Untethered Healthcare

Portland, OR – Dr. Leroy Hood of the Institute for Systems Biology (ISG) gave the keynote speech today at Supercomputing 09 conference on the emergence of P4 (predictive, personalized, preventive and participatory) medicine. Dr. Hood has serious street cred when it comes to making predictions about the future of healthcare, having co-founded some twenty startups including Amgen. He believes that medicine will become primarily an information science in ten years with the convergence of personal genomics, systems biology and grid computing. These tools will help create billions of physiological data points for each individual, enabling the creation of predictive and actionable models to explain emergent behavior.

How does this vision translate to primary care?  According to Dr. Hood, you will first sequence  your personal and family genomes and then show up at your doctor’s office every six months for a wellness assessment.  Your physician will use a handheld blood assay device to profile 2500 blood proteins from a pinprick sample. The new assays will identify disease-related changes to organ-specific, blood protein fingerprints long before you become symptomatic. New therapies will stem from personal banking of stem cells and new drugs to “reengineer disease-perturbed networks.” “All the current business models will change,” says Dr. Hood. The time horizon for this vision becoming a reality? “Five to eight years.”

Given that the average timeline for a radical new technology to become ubiquitous in healthcare practice is seventeen years, I don’t know if I believe Dr. Hood’s timeline prediction. On the other hand, if he is right and the technology were available inexpensively today or in a few years time, wouldn’t you ask your physician for it?  Your thoughts?

Perhaps it is an unfair question. Any industry can be re-invented or even disrupted. That is the history of industry. Maybe the better question is what sectors of the medical device industry have become static and how are they likely to be impacted, not only by  changing how things are done, but also by the larger sea changes happening in healthcare. In some ways, the medical device industry today is similar to the mini-computer industry in the 1980s. In the 80′s, companies like Digital and Wang sold vertically integrated business computer systems, all based on proprietary hardware and software. Customers usually standardized on one vendor because systems from one mini-computer company were  incompatible with systems from another. They couldn’t  share data without a lot of custom development. Software was not portable between vendors and systems were expensive. Sound familiar?

In the mid-1980s IBM introduced the PC. Not only did it cost an order of magnitude less than a mini-computer, it had an open architecture with a published hardware/software specification. It was easy to modify and extend, write new software for and even replace hardware components. Design a better graphics card for it, write some custom CAD software and you could disrupt an industry like design automation, which was itself built on top of the old mini-computer silos. Digital and Wang and a host of other companies who depended on them never knew what hit them.

Hospitals today standardize on medical device vendors for some of the same reasons that people standardized on mini-computer vendors in the 1980s. Basically, they have few alternatives and plug-and-play between vendors doesn’t really exist yet except for HL7.  ( When Welch Allyn recently came up with the Flexiport cuff to connect a single blood pressure cuff to devices from different vendors, this simple plug-and-play idea was a huge success.  Nurses often waste time daily hunting through supplies looking for the right cuff to fit a particular device. ) Yet there is nothing preventing someone from developing a truly open PC-like architecture and toolset and radically lowering costs of both development and ownership.  Commoditization is already a fact of life in segments like patient monitoring. Every year at Medica there are several new monitoring companies with lower-price products. However, commoditizing only deals with price/performance. It doesn’t really change the game. What happens when someone develops the equivalent of the IBM PC for the medical device industry? In some ways this is already happening bottom-up in patient monitoring with many high-level monitoring parameters available from multiple sources as off-the-shelf modules. One trend is clear, regardless of whether the full open-source, medical PC shows up: the software content of devices and their ability to synthesize multiple data streams into actionable information for a variety of user skill-levels is going to become a lot more important as hardware becomes cheaper, smaller, wireless and more re-targetable. (After all, the era of the disposable patient monitor isn’t far away. )

Better  platforms and tools are only one part of the puzzle. Faster time to market is the other. Introducing Google’s concept of “fail faster to learn and succeed quicker” implies re-thinking the development process from the end stage backwards. Integrated design and test, much more sophisticated use of simulation, and re-thinking when and how to move under design controls are all elements of making the traditional development process more lightweight and Agile. Your thoughts?

Developing medical devices is traditionally seen as a linear process with little formal feedback between stages. The  FDA’s Quality System Regulation (QSR) even tends to enforce a linear, waterfall method of product development.  Traditional waterfall development can contribute to long development cycles, costly late stage defect discovery and even missing the mark on market acceptance. However there is nothing in the QSR that precludes adopting a more iterative, Agile method of product development, even in a regulated industry. If you think about it, all product development is iterative anyway. It just gets more expensive to find and correct your design mistakes and latent defects in later stages, but you still have to loop back and correct them. Most medical device companies’ quality systems grew up in the gate-style product life cycle era. Smart companies are re-thinking their internal processes to allow more Agile development. It has been my experience that most engineers feel more empowered and productive using Agile methods, even if it means very tight tracking of tasks, daily stand-up meetings and higher accountability. Real teamwork and real results always feel better than slipping schedules and development death marches.

Moving to Agile does not mean having to throw out all the existing SOPs and starting over. Agile is a set of practices, but you don’t have to adopt them all. In the regulated world, it is still best to fully understand the requirements before you start design. The most useful techniques can be adapted to regulated industry style development and phased in, often under the existing SOPs.  Even moving to daily scrums, project velocity tracking and enhanced collaboration tools can make a significant difference in schedule predictability. I’ll talk more about what this looks like in later posts. I’d appreciate any comments on other’s experiences using Agile methods in medical device development.