Now for a Little Usability, April 17, 2008).  One of the comments on the post got me looking more closely at HealthVault, and thinking about how it can help with the current dismal state of the art regarding the search for consumer health information on the Web.

My thesis work tapped into a huge body of server log data  that accompanied a Microsoft Research grant to my thesis advisors, Lada Adamic and Suresh Bhavnani,  as part of the research program leading up to last summer's Microsoft Live Search Summit. The data set provided a remarkably clear picture of search behavior "in the wild". The bottom line: the quality of the result sets returned by major search engines from the queries of consumer health information seekers was questionable at best.

This came as no surprise, for two reasons. First, consumers seeking any kind of information, even about a subject as critical as a mortal illness, tend to submit terse, vague queries. Search engines are hard put to discern the user's intent. Their general-purpose algorithms do a pretty good job of bringing relevant information into view, but it is not uncommon for outdated or unscientific information to appear in the midst of (or even ahead of) authoritative results.

Challenges facing translational biomedical research

I was at two different meetings last week at the NIH in Bethesda, MD, both of which were related to the fostering of clinical research networks. The first meeting, called "Clinical Research Networks: Building the Foundation for Health Care Transformation", was the culmination of the NIH Roadmap "Re-Engineering the Clinical Research Enterprise" program, in which a dozen academic health centers and other research institutions fulfilled contracts to do research and develop tools and techniques for the creation, care and feeding of clinical research networks.

The network concept was defined broadly to include disease-focused networks, practice-based research networks (PBRNs), transplant donor/recipient networks, and others. There were keynote speakers from the NIH leadership, but the bulk of the meeting consisted of panels composed of principal investigators discussing the lessons learned in the fulfillment of their contractual obligations. While each contract produced some sort of intellectual property, but the lessons learned were the primary goal of the program, and the lessons were rich and diverse.

"Better Than Free": Lessons for Open Source healthcare software systems

In The Technium, Kevin Kelly's "Better Than Free" post should be a wake-up call for Open Source project leaders, in particular those in the healthcare arena. I first got wind of his post via The Edge, and when it popped up again in Seth Godin's blog, I had to stop and take a look.

This super-distribution system [the Internet] has become the foundation of our economy and wealth. The instant reduplication of data, ideas, and media underpins all the major economic sectors in our economy, particularly those involved with exports -- that is, those industries where the US has a competitive advantage. Our wealth sits upon a very large device that copies promiscuously and constantly.

Yet the previous round of wealth in this economy was built on selling precious copies, so the free flow of free copies tends to undermine the established order. If reproductions of our best efforts are free, how can we keep going? To put it simply, how does one make money selling free copies?

I have an answer. The simplest way I can put it is thus:

When copies are super abundant, they become worthless.
When copies are super abundant, stuff which can't be copied becomes scarce and valuable.

When copies are free, you need to sell things which can not be copied.

Well, what can't be copied?

A great question. Fortunately, Kelly has an answer ready, though I think he left out one important point.

Health 2.0, as seen by those canny Brits at the Economist

In the most recent Technology Quarterly from the Economist, offers a guardedly upbeat assessment of the state of consumer health information, in an article entitled Health 2.0. I apologize if the link is for subscribers only (I didn't check at the time I read it), but I value their opinion as trend validators.
 

Misinformation is another worry. On the internet, as the old saying goes, nobody knows you are a dog—or an idiot, notes Dan Keldsen of AIIM, a non-profit association based in Silver Spring, Maryland, which helps companies manage digital information. And too much health information can confuse people, says Monique Levy of Jupiter. But a survey by the Pew Internet & American Life Project in Washington, DC, suggests that although user-generated information offers consumers more health options, the upside outweighs the risk, says Pew's Susannah Fox. Nearly one-third of the 100m Americans who have looked for health information online say that they or people they know have been significantly helped by what they found. In contrast, only 3% reported that online advice had caused serious harm.

A lot of user-generated health information is accurate. A panel of neurology specialists judged that only 6% of information posted in the epilepsy-support group of BrainTalk was factually wrong, according to a study published in 2004 in the British Medical Journal. And with enough people online, misinformation is often quickly corrected. Inaccurate posts on the website of the Association of Cancer Online Resources (ACOR), for example, will be pointed out within two hours, says Gilles Frydman, the founder of the association, based in New York.

As a rider on the Health 2.0 bandwagon, I bask in the Economist's praise, deservedly or not. But I also have done some empirical research of my own to back their opinion up.

More on the $1000 genome: Gary Wolf on The Quantified Self

I found a more skeptical take on the 23andme story I reported on in my last post.  It's in the Kevin Kelly blog, and the post (not by Kevin himself, but that's OK, Gary's cool) is called The Quantified Self.

Still, within all of these stories I caught the first hint of a backlash. This backlash has little to do with the main objection all of these writers presented, which was: do I really want to have this important knowledge about my genes? I don't think this objection is serious. I don't even get a sense the writers seriously entertain it. I suspect it is a rhetorical effect, an exaggerated worry that lends interest to the promise at hand. (Gray goo serves this function for nanotech.) Perhaps it is the nerd in me, but I have difficulty imagining that accurate medical self-knowledge would be rejected by consumers because they are afraid to know.

Anyway, there's a bigger problem. The companies offering these tests carefully insist that they are not be taken diagnostically, and acknowledge that that are not currently accurate indicators of clinical risks. They give these caveats, while of course also trying to satisfy our desire to get accurate clues to our genetic predispositions to scary diseases.

There's more detail and references than I can describe in this necessarily brief post, but here's a taste:

THIS JUST IN: Announcing the arrival of the $1,000 Genome

OK  I admit it's not quite what the NIH is referring to in their $1,000 genome RFAs (RFA-HG-06-020 (R01), RFA-HG-06-021 (R21) and RFA-HG-06-022 (R21/33)), but it's a nice start, and it's available now.

I just got hold of a press release from a startup called 23andMe. The title is  23andMe Launches Web-Based Service Empowering Individuals to Access and Understand Their Own Genetic Information.

MOUNTAIN VIEW, Calif. -- November 19, 2007 -- 23andMe, Inc., a privately-held personal genetics company, announced today that it has officially launched and has begun offering its services directly to consumers in the United States. 23andMe is helping individuals understand their own genetic information through the latest advances in DNA analysis and web-based interactive tools. The Company's service will enable customers to gain deeper insights into their ancestry and other inherited traits which are marked in an individual's genetic code.

The 23andMe service allows individuals to:

* Search and explore their genomes;

* Learn how the latest research studies relate directly to traits identified in their genome;

* Compare their genomes to family and friends who are also 23andMe participants;

* Discover their genetic roots and find where they sit on the tree of human genetic history; and

* Give individuals the option to actively participate in a new research approach.

Linda Avey, co-founder of 23andMe, said, "Over the past several years, significant advances have been made in the field of genetics. The mission of 23andMe is to take the genetic revolution to a new level by offering a secure, web-based service where individuals can explore, share and better understand their own genetic information."

Anne Wojcicki, co-founder of 23andMe, said, "23andMe is designed to provide our customers with scientifically accurate, high-quality information about their own genetic code in a format that is easy to understand and use. We believe this information provides intriguing insights into an individual's genetics, with the goal of expanding the collective knowledge base by enabling active participation in research."

But wait. It gets better.

A Positive Trend: New Search Engines from Healia

I found some potentially great news regarding consumer health search at  beSpacific: New Search Engines Focus on Biomedical Literature and Clinical Trials Information.

"Healia announced today the launch of two new search engine tools - Healia PubMed/Medline Search and Clinical Trials Search - to help consumers find relevant biomedical literature and clinical trials information...Healia PubMed/Medline Search is specifically designed to help consumers retrieve abstracts (summaries) of scientific articles published in biomedical journals in a user-friendly way. It searches the National Library of Medicine's (NLM) PubMed/Medline dataset, which includes more than 17 million abstracts and citations from approximately 5,000 biomedical journals published in the United States and worldwide since the 1950s."

Omitted from the quote is that the Clinical Trials portal data comes from ClinicalTrials.gov, the National Library of Medicine's database of active clinical trials.

Healia is doing a great service, assuming these work as advertised. I have every reason to believe they do, but I haven't had time to try them out. Hopefully this trend will grow over time - providing consumer audience-tailored search to public databases whose official front ends are targeted at professionals, or not designed to be as user-friendly as possible, or both.

Kudos to Dr. Eng and the crew at Healia!