I recruited Dave Dinatale to join me at the hack-a-thon a few weeks before the event. Amber Zimmermann heard about it and thought it would be fun to participate just a few days before the event. The morning of the event, several folks wandered over to our table as they considered joining our team, but it was a brave Joel Sutherland who decided to take a gamble on us.

We tossed around a bunch of ideas, but couldn’t quite decide what was an important enough problem to tackle. Eventually, we ended up on the HealthyPeople.gov website, which has set forth a list of national health objectives for the year 2020. Our eyes wandered to the list of New objectives, which included some targets related to sleep health.

A quick call to Dr. Mark Aloia provided us with some scientific leads on the relationship of sleep apnea to driving accidents, which lined up with Objective SH-2. The remainder of the day was spent iterating on searching for relevant data (BRFSS, FARS, CMS Medicare), thinking about how the datasets relate to one another, understanding the relevance of the data for patients and professionals, and – of course – programming. My personal highlight was finally learning how to mapreduce a big dataset using AWS. The result: “No Sleep Kills” at http://www.nosleepkills.org. Provocative, but we needed to get the public’s (and the judges’) attention!

In the end, it was a great honor to be selected as the winning team. There’s more work to do to polish the site and the data, but we were thrilled with what we were able to accomplish in just a few hours. Can’t wait until the next code-a-thon!

Interestingly, the level of social engagement was up this year over last. Using the number of retweets and replies as a measure of message amplification and person-to-person interaction, respectively, 2010 showed significant improvement. Retweets accounted for 24% of the total traffic versus 16% in 2009. Replies were up to 19% versus 11% in 2009.

These results suggest that the Twitter audience is growing at RSNA, and that audience is becoming more engaged in this form of social media.

Figure 1: Analysis of tweets containing the #RSNA10 hash tag, decomposed into tweets, re-tweets, and replies.

Figure 2: Top 15 tweeters (left), re-tweeters (middle) and repliers (right). Analysis of all tweets containing the #RSNA10 hash tag.

It turns out that Israel Railways only counts operating trains in its calculation. So, all trains that would have been traveling on the Haifa – Tel Aviv line were selectively taken out of the equation! “Lies, damned lies, and statistics.”

Motivation for this creative accounting? On-time performance is used by the Israeli government to set subsidies for Israel Railways.

A first jury convicted this individual, without bearing any significant risk (beyond a weight on their conscience) that their judgment may be wrong. A second jury then awarded damages that are paid for by the state. Once again, the jury bore none of the financial downside of their decision. That is, until the State of New York decides to raise the jurors’ (and your) taxes to cover the now larger budget deficit.

Conference organizers generated hype around Twitter and pre-announced that Twitterers should include the #RSNA09 hash tag to label their RSNA related tweets. A live feed of #RSNA09 tweets was posted prominently on the main RSNA conference web page. Yellow-shirted Twitter teams walked around the conference to promote the medium by drawing attention to themselves (yellow t-shirts in a sea of suits) and sponsoring Twitter-mediated giveaways. 

As a result, 163 Twitterers sent a combined 857 tweets in the nine days around the official four and a half days of the conference (see Figure 1).  The top 15 Twitterers, who generated over 55% of the total tweets, were overwhelmingly represented by commercial vendors and the RSNA staff (see Figure 2). 

Figure 1: Analysis of tweets containing the #RSNA09 hash tag, decomposed into tweets, re?tweets, and replies.

Figure 2: Top 15 tweeters, re?tweeters and repliers. Analysis of all tweets containing the #RSNA09 hash tag.

A lexical analysis showed that 141/857 (16%) of the #RSNA09 tweets were re-tweets, indicating that one person thought it worthy to repeat what another person had said. An additional 80/857 (11%) of the #RSNA09 tweets were directed at, or explicitly mentioned, another Twitterer. Using the re-tweets and referential/conversational tweets as a measure of social interaction among conference attendees and vendors, this analysis shows that there was only a moderate amount of Twitter-mediated social interaction at RSNA 2009 (27% of all tweets). 

Was Twitter an effective outbound marketing channel at RSNA09? Perhaps; the silent customers, including physicians and hospital administrators, who may have been following the #RSNA09 twitter feed, may have read vendors’ tweets and subsequently followed a hyperlink or visited their exhibit at the trade show. The entry cost for vendors to use this communication channel is negligible and their actions are immediately observable, which facilitates a rapid competitive response (see Figure 3). Yet, the modest level of social interaction suggests that the Twitter audience was not fully engaged by this social technology. 

Figure 3: Three large medical imaging vendors: cumulative tweeting by day at RSNA 2009. Philips and Siemens demonstrated a relatively consistent level of twittering, while General Electric had a surge of activity on the second day of the conference.

My use of the term “social technologies” is a conscious choice over the ubiquitous “social media”. The latter connotes a communications strategy aimed at pushing out a message, as the marketing function has traditionally done through newspapers, magazines, and television. The tools that enable social interaction on the internet are capable of offering companies much more than an outbound marketing channel. When used effectively, Web 2.0 enables individuals to come together as a community. I posit that Web 2.0 has flourished in settings where individual users perceive a direct benefit from their engagement, and where they feel empowered and free to interact. Companies that can create the right environment for their customers will reap much more benefit than can be achieved from “just another outbound marketing channel”; these companies will be able to harness their communities’ collective wisdom for driving innovation processes and rely on customer-to-customer interaction for word of mouth marketing and for solving simple service requests.

One of the key, fundamental differences between the old and new channels is directionality. Print and television are largely unidirectional, allowing vendors to disseminate a carefully crafted message that complies with FDA regulations. The internet made it easier for customers and vendors to engage in instant dialogues, which can still be controlled and made to comply with regulations on labeling. On the other hand, Web 2.0 social technologies bring together groups of people so that they can interact with one another. These tools benefit from network externalities, but as the number of engaged users grows, the resulting multilogues become harder, if not impossible, to control. Further complicating matters are tools like Google’s SideWiki, which allows a user to visit any site and to leave comments in the margins of the web page. The comments are stored on Google’s servers and are adjoined to the site’s web page only in the users’ web browser. The website owners have no control over the comments that visitors leave in the SideWiki [1] but a naïve third party may go to the company’s website, stumble upon the SideWiki discussion, and think that the company is somehow involved. Legal and regulatory teams fear that a visitor may discuss an unapproved application of their medical device, and that the FDA will hold the company responsible for off-label promotion. Unfortunately, the FDA has not provided guidance and companies can’t control and don’t have the resources to respond to every SideWiki comment on every page that may be left on their websites.

While incumbents stay operationally close to their mainstream customer physicians, as Bower and Christensen propose, medical technology firms do spend considerable strategic resources on a small minority of visionary and key opinion leader physicians. These firms recognize that the physician community is conservative in its adoption of new technologies, and that the evolution of patient care is initiated at academic medical centers before diffusing into the mainstream community, a process that can take 7-12 years. I would therefore contend that firms are adept at observing the emerging clinical trends, one of which may spark the next disruptive innovation. Whether a firm is organizationally capable of taking a calculated gamble and investing directly in one or more of these new technologies will ultimately depend on the culture and risk profile of the firm. Overall, though, I believe that, in the medical technology industry, the incumbent firm will worry less about being unseated by a new entrant, largely due to the significant entry barriers in this heavily regulated industry.

By definition, disruptive innovations in health care correspond to technologies for which there are no substantial equivalents. Therefore, companies must seek regulatory approval from the Food and Drug Administration (FDA) using the Pre-Market Approval (PMA) application process. Applicants must demonstrate the safety and effectiveness of their device for each specific clinical application (“indication” or “intended use”) using bench-top, pre-clinical (animal) and clinical (human) trial data.[1] This clinical trial work, per indication, can easily add three to five years to the already committed technology development time, at additional cost of $10-20 million per year.

Assuming that a firm was to obtain regulatory approval, it would then need to generate evidence to convince the Center for Medicare and Medicaid Services to issue a billing code for the new procedure and the third-party payers (e.g. Medicaid, private insurance firms) to reimburse physicians for performing the procedure. A positive coverage decision from insurers can easily take several more years for disruptive technologies, which are likely to be more costly, yet lack the long-term outcome evidence justifying their use over the cheaper existing methods. While a small venture-capital backed firm may have the financial resources to execute these steps, they will lack the experience and tacit knowledge required to successfully maneuver through the, often bureaucratic and political, regulatory and reimbursement systems, let alone scale-up for full production and distribution. Similarly, the new entrant’s financiers will want to see a return-on-investment on a time horizon which is shorter than the time needed to reach the mass market.

In contrast with Bower and Christensen’s example of the computer industry, the factors presented here pose significant entry barriers – in cost and time – for new entrants in the regulated health care industry. Time affords incumbents the opportunity to respond competitively. On the other hand, new entrants’ inexperience in navigating these obstacles, coupled with their investors’ desire to cash out, increases the probability that an incumbent, with the experience to commercialize the new medical technologies will take the opportunity to become a strategic partner, or owner, of the new entrant, as its disruptive technology matures and starts to threaten the incumbent’s existing technology markets.