Skip to content

Automated Care: Thermostat of Health or Ponzi Scheme?

September 4, 2012

The last post on the cHeath Blog was mid July.  I think that is the longest hiatus since I started blogging in March of 2010.  But the summer was great in New England. Lots of warm days and cool nights and plenty of sunshine. So as we close the books ‘unofficially’ on summer 2012, I’m back.

I also want to give a shout out to my colleague Gina Cella who was instrumental in thinking up one of the analogies used in this post (the Ponzi scheme).

Though I haven’t posted in six weeks, I’ve been thinking a lot.  One of the concepts that is occupying my mind is that of automated care.  The last time I wrote about automated care was February of 2011 (Emotional Automation, Revisited).  Lately I’ve been thinking about it more and more. The burden of chronic illness continues to rise and the size of the provider work force is not keeping up.  This manifests as overworked, unhappy providers, particularly in primary care.  

Sooner or later we’ll need to come to the conclusion that some of the care processes must be automated.  Other industries have done this and whether it is pumping your own gas or checking yourself in at the airport, we seem to like it just fine.  In fact the airline industry plans to take it to a new level.  But in healthcare, we don’t just employ one person where other industries have automated, we employ three or four to do redundant work.  We have a long way to go.

However, the idea that a software agent, or a robot might take on various aspects of your care tends to ‘creep out’ both patients and providers.  Maybe it’s because they fear that robots will run amuck as is illustrated in this amusing video

Yet, folks like my friend Tim Bickmore have capably shown that in some instances, patients actually prefer a software agent to a person.  There are several examples, in addition to Tim’s work, that show how software agents can indeed be caring.  Buddy, a virtual companion from the new company Geri Joy is one such example.  It’s a cuddly representation of a dog that appears on a tablet, but responds to voice and to touch in the way a real pet would.

Effective automated care includes feedback loops and emotional responses.  Feedback loops typically involve some measured parameter. The idea behind automated care is to send effective, caring messages to an individual based on the feedback loop.

A great example of a feedback loop we take for granted is the thermostat.  We’re all better off because of internal climate control whether it be in a New England winter or the summer in Abu Dhabi.  Thermostats work well and fade into the background. We don’t even realize they are working most of the time.

So it would be in the case of an effective automated care structure.  Some signal would come in from the patient’s remote monitoring device(s) and a caring response would go out to the individual.  When that process works well, the individual will be comforted and pleased, as in the case of Tim Bickmore’s relational agents or Geri Joy’s Buddy.  If, however, the feedback loop is not crisp, it would be more like the cleverbot video linked above.  Or like a Ponzi scheme, where there is a positive feedback loop with no control and eventual implosion of the system.

In a Ponzi scheme, ‘A’ produces more of ‘B’ which in turn produces more of ‘A,’ a classic example of a positive feedback system. The ‘A’ is profit and the ‘B’ is new investors. Profits are channeled back to new investors which, left unchecked, can lead to rapid growth towards collapse.

When we use information from the past to influence behavior in the future, that is feedback.  A heart failure patient eats pizza and gains three pounds, putting him or her at risk for a visit to the emergency room. Cause and effect. But, when we share that data with the patient, he or she will be more aware of the cause and effect, and will likely avoid eating pizza in the future.

All of you are leaders in creating the new generation of healthcare.  It is well recognized that we must act boldly. We have the capability of building systems that are more like Buddy or Tim’s nursing agent.  We must avoid building systems that are like the cleverbot example.

It’s in our hands.  Are you up to the challenge?

18 Comments leave one →
  1. September 5, 2012 3:53 am

    Welcome back, Joe, and I’m glad that you had good and thought-provoking summer weather in New England. Old England, alas, was not so lucky but we had good reason to cheer – and we still have, while the Paralympics reaches out to celebrate sporting achievements that testify to human endeavour.

    I mention this because, as my colleague and I saw when visiting a hospital in Sweden near the Arctic Circle, ‘ideas are born where they are needed most’. That’s a translation of the tag-line from their public services network in the Norbotten region and it’s hugely relevant because for half the year the place is seriously snowed up and the ability to visit hospitals and clinics is severely limited.

    Your observations on the general non-acceptance of automated health-care would seem very strange to folks and communities in this relatively isolated region but it is also true that the health workers include very human interfaces like Skype that can be invoked when required. It struck me that because of these pressures their health services have in many ways evolved into a far more efficient and effective system.

    But I was prompted to comment on your note for another reason – one that you allude to comparing health service innovations to the automation seen in other sectors. Your note should be required reading for all those academics and policy developers who are working on understanding the ‘digital economy’.

    These studies generally fall into two categories – those that focus on the economics of the firm and those that are focused on places, i.e. the smart/connected/mega/future Cities/Communities.

    This latter group debates endlessly about infrastructure provision but nowhere near enough about the essential motivations – what it should/could be used for and what work is needed to bring this digital economy to life.

    Here in the UK there’s been a recent reshuffling of government ministers. Jeremy Hunt, who who had oversight of broadband connectivity (and culture and sport), has been shifted to the Department of Health. Let’s hope that he takes whatever insight he garnered about digital infrastructure with him and applies it to the health of UK citizens wherever they might be living.

    • September 5, 2012 5:44 pm

      thanks for your very thoughtful comment, David and best wishes

  2. September 10, 2012 5:49 am

    There has been a fundamental shift, from a service perspective, in what used to be “what can you do for me?” to “what can I do for myself?”… DIY is all around us (ATM, Airline kiosks, RedBox, Coinstar, self checkout at the grocery store, etc)… Self service healthcare is at its infancy. Watch for this space to heat up as retail pharmacies enter this space, replacing their old, outdated blood pressure machines with new innovative health kiosks which are connected to local providers.

  3. September 26, 2012 11:44 am

    Thanks for the provocative framing Joe. We articulated the promise of a Health Advocate Avatar in 2004 and have seen many health leaders adopt the idea so it’s good to have you focus on the feedback needed as we automate some of the learning needed for smart self care. The knowledge revolution represented by technologies such as IBM’s Watson shows people can get answers to questions from machine based intelligence and thus become more intelligent in new learning and better decision making. Or not. The difference may reside at the wisdom level rather than in knowledge, which is why we need to introduce the question: what level of ethics will guide the automation of care processes? We’ve seen business ethics guide medical decision making (e.g., Gawande’s article) to see patients as profit centers.

  4. September 26, 2012 11:50 am

    Joe: I love your framing of the approach and that is exactly what we have done with our online version of the NIH’s Diabetes Prevention Program (DPP). We automated the protocol, curriculum, and behavior change approaches of the orginal in-person intensive lifestyle intervention so that instead of one coach supporting 30-40 patients over a 12 month period, one coach can support 500 patients! Our results are comparable to the original one-on-one intervention but at a fraction of the cost. A recent ROI study demonstrated a 4.3 to 1 Return on Investment within 24 month of enrolling in the program. Over 3500 patients at 25 different sites have used our program — called Virtual Lifestyle Management serivce (VLM). If anyone wants to learn more go to or contact me at 310 444 0636

  5. September 26, 2012 12:04 pm

    I think robotics and software needs to advance quite a bit before automated care can become a reality.

    In the meantime social media seems to be a way to provide support for folks living with chronic illnesses. The Diabetes Online Community isn’t a formal organization but many have been helped in a variety of ways by the DOC. This support isn’t the same as monitoring an illness, but I’ve seen examples of people being helped dramatically by online communities formed by people who live with the same illness. It’s a partial solution currently, but I think it will evolve faster than an automated one on one solution.

    • September 26, 2012 12:18 pm

      social media are an important tool, no doubt, as are other motivators such as games, incentives and coaching/reminders

  6. September 26, 2012 12:12 pm

    People don’t mind doing things for themselves. More importantly, if they can have things done for them “automatically” or in a time-saving way, that can be compelling. The kicker would be an automated solution which is so life-changing or empowering for a particular health issue that it becomes essential.

  7. Randy Williams permalink
    September 26, 2012 3:52 pm

    Automating the response of the healthcare system to the patient makes some sense, and as you point out, Joe, some patients prefer an automated agent. The challenge to your thesis, in my opinion, is whether we want the feedback loop to INVOLVE the patient actively or not (passive agent). If you believe, as Pharos does, that the patient is an integral actor in the loop, then the only way to drive behavioral change is to involve the patient as an active participant in the loop. A great example of this is the power of the “know your number ” campaign around cholesterol management. I worry that the concept of automation in connected health is being taken to the level of automating the patient out of the process (aka embedded sensors, for example). To use your analogy, I think controlling the temperature in the building is different from teaching the inhabitants to consume less fossil fuel.

    • September 26, 2012 5:00 pm

      Not surprisingly, we think alike. Consumer/patient involvement is critical. To me the data collection part should be automated (a la embedded sensors), but the engagement part (the know your number part) must be quite active.

  8. September 26, 2012 11:10 pm

    Reblogged this on lava kafle kathmandu nepal.

  9. September 29, 2012 4:48 pm

    (Disclosure – I am Diabetech’s co-founder & CEO) Diabetes Care recently published a peer-reviewed, randomized controlled clinical trial; “Integrating an Automated Diabetes Management System in the Family Management of Children with Type 1 Diabetes” in March of this year. ADMS is an automated approach incorporating proprietary wireless devices that began with a series of trials commencing in 2002 and includes nuanced feedback loops (analytics, games, surveys, etc…) for adults, kids, type 1, type 2 and gestational diabetes. The future has been here for quite some time. Getting the payment mechanism in place is what continues to elude progress for the masses. An index of these studies can be found here:


  1. Automated Care: Thermostat of Health or Ponzi Scheme? - The Doctor Weighs In | The Doctor Weighs In

Leave a Reply to Bernard FarrellB Cancel reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: