Thursday, February 17, 2011

Dr. Watson treats a patient with back pain!

The "amazing" performance of the IBM Watson in the Jeopardy contest vs. 2 of the best past performers has lead to headlines like

  • Machine stomps on humanoids
  • Watson is creaming the humans
  • Ken Jennings himself wrote "I for one welcome our new computer overlords"

Yesterday we saw the news of IBM partnering with Nuance (the makers of Dragon) to develop a physician assistant.  The concept being this would be way to rapidly consider all the related texts, reference materials, prior cases, and latest knowledge in journals and medical literature to gain evidence from many more potential sources than previously possible.  


One can imagine a day in the future where patients will give their history over a phone to a computer that triages them to appropriate care.  So here is what might happen when some one with back pain calls in:








Kidding aside, here are some examples where a supercomputer could be useful as a physician assistant:
  1. Have all the EHR data for an individual be fed into the computer.  Then when a physician wants to know something like:
    1. Is there a relationship between the rise in creatinine and any medication or procedure? Or
    2. What could be the cause of the hyperkalemia?You could get an answer like creatinine increased 3 days after starting Lisinopril for example OR the pt developed non-gap acidosis and hyperkalemia and this is probably from a type IV RTA due to his diabetes.
  2. Have all the EHRs from major hospitals in a city be connected to the Computer.  Then the computer would know the prevalence and incidence of various conditions or the resistance patterns of microbes to antibiotics in that area.  You could then ask the computer to consider the pretest and posttest probabilities and likelihood ratios of various conditions and tests and spit out the odds.  This would help us in appropriate test ordering and medication prescriptions.
  3. Record my conversation with the patient, a video of my examination of the patient and list all my queries (that I ask the computer to do on the EHR) and create a note! 

Saturday, February 12, 2011

Why the NHS Couch to 5K Program Works!

The National Health Service in UK has hit a home run with their plan to get couch potatoes to run 5K in 9 weeks.  Why did it create such a buzz when hundreds of other sites have exercise plans and don't seem to be quite as popular?


Bandura wrote about self-efficacy and I summarized this recently.
The key point is that for a person to adopt a behavior like exercise to improve their health, 2 things must happen:

  1. They must believe that they can do the exercise
  2. They must believe that the exercise can help improve their health.
Thus, most couch potatoes will believe that if they could run 5K they would be healthier but they don't believe they can run 5K.  On the flip side most couch potatoes probably can walk 15 minutes slowly and they believe they can, but they probably believe that that amount of exercise will not make them healthier.  So they don't even start exercising.

Here is a short example of how a doctor might be able to use this information to help motivate a patient.  This is just for educational purposes and patients should consult their doctors before trying this out.


What the NHS program does is give people hope that anyone (who is reasonably healthy) can go on and run a 5K.  They provide a series of podcasts with music and commentary that guides the listener to walk and jog at appropriate times gradually getting them to jog for longer periods each week.  It seems to work because:
  1. There is no mention about how fast to jog/walk.  (I tell my patients who cannot jog to instead just walk as fast as they can when the podcast asks them to jog).
  2. The first 3-4 weeks are very achievable for most people and creates a sense of confidence, 
  3. There is a very active and supportive forum where people can see how others struggled like them and were able to succeed.  Some of the comments of people who were not sure they could get through a particular week, and were able to succeed are heartwarming and very encouraging for others.
It is quite possible that when folks get to week 4 or 5 they will already be seeing changes in their health, how they feel and that will create a cycle of success --> results --> self-motivation that will then be self-sustaining.

Big question is what will NHS come up with for those who complete the 5K?  Will it even be needed?  Will folks to struggled so hard to get to this goal, need ongoing motivation and tools to keep going?  Will they want to run faster, further, longer?  It will be interesting to follow the forums.    




Saturday, February 5, 2011

PowerPoint Bulleted List Theorem

Most of us associate PowerPoint with lectures in large darked halls with many slides of bulleted lists accompanied by a droning voice.  So what is wrong with the bulleted lists in PowerPoint?  I was preparing to conduct a workshop on PowerPoint and Education and found some theoretical basis for why we should not use bulleted lists when presenting.  As you see, I do use them freely in documents.
[This is presented as a theorem just to make it interesting.  It is just my very simplistic interpretation of work done by many people and better presented elsewhere - see references at bottom.  Readers should refer to the Atkinson,Mayer reference below for an excellent and more detailed description of these concepts.  My hope here it to get readers interested both the cognitive theories related to this topic and the practical applications of these.]

Compare this:
A bulleted list in PowerPoint
To This:



Statement:
Using Bulleted lists while narrating during presentations is detrimental to students' learning.

Axioms:

  • Working memory (formerly called short-term memory)
    • Processes incoming data/information
    • Connects it with existing knowledge/wisdom
    • Encodes it into long-term memory
  • Working memory has limited capacity to process information
    • It has 2 separate channels 
      • Verbal/auditory input
      • Visual input
    • Each channel has a limited processing capacity
    • Text is processed by both visual and verbal channels (you know now where this is going right?)
  • Meaningful learning requires substantial amount of cognitive processing in both channels 
    • Select and pay attention to incoming data
    • Organize the data
    • Integrate it with prior knowledge

Argument:
Information presented in a manner that overloads the processing power of the Working Memory makes learning difficult.
Bulleted lists which are multiple concepts presented as text are processed by both the visual and verbal channels.
When you start talking around these lists, the words you speak are processed by the verbal channel.
The audience struggles 
  • to correlate the text on the screen with the words you speak
  • to grasp which bullet you are talking about
  • to decide whether to just read the slide or listen to you talking
This struggle is not germane to getting a deeper understanding of the presented material.  It actually takes away from the learning process.

Conclusion:
Presenting (long) bulleted lists while narrating during presentations is detrimental to learning.

Quod Erat Demonstrandum!

Corollary1:
While narrating in a presentation, showing an appropriate image on the screen is better than showing a bulleted list.  This leverages the dual channels to facilitate learning.
Corollary 2:
Putting both an image and a lot of text on a single slide is detrimental to learning.  This can overwhelm the visual channel.

References:

  1. Working memory (Baddley and Hitch) [PubMed][Wikipedia]
  2. Dual Coding Theory (Paivio) [PDF article][Wikipedia]
  3. Cognitive Load Theory (Sweller) [EduTech Wiki]
  4. Select Organize Integrate Theory (Mayer)
  5. Cognitive Theory of Multimedia Learning (Mayer and Moreno)
  6. List of learning theories at Learning-theories.com
  7. Richard Mayer's full text article [PDF]
  8. 5 ways to reduce PowerPoint overload (Atkinson and Mayer)


Some of the biggest advantages of using tools like PowerPoint are:

  1. Use of Multimedia Elements
  2. Integration with Audience Response Systems
  3. Creating branching/non linear presentations based on audience needs
  4. More legible text, clearer images
  5. Ability to re-purpose/reuse material from other presentations (can be dangerous)
  6. Options for distribution and sharing
Thus, it is quite obvious, why PowerPoint (and other technology) is used so much in education.  It is very important to remember the steps for using technology in learning:
  1. Understand how people learn
  2. Think about how educators can Facilitate this learning process
  3. And only then think about how technology can help improve this facilitation process.  
When we keep these three steps in mind while designing our presentations, it will lead to better use of PowerPoint.  When the presentation "fails" it is most likely because we ignored one or both of the first 2 steps and jumped straight into the technology (PowerPoint).