Future of Health IT: Trends and Scenarios

The Internet of Things and its implications for health IT

The Internet of Things (IoT) refers to the enormous wave of sensor and actuator devices becoming manifest on the Internet, each with its own address that allows its sensor(s) to be read and/or its actuator(s) controlled via other devices, including but not limited to those providing human-computer interfaces.

IBM is working with OASIS to promulgate a standard for communications with and between IoT devices. The connection to health IT may not be intuitively obvious, but this trend is important. Standards-based technologies are what drive the Internet and the World Wide Web, not to mention just about all information and telecommunications technologies.

Some pundits disagree with the need for standards, for example Kishore Swaminathan of Accenture, mostly based on the premise that device inter-communications will be highly specialized and conducted over proprietary networks. I can only say that I heard such arguments in the 1980's with respect to electronic mail and social online environments like Compuserve and America Online.

Network effects put paid to proprietary email and online service protocols. Critical mass is immensely powerful as a driver of societal and technological evolution. 

What will the IoT mean for the world of healthcare? Ubiquitous monitoring of vital signs is just one example of how the IoT will be of benefit to healthcare providers and patients.

For decades, computing devices have been the machine equivalent of Helen Keller, with an extremely limited amount of sensory input from the real world. That is changing rapidly. Individually and in distributed configurations with a facade of individuation, computers are beginning to access the same sources of real-world experience that are available to us real human beings.

I have long been skeptical of the concept of the Technological Singularity promoted by Vernor Vinge and Ray Kurzweil, in which machine intelligence surpasses human intelligence, but thinking back to Helen Keller's sudden transformative leap in comprehension gives me pause. It is no longer unthinkable that at some point, our thus-far-unintelligent machines will be capable of independent learning. Given their tirelessness and access to vast quantities of data, a singularity event no longer seems out of the question.

Does that mean I believe that machines could render their human progenitors irrelevant? No. Is my mind open to the possibility? Reluctantly but definitely. Were it to come to pass, would I want to live to see the day? Likely not.

Can Open Source solve the health information integration problem?

I have a vested interest in better health informatics at the primary care level, both because I am a patient and because of my consulting relationship with Cielo Medsolutions LLC. In the current political environment, with its emphasis on throwing money at the problem, I fear for the public health. Bad health IT is worse than no health IT.

How would I define good health IT at the primary care level? I define it in terms of accuracy and completeness.  For your electronic health record (EHR) to be accurate and complete, the information in it must be up-to-date and cover all key facts about the state of your health and the care provided to you. Because many patients have multiple providers, this means that your EHR must be available to all participating care providers, both to read and to update.

Providers often use different EHR systems from disparate vendors, most of whom store data in proprietary formats, and each of whom has a fiduciary responsibility to its shareholders to resist dilution of control over the health information managed by the vendor's software. Moreover, providers belong to different HIPAA covered entities (i.e. custodians of records containing HIPAA-protected health information).

Thus the challenges in achieving a seamlessly integrated health record are daunting, spanning the technical, political, and regulatory realms. Nonetheless, their is a significant push at the Federal level with bipartisan support to achieve health information integration.

Third-party payers have already compelled a degree of standardization with respect to some of the information in the EHR, though the information they need for efficient claims processing is only a subset of what is needed from the care provider perspective. Adequate or no, at least it's a start.

In addition to the provider and vendor challenges, both because providers can make data entry errors and because the patient does not always take the provider's advice, the record must also be available to the patient to read and update. This latter problem is a matter of providing a patient portal to the integrated EHR shared by the providers, and while it presents its challenges, this is an easier problem to solve than getting to the integrated EHR in the first place.

Solving the political challenges, both inter-vendor and at the governmental level, is outside my scope. I do, however, have some insights into the technical challenges to health information integration, in particular the syntactic and semantic obstacles that must be overcome.

Getting information out of proprietary systems is essential to making it available for integration, and there do exist commercial entities who specialize in "cracking" proprietary schemas. Recently, though, I learned of the existence of an Open Source initiative in this arena, which brings me to the interview linked below from opensource.com. In it, Jason Hibbets of Red Hat interviews Matt Mattox of Axial Exchange, a company that provides an Open Source data interchange appliance and sells services related thereto.

I'll provide a taste from the interview, and leave you to read it (or not) without further commentary on my part. I'm still investigating them so I can't provide an opinion of any sort with respect to Axial, but I do have to say I find their business concept highly intriguing.

When you take a system where suppliers create their own demand--which is what happens when physicians both diagnose and treat illness--and combine it with model where a third party picks up most of the fees, costs can rise quickly.

In terms of the care model, the physician practice and acute care hospital were designed in the early 1900s. Medical knowledge has evolved dramatically since then. Today, there are relatively inexpensive ways to treat conditions that a century ago were understood only by their symptoms. We have an opportunity to re-organize our institutions around the state of the science.

These changes won't happen overnight. The first step is to extract data from locked-down proprietary systems so that health information can be made available to the appropriate people at the appropriate times. Once the data has been freed, we can begin to innovate around new care and payment models. Freeing clinical data is the problem we're working on at Axial.

via opensource.com

Personal Tech: Medical Apps for the iPhone

From NY Times Personal Tech: Medical Apps for the iPhone

After a lot of time reading "best medical apps" stories online, asking Twitter users for their suggestions and reading online reviews, I finally boiled the list down to about 50 promising apps. I tried them out and further winnowed the list to a bunch that I ultimately demo'ed in my talk. Eventually, the video of the talk will be posted at ted.com/talks, but in the meantime, here's what I covered.

(I haven't really given them full-blown testing, so read the online reviews before you spend good money on them. Except for the free ones--you've got nothing to lose!)
--Dave Pogue

I read David Pogue's personal tech blog at the NY Times to keep up on consumer IT trends, but this post from a couple weeks ago seemed really relevant to the readers of FutureHIT. I've been putting off posting because I wanted to say something eloquent, but work and home life have made that impossible. I'll settle for a few hints at what I believe are the most interesting among the items Pogue lists. I'm focused on the ones targeted at health professionals, but the consumer tools have a lot of potential as well.

• OsiriX. "An amazing viewer of medical images (X-rays, scans of all sort)".  I think this is probably just a DICOM viewer, but it requires special server software, so it may be adding some value. Regardless, it's very cool.
• AirStrip OB. "Lets an obstetrician monitor a patient’s status, right down to the baby’s heartbeat, from elsewhere in the hospital (or the town)."  As Pogue points out, this is a good pointer at the types of remote monitoring applications we can expect to see as ubiquitous/pervasive computing moves into the mainstream.

Actually I will mention a thought I've had about one of the consumer apps...

• Retina. "It’s for color-blind people like me. You hold it in front of something—clothes in your closet, for example—and it tells you by name what color you’re seeing."  This is an examplary augmented reality application, intriguing not so much for what it does -- which is of great value, I don't mean to denigrate it -- but for what it portends.  As image analysis becomes more sophisticated, it's easy to picture diagnostic tools that will look at the patient in visible light and infrared and point out diagnostic signs the clinician might overlook. A more prosaic application not yet built but within the reach of current technology would be a program that "looks" at the patient and points out relevant facts and decision guidelines based on the part of the patient's body on which the phone's camera is focused.

I think all this is intimately related to the technology I spotlighted in my recent post on the Open Data Kit, which is aimed at moving advanced tech for the clinician into community settings. As I said in that post, these types of tools will serve the developed world as well as the developing world, because even in the most advanced economies there are segments of the population that are underserved because they can't or won't go to a traditional healthcare provider setting.

There are challenges in all this that take some of the rosy glow out of the picture, notably patient privacy and truth maintenance, i.e. making sure the various information stores relied on by the remote devices have the most up to date information, and can update it as well. In addition there is the very real concern I feel as a patient that healthcare providers may become overly dependent on distributed computing and communications technologies, and be at a loss when connections break down (and they will, oh yes, they will).

Nonetheless, on balance the future for these technologies is bright and getting brighter, and the value they add is potentially enormous.

Wiser OWL learns to unravel doctor talk (The University of Manchester)

I saw a really good news bit today, actually two days old but that's how far behind I am on non-critical email:

A good example of the scale of the problem facing medics and computer scientists is the NCI Cancer Thesaurus that has swollen from 20,000 medical terms in 2004 to over 50,000 terms today.

As new terms are being added all the time, ensuring that all these terms are described, updated and linked together correctly is a mammoth task for humans. However, by using OWL 2 definitions can be written in such a way that computer programs can tirelessly update these terms, enriching the structure of the Thesaurus and pointing out where there are errors.
 
"The first stage was writing the NCI Thesaurus in the original version of the language, OWL, but now OWL 2 enables computer programs to interpret these terms in a much more human-like way, for instance reasoning that if a fracture is located on a bone which is part of a leg then that fracture is a fracture of that leg," said Bijan Parsia of the University of Manchester’s School of Computer Science.

via www.manchester.ac.uk

I can't write more about it now, but it's a critical step in the right direction for the evolution of the Semantic Web. Some maintain it hasn't even been born yet, much less started to evolve. There's an element of truth in that, but it's a sign of the change in our sense of time that we would expect something as complex as the Semantic Web to  move along more quickly than it has so far. Kudos to the folks at Manchester and Oxford who made this manifest earlier this week.

The future of computing and communications: random thoughts, part 5

The New Scientist recently published an article entitled "Cell phones could be used to build 'audio Internet'". (I'm not sure if the article requires a subscription to obtain full access, and I apologize if it doesn't, but the link should at least lead to the abstract in that case.) This is one more indicator of the proliferation of modalities for human-computer interaction (HCI) that has been a major theme of my random thoughts in recent months.

The spoken web is an attempt to bring the power of the internet to rural India. Despite India's highly touted knowledge economy, around 70 per cent of its billion-strong population lives outside its cities, and most of them earn just $4 per day or less. And even for those who can afford computers, attempting to access the internet remains a futile exercise, either because they cannot read or write or the information on the web is simply not relevant to them.

The answer, says Tapan Parikh of the University of California, Berkeley, is speech. "An audio format would provide much more access and opportunity for local people to contribute," he says. "While a farmer may not be able to write a memo, or an email, or a summary of his work, he can easily talk about it."

Between a quarter and a third of India's one billion-plus population have access to a cell phone, with the number growing at 8 million per month. This is far greater penetration than the World Wide Web in India, especially in semi-rural and rural areas. Moreover, the content of the Web is not useful to India's poor, either (or both) because they cannot read or the content is irrelevant to their daily lives.

So how to leverage cell phone technology to bridge the Digital Divide? The answer posited by the New Scientist article is the spoken web.

...Conceptually, the spoken web is a network of VoiceSites, just as the internet is a network of websites. A VoiceSite can only be accessed by a phone, and only requires the user to be able to speak and listen. Callers can create their own VoiceSites or access those of others. They can also surf the spoken web, jumping from VoiceSite to VoiceSite using speech.

In developing the system, the team took their cues from the growth of the internet. "The thing that really fuelled the adoption of the World Wide Web is the ease with which you can create websites," says IBM IRL's Amit Nanavati. So they focused on developing a simple way to create VoiceSites.

Say a plumber wants new customers. First he calls a number and software called VoiGen then guides him, in his local language, through the process of setting up a VoiceSite. Relevant information, such as a welcome greeting and the plumber's contact details are recorded. Behind the scenes, the VoiGen system then creates a VoiceSite. A phone number, analogous to a URL, is then assigned to the plumber.

Anyone who calls the plumber's VoiceSite number is greeted with his welcome message, and then given help to navigate the information. For example, the caller can say "address" to hear the plumber's address, or "appointment" to request a call-out with him.

Conceptually, the spoken web is a network of VoiceSites, just as the internet is a network of websites. A VoiceSite can only be accessed by a phone, and only requires the user to be able to speak and listen. Callers can create their own VoiceSites or access those of others. They can also surf the spoken web, jumping from VoiceSite to VoiceSite using speech.

In developing the system, the team took their cues from the growth of the internet. "The thing that really fuelled the adoption of the World Wide Web is the ease with which you can create websites," says IBM IRL's Amit Nanavati. So they focused on developing a simple way to create VoiceSites.

Say a plumber wants new customers. First he calls a number and software called VoiGen then guides him, in his local language, through the process of setting up a VoiceSite. Relevant information, such as a welcome greeting and the plumber's contact details are recorded. Behind the scenes, the VoiGen system then creates a VoiceSite. A phone number, analogous to a URL, is then assigned to the plumber.

Anyone who calls the plumber's VoiceSite number is greeted with his welcome message, and then given help to navigate the information. For example, the caller can say "address" to hear the plumber's address, or "appointment" to request a call-out with him.

The vast majority of people on this planet are poor and illiterate, but more and more of these folks are acquiring access to cell phones, partly due to initiatives like GrameenPhone, but mostly due to the fact that cell phone infrastructure is profitable even in developing countries, in turn due in part to compelling economic reasons to want telephone access.

One example: I have read accounts of fishers in Kerala, one of India's poorest states, using cell phones to find out where to get the best price for their catch, eliminating their dependence on a single local bulk buyer. They can learn the price at various ports within sailing distance and mentally compute the tradeoff between the cost of traveling to a given port and the value of their catch at that port. A classic reduction of market friction.

Clay Shirky wrote an acerbic commentary 2002 on the tendency to sentimentalize the "digital divide". In Half The World, Shirky wrote:

Because wireless infrastructure does not require painstaking building-by-building connectivity, nor is it as hampered by state-owned monopolies, it offers a way for a country to increase its telephone penetration extremely quickly. By 2001, there were 25 countries where cell phone users made up between two-thirds and nine-tenths of the connected populace. In these countries, none of them wealthy, a new telecommunications infrastructure was deployed from scratch, during the same years that keynote speakers and commencement invitees were busily and erroneously informing their listeners that half the world had never made a phone call.

This shift suggests that characterizing the "digital divide" in terms of access to broad- and wideband connectivity is naive. The power a cell phone provides to a poor fisherman is impressive, as one can see from this quote from a 2006 Washington Post article:

Rajan said the dealers don't necessarily like the new balance of power, but they are paying better prices to him and thousands of other fishermen who work this lush stretch of coastline. "They are forced to give us more money because there is competition," said Rajan, who estimated that his income has at least tripled to an average of $150 a month since 2000, when cellphones began booming in India. He said he is providing for his family in ways that his fisherman father never could, including a house with electricity and a television.

The spoken web could extend the leverage described by this fisherman to myriad small businesses and to individuals in the general population of the developing world. As with the World Wide Web in its infancy in the early '90's, it is impossible at this point to imagine the potential of the spoken web.

Another question: when high-bandwidth connectivity does reach into regions where the population is largely illiterate, will we see the emergence of an audiovisual web, which could merge the spoken web with multiple visual and audio media? Only time will tell, but the possibilities are not remote: in India a little more than a decade ago there were no cell phones, and Web-enabled multimedia cell phones were a pipe dream. Wherever a cellular network exists now, the potential for a 3G/4G Web-enabled cellular network already exists. Unless a better technology emerges in the interim, an audiovisual web targeted at semi-literate and illiterate population segments seems to me to be inevitable.

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.

The future of computing and communications: random thoughts, part 2

I am testing a new software package I just bought called Dragon NaturallySpeaking, which I was lured into buying due to a substantial discount offered by a spam e-mail that I accidentally looked at -- a real success for the spam industry. It's turning out to be very successful at translating my voice into text.

I tried out voice-to-text software in the late 1990s, and decided then that it was not ready for prime time. This is one example of the kinds of interfaces and affordances I wrote about in my last blog post. If voice-to-text software is working as well as this seems to, and thus far I have had to make only a couple of corrections in this post, I can envision situations in which a keyboard is not feasible as an input device today, but where voice-to-text would be a very acceptable substitute, for example as an aid to surgery where no one in the operating room can spare their hands for a keyboard, least of all the surgeon.

I am still a complete novice at using Dragon NaturallySpeaking, which means I am still spending a lot of time learning how to use my voice to do a number of things that come naturally when using a mouse -- or at least naturally to someone who has used a mouse for as many years as I have. The program actually seems to work quite well at navigating through and selecting text in word processing programs, including Microsoft Word and the Firefox page where I am creating this blog post.

What I have written so far has taken me most of an hour, mostly due to the learning curve that is a natural part of learning an entirely new way of using a very familiar technology. What it reminds me of most, though I haven't done it in many years, is training a dog. In training a dog, part of what happens is you training the dog, and the other part is the dog training you. With respect to Dragon NaturallySpeaking, the bulk of the time is spent on the second part; that is, the dog is training me most of the time.

The most encouraging thing about the experience of the past hour or so is that Dragon NaturallySpeaking has failed to recognize only two words out of what I have written. Most of what I am learning is how to revise text as I write, and the vast bulk of the revisions are due to me changing my mind about what to say rather than errors on the part of the voice-to-text program. [I did find 3 or 4 more misrecognitions on reviewing the text.] When I watch closely what I do with the mouse under similar circumstances, I expect I will find that the ratio holds there as well; at this point, after using a mouse for more than 25 years, using the mouse is second nature to me. If I keep using Dragon NaturallySpeaking on a regular basis, and especially if I use it exclusively for some period of time, I suspect that its way of doing things will become second nature to me as well.

I don't think I will be able to use Dragon NaturallySpeaking on his regular a basis as I would like, though. I do a lot of work in circumstances where I am surrounded by other people, and where speech is not encouraged -- our open plan office at the University, for example, or at the public library. I can imagine using it in a coffee shop, but I suspect it will look very strange to the others around me. Then again, they already think I'm pretty strange, so maybe it won't matter. :-)

The future of computing and communications: random thoughts, part 1

I've been giving a lot of thought over the past several months to the future of information technology and its implications for society and culture in the 21st Century. We face a future in which we will be - already are being - deluged with toys, but which will also have - already does have - many dark facets as well.

This summer, when I started working on this post, I had been looking at the whole spectrum of social and technological changes afoot in the world, and realizing the momentous changes I had predicted in my 2002 Rogue Wave white paper have finally begun to gain momentum. 

This post, or the series of posts I hope it is going to start, is the reason I have been so quiet, and I apologize to my regular readers for this decrease in productivity. The problem is that the signs of change are so diverse and multitudinous that it is difficult to categorize them, and once categorized, to prioritize them. I got stuck on trying to wrestle this into a rational framework. As University of Michigan bioinformatics guru Dr. Brian Athey often says, "The perfect is the enemy of the good," and my months of near-silence are definitely an example.

As an example of my various dilemmas and conundra, I have an iPod Touch that came free with a MacBook I purchased over the summer. I am puzzled by it to no end. It's clearly innovative and has some downright revolutionary aspects, but it's hard to point out which feature or combination of features make it a harbinger of a technological sea change. I don't mean the iTouch in particular; it's really an iPhone without the phone, but with wi-fi instead of 2.5/3G connectivity. The iPhone OS is what they have in common, and that's got some revolutionary HCI (human-computer interaction) features, among them the first broad commercial implementation of a multi-touch gestural interface (the mouse and similar devices being the single-touch ancestor). But is the iTouch really just a dumber iPhone, or is it revolutionary in some way of its own?

The only way to sort through these ideas is to try to put them in writing, and that is the task I am setting for myself in this series of posts.  Let's start with a look at the future of HCI, the realm of user interfaces and affordances.

First let's consider this YouTube video of a new technology from Image Metrics, the tech wizards behind the latest and most controversial Grand Theft Auto game. It's less than two minutes long, and yes, you do want the sound on when you watch it, and yes, it's safe for work.

Do you begin to see some potential here?

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.

The semantic web | Start making sense | Economist.com

The Economist recently published an article called The semantic web | Start making sense.

SOME new ideas take wing spontaneously. Others struggle to be born. The “semantic web” is definitely in the latter category. But it may have found its midwife in Reuters, a business-information company.

The semantic web (or “web 3.0”, as some people are trying to re-brand it), is the name given to the idea that the pages of the world wide web ought to carry more than just the meaning they are intended to convey to the human reader. They should also, the thinking goes, be tagged and flagged in ways that machines can make semantic sense of, as people make semantic sense of language. That way, machines could make instant connections that would take serious amounts of time for people to see, or might even elude them altogether.

To this end, the web's übergeeks, the World Wide Web Consortium, have approved all sorts of snazzy acronyms that are supposed to help. The Resource Description Framework (RDF), for example, is supposed to standardise keywords, important dates and so on in a machine-friendly manner. The Web Ontology Language (OWL) will then pick these up and make sense of them. And if those don't work there are hCards, hCalendars, hReviews and other so-called microformat flags that will wave themselves to indicate where to look for various types of data.

It sounds a mess and it is. As a result it has been hard to persuade those who post web pages to include all the semantic-web stuff in their postings, too. Such marking up, as it is known, goes against the whole spirit of the web, which succeeded where similar ventures failed precisely because it was easy to use.

I'm happy to see a company like Reuters attacking the problem. This may be the factor that gives the Semantic Web the critical mass it needs to take off. The domains in which Reuters works are much simpler than the biomedical realm, and that's just where the first real successes can and should occur. As Lao-Tse is reputed to have said, "The journey of a thousand miles begins with a single step."