home email us! sindicaci;ón

Making Artificial Intelligence “Real”

by Scott MacStravic

I have copied, read, annotated and filed hundreds of analyses and evaluations of proactive health management (PHM) in the past ten years.  These have come from academic and research organizations, customers of PHM programs, the suppliers thereof, government studies, and virtually every possible source likely to conduct such analyses.  And the thing they almost all share is a myopic focus on a small portion of the dynamics of PHM effects, from the beginning to the end of the “causal chains” involved in achieving them.

Part of this myopia is simply practicality.  I still recall a story in the cartoon strip “Prince Valiant” where the hero visited a land where all its people could and did consider the entire set of consequences of everything they thought of doing – from now until forever, across the entire universe of things affected, and therefore could never do anything – caught inevitably in “the paralysis of analysis”.

Moreover, evaluating everything, or even just most of the important things that happen, during and as a result of PHM interventions, costs money.  If it comes to the point where evaluating the results and returns on investment (ROI) from such interventions threatens the possibility of achieving positive ROI, then it becomes a case where: “If ignorance is bliss, ‘tis folly to be wise”.

But we have made enormous strides in artificial intelligence (AI) in recent years.  I have visited and listened to the people at MedAI, in Orlando, Florida, for example, who keep winning international awards for their AI and predictive prowess, which includes predictive modeling that is used in analyzing healthcare expenditures for targeting PHM interventions.  The ability of AI to capture the interactions of hundreds of factors means that computer intelligence, at least, is perfectly capable of considering complex the complex systems that affect health and healthcare expenditures.

Ray Kurzweil, in his book The Singularity Is Near: When Humans Transcend Biology (Penguin 2006), notes that the rate of information technology growth is exponential vs. linear.  Our capacity to identify and understand systems is doubling every decade, and artificial intelligence will transcend the storage and analytical capacity of the human brain before the middle of the 21st century.

One of the health-significant examples in his book is the identification of the human fat storage gene, already identified.  He predicts that AI will provide the basis for the creation of drugs that will control the operation of this gene within five to ten years.  This will enable us to eliminate the overweight/obesity epidemic that is already sweeping most of the world.  In turn, this will dramatically reduce the rates of diabetes and related conditions caused by weight problems, for which linear predictions have already been made that foresee the swamping of the sickcare system because of weight problems.

The challenge in PHM is to identify the chain of cause and effects in operation therein – from PHM intervention to human behavior change to physiological change to reduced sickcare costs, worker absences and presenteeism, to improved productivity and performance.  Armed with data on each factor in this complex system, we will be able to understand and predict the meaningful effects of PHM interventions on individuals and populations.

In turn, this will enable us to not merely evaluate past PHM interventions, but to predict which existing interventions, and simulate new interventions, in order to pick the most effective and promising.  We will have the ability to achieve evidence-based health, in the same way as we have developed evidence-based medicine (EBM) in the past fifty years, but in a far shorter time, thanks to the “law of accelerated returns” that Kurzweil describes.

The widely mixed and invariably myopic evaluations of PHM interventions up to now have typically focused on far too few years, far too few factors, and far to few dimensions of value to capture even what is already happening.  As Kurzweil noted, quoting the philosopher Schopenhauer: “Everyone takes the limits of his own vision for the limits of the world.”  With AI, we will no longer be so limited, but be able to take PHM to its full potential, instead of being stuck in narrow and equivocal evaluations of its current achievements.

1 Comment »

  Paul wrote @ July 27th, 2007 at 10:54 am

Very well spoken. We must be wary, however, of relying on AI lest our minds atrophy. AI should be used as a tool not a crutch. Also realize that although we may not be so limited in our thinking with AI as our guide but people still make decisions, good or bad, based on the recommendations provided.

Dr. David Crowder, a retired surgeon and consultant for Foundation Coal West, had the insight to recommend sending patients to the best medical facilities possible for the care of their employees rather then opting for the cheapest solution. Foundation Coal West reduced health care costs by over 2.5% by following Dr. Crowder’s recommendations. Too many companies, unfortunately, take the route of cheap health care which cost more in the long run. These people are the decision makers that will be deciding how to use recommendations from future AI systems.

Evaluation everything, taking into account all consequences, is the very conundrum faced by the myriad governmental and non-governmental organizations that are trying to create a uniform personal health record. Their goal is one standard and forcing every EMR and PHR vendor to conform to that standard. Is this the right direction? Maybe AI will answer that question for us some day. The makers of the EMRy Stick personal health record don’t think so. We have taken the approach of a universal translator which is capable to speaking the language of all EMR systems. Maybe someday an AI program will prove this decision correct.

Your comment

<a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <code> <em> <i> <strike> <strong>