Your Pager Beeps – PULL OVER: YOU WILL HAVE A SEIZURE IN 5 MINS
Posted by evolvingwheel on March 16, 2007
I couldn’t help being excited about the option – an invaluable opportunity to save lives and prevent collateral damages. You should check out this [article] in TechReview. John Guttag, head of the Dept. of Electrical Eng. and Computer Sc. at MIT, along with a student is in the process of developing an early warning system for seizures and other health risks by analyzing real time medical data from the body. The details of the development can be found in the text. I will talk a little bit about possibilities and risks associated with the commercialization of such detectors.
The system analyzes real time medical data from the body and feeds it to a highly sophisticated algorithm that processes the pattern. The tool is fed with several patterns that result to death or near severe outcomes. The system then matches the current diagnostics with possible red-alert patterns and tries to beep you about a highly probable circumstance. The idea in practice is noble. However, there are several factors that need to be considered.
In industrial early warning systems, data mostly respond to mechanical outcomes that are predominantly governed by linear factor based rules. Even in polynomial representations, the changes in conditions correspond to a graphical pattern that consistently extends to a proven outcome. However, in human body, the response is dependent on individual human factors and unique physical conditions. If the tool records thousands of graphs and tries to match it with a sudden peak in sugar level or a clot in Mr. X, it may not necessarily lead to the same outcome. Again, on the contrary, the large sampling data and high probability of an event might correlate and the outcome could still be matched. The thin line separating a success from a failure should be considered diligently.
Next, the opportunity for computation is enormous. If there is a service running supercomputers and analyzing data and sends the result securely to the attached device right on time, it could save so many so much. But, would the services be privately available? Who will control the quality control for highly sensitive medical diagnosis? How will a consumer choose a service? Who will validate the consistency?
Well, I hope we have answers to all these issues soon and see such a product in the market!