PCAST on Health IT

The President's Council of Advisors on Science and Technology (PCAST) has recently published a report to the White House on the impact of Health IT on Healthcare. PCAST is an esteemed group of appointed scientists and engineers whose mission it is to provide an unbiased assessment of matters technical and scientific.

PCAST's report comes at a crucial time in U.S. healthcare; after the passage of legislation (the Patient Protection and Affordable Care Act, PPACA and HITECH) intended to save healthcare and jump start health IT. This report provides a vision of healthcare in the near future; what obstacles prevent us from getting there and what should be done to surmount those obstacles.

There are a number of good use cases in the report that provide some very compelling arguments for better health IT and how it would benefit the population. It is not surprising that these use cases require that medical information be exchanged between healthcare providers, i.e. an exchange of a patient's information between hospitals and doctors. At this point, however, the government has not actually specified a standard language to be used. In this report, PCAST is recommending that the government step in and explicitly state those standards.

As in most silo'd industries, healthcare organizations cannot exchange information between organizations because each uses their own flavor of healthcare terminology. This is nothing new and many Standards Development Organizations (SDOs) have been working on interoperable healthcare terminology for some time, i.e. ISO, HL7 and IHE to name a few.

PCAST is advocating for health information to be represented as "meta-data tagged data elements" "some kind of extensible markup language (an XML variant, for example)." This is in contrast to the message and document structure used in current information exchanges. Information represented in this way would provide more flexibility, and security, than current methods.

Although it is never explicitly referred to in the report, this method of representing data starts to look like the semantic web technology of RDF. Indeed, one of the use cases could be taken as a vision of what the semantic web could do. One very compelling aspect of this method is the lack of a central data repository, that the data would be assembled from distributed sources as needed, reducing the huge effort of assembling and maintaining such a repository, not to mention the security aspects.

HHS has made a great effort in pushing through the first stage of the HITECH Meaningful Use Criteria which is to be implemented in 2011. They are now poised to move forward with stage 2. PCAST has laid down a considerable challenge but one where as patients, we could start to see some real improvements in our healthcare.

The Bombs and the Bees

In an effort to solve the mystery of what is causing bees to spontaneously die off around the country; scientists from two distinctly different disciplines have joined forces; entomologists and military scientists. I am always fascinated at how groups in dissimilar fields can connect to find solutions to problems. Not necessarily how they would work once they have found they have shared goals, but what got them together in the first place. In most cases scientific and technical disciplines become very insular; a strategy designed to focus on predefined goals and exclude extraneous influences from disrupting the process. All too often we see that another group has solved a similar problem or has the makings of a solution but the groups have no visibility of one another due to the lack of cross disciplinary communication.

In the case of the bees, entomologists have been focused on finding some pathogen that could be the cause of the bees' plight; Colony Collapse Disorder (CCD). The process of finding a unique protein in the bees' tissues is like finding a needle in a haystack, a haystack the size of Montana. But the good news is that the technology that can find this protein does exist, it's just not sitting in every entomologist's lab. Enter the U.S. Army. Just so happens that they do have this kind of technology sitting in their lab, intended to be used to quickly find harmful pathogens in solders. It was new, so new that they had yet to try it out in the field. In lieu of sick solders, bees would do quite nicely.

So the Army gathered up the dead bees and ran them through their machine and found the guilty virus. But how did the Montana entomologists ever get hooked up to the Army team on battlefield diseases in the first place? Reading journal articles, attending conferences, searching the Internet? Nope. The brother of one of the Army researchers, a tech entrepreneur, saw an interview of a bee expert on television. He then called his brother and set up a meeting. Good old fashion family connections.

I am all for serendipitous meetings, it is always exciting and to discover previously unknown mutual interests. But I can't help but wonder what if we could coax the process along, maybe remove a few barriers to the chance meetings. This is the idea behind the VIVO project that I blogged about last month. The challenges of staying abreast of developments in a single scientific discipline are indeed daunting; to ask scientists to keep up with developments in multiple disciplines is impractical. If there were a system that could perform this task, like the TV watching brother, this may remove some of the barriers that currently exist.

The Power of Pull

David Siegel has taken a futurist's view in describing what a world with the Semantic Web would be like in his book Pull: The Power of the Semantic Web to Transform Your Business. In this world it seems that everything works seamlessly: machines can talk to other machines and to humans and cars and houses and they all seem to know what we want before we know it ourselves. No doubt that adding more intelligence to the Web by organizing the information and adding connections between data has unimagined potential. But Siegel is busy imagining it nevertheless.

This book is deliberately non-technical; describing what the Semantic Web can do with very little emphasis with how it can be done. As its title attests: this is a book for business managers who will be given plenty of examples of what's in store in the Semantic Web future. Techies will also get something out of the "big picture" view and be motivated to plunge deeper into the technical details.

The title "Pull" comes from the concept of actively pulling information and knowledge when and where it is needed, instead of having information pushed at us. This is intended to give us, the consumers of information, some control over its consumption. There is now so much information, both useful and not, that is being pushed on us that we tend to block it all out. If we could control that stream, and receive only that which we need, it would make this information much more useful.

Siegel points out an interesting trend in health information; that patients are demanding and receiving more control over their own medical records. They are demanding the right to pull their data from hospitals, pharmacies and insurers and are not content with waiting for these entities to push just that information they wish to the patient. This is indeed taking place, encouraged by provisions in the American Recovery and Reinvestment Act which requires doctors and hospitals to provide patients with their information after every visit.

But will advertisers and media companies sit idly by and let consumers take complete control over the information that reaches them? Siegel is fundamentally a cheerleader in the book, building excitement and enthusiasm for this brave new world without really analyzing the downsides. One significant downside is the imminent loss of privacy; data on every facet of our lives will be spread all over the Internet, potentially available for misuse by malevolent forces.

Siegel provides lots of links to current Web sites that provide a glimpse into what we may expect; this field is advancing so rapidly the book may be out of date in a year. In fact, if Siegel's vision comes true, all books may be obsolete. But don't let that stop you, read a copy now so you know what to expect. As Vinod Khosla is quoted on the back cover "There are at last ten killer business ideas in here."

VIVO: Social Networking for Scientists

With the wide spread adoption of Facebook by the general public social networking has hit the main stream. Facebook's strategy of keeping their interface as simple as possible has made it easy to use and accessible to many but has limited its usefulness in expressing and transferring knowledge about its users. This has opened the door to social networking products that hope to target niche markets that Facebook has underserved.

Jive has entered the enterprise market for social networking, while Sermo has targeted physicians. Ning offers a platform for building social networks. ResearchGATE, Nature Network and sciencestage.com have all targeted the scientific disciplines. And now there is VIVO: Enabling National Network of Scientists an effort to create a social network for scientists. VIVO is a project in late stage development/implementation by University of Florida, Cornell and others and funded by a grant from the National Institutes of Health (NIH) that will offer scientists a way to quickly scan for potential collaborators, bridge domain silos and see what their colleagues have been up to.

But unlike many of the other offerings, VIVO has an underlying benefit that the other social networks do not: Semantic Web technologies. In order to support a semantically rich and machine processible environment, the developers of VIVO have created a formal ontology for the academic domain. This ontology now provides semantic interoperability across a university domain and between university systems.

Will we now be hearing from scientists a complaint that's grown louder in the general public: that they couldn't meet the grant deadline because they were messing around on VIVO for several hours? Undoubtedly. But collaboration is a valuable goal and will pay off in the long run by reducing the barriers between scientists and disciplines that have slowed the spread of ideas in the past.

Open Data in Alzheimer’s Research

I've long been a fan of open access to scientific documentation; providing free online access for scientific work, especially that funded by public sources. A recent research project on Alzheimer's has taken this a step further by providing access to experimental data as it is produced. This is a major paradigm shift in scientific research, one which could provide both real benefits to the research and upset the traditional scientific culture.

Benefits include increasing the pool of researchers analyzing the data, i.e. more eyes and minds that may lead to more discoveries. The data set has been downloaded over 3200 times, which is some indication of distribution.

But this is a major departure from how research has been done and it conflicts with some of the foundations of scientific research. Scientists are rewarded for what is considered to be their discoveries, inventions, ideas. They are incentivized to keep this information private until they themselves have analyzed it and have made what discoveries they could. It makes clear sense that if you are the one who wrote the grants, received the funding, ran the experiments you should be the one to get the credit and ensure future funding.

Even more remarkable is the fact that the Alzheimer's Disease Neuroimaging Initiative, ADNI, is a public-private collaboration including some 20 pharmaceutical companies. The drug companies can still profit off of drugs they may eventually produce, but the data will remain in the public domain.

There have been a number of important papers that have been published recently and some very significant results on early diagnosis of Alzheimer's, a fact that may be attributed to this new paradigm. More than 100 studies are now underway on drugs that may stop or slow the disease.

Other large research studies have taken notice and are using the Alzheimer's project as a model for collaborative research. This may not be a model for all future research, but it does provide a glimpse at how collaboration could advance research in certain areas.

Top Down or Bottom Up?

If systems are thought of as having a vertical structure, building them would proceed from one end to the other; either start at the top and work down or start at the bottom and work up. An iterative approach may require multiple end to end swipes but they would most likely proceed in the same end-to-end fashion.

Consider a large construction project, such as a bridge. A vision of the completed bridge, and its connections to a road or rail system, would exist before the first bulldozer rolled. This is a classic example of a top-down project. And the planning stage is not even really the top of the system; consider the budgets, laws, regulations, and standards that must be in place even before the initial plans are drawn up.

The bottom-up approach usually starts at the level of implementation; the construction site, an empty lot. Solutions are implemented quickly and meet immediate needs, i.e. a shed, mobile home to fill a need for shelter. These solutions, by their nature, may not take system wide impacts into consideration and may in fact create issues with other parts of the system; where's the bathroom? This is not to say that these types of solutions are not valuable; the key is knowing when a bottom-up approach can be taken and when a more top-down solution is necessary.

One frustration we all have with top-down designs is that they tend toward the one-size-fits-all approach; the designers cannot anticipate all possible scenarios in which the designs will be implemented. We may be familiar with some examples of this such as Medicare reimbursements incentivizing unnecessary treatments, government programs intended to encourage home ownership bringing down entire economies. Top-down approaches, left unmanaged, are prone to abuse simply because they cannot predict what the future will bring.

The recently enacted HITECH Meaningful Use criteria constitute a top-down edict instituted by the government to the healthcare industry: if healthcare providers and hospitals comply with the edict then they will receive incentive payments, if not they will be penalized. The problem is that these rules have taken the focus from providers and hospitals directing their own information technology strategy to one that is focused on meeting needs that may not directly benefit the providers and hospitals. There is an assumption that the end result will be to the benefit of all stakeholders, but that won't be known for many years and in the meantime the costs, only partly defrayed by incentive payments, will be borne by the providers and hospitals.

Can a transformation, such as that being attempted by the U.S. healthcare industry, be accomplished through a bottom-up process? In fact bottom-up processes are in continuous practice, solving short and long term issues. To expect a bottom-up process to come up with a national infrastructure, in a well established industry, would be quite extraordinary.

Summer Health IT Summit

The Summer Health IT Summit was held in Denver this week drawing an energetic group of health IT decision makers from across the country. The topic of the conference was the Adoption and Implementation of the Meaningful Use Rule. The presentations ran the gamut from how to get started with installing an EHR system to integrating clinical and research databases in a research hospital. The crowd was evenly split between hospitals/providers and vendors with a few consultants thrown in.

Two points that were brought up repeatedly were the requirements for interoperability and for privacy and security. It was not clear that participants had a good handle on either of the concepts although Lynn Vogel, CIO of MD Anderson, made the claim that interoperability is one of the most overused words in healthcare. Although implementation strategies were in short supply, participants seemed to be looking to the Federal Advisory Committees, the HIT Policy Committee and HIT Standards Committee, for guidance.

There was an interesting distinction that was drawn between public and private Health Information Exchanges (HIE). With the advent of the HITECH Act, much of the focus has been on public HIEs which would connect hospitals and providers to state and federal networks. Private HIEs are being developed to network hospitals within a single IDN, a hospital to its providers, or hospitals using the same EHR vendor (re: Epic). Although the business model for public HIEs is unproven, and past failures abound, private HIEs have shown to have a valid business model and show compelling ROI.

There were several participants that voiced some displeasure at the top down approach handed down by the government. As with any top down, broad based approach, the unintended consequences of these rules cannot be known at the outset. A bottom up approach to solving the health IT problems may be preferred by hospitals because they may feel that they will get the results that are most valuable to them. Hospitals have had ample opportunity to solve these problems and most have not availed themselves of that opportunity, so, like it or not, the Meaningful Use Rules must now be complied with.