BioWatch, False Alarms, and Adaptable Alarms
Post: July 10, 2012 11:35 am
Author: Rafe Sagarin
Source: ETVOL Exclusive
This article from the July 7, 2012 Los Angeles Times
documents some of the vexing false alarms that have plagued the U.S. Department of Homeland Security’s multi-billion dollar BioWatch system. The problem of the type of alarm systems humans make, whether they are $5 home smoke detectors, million dollar tsunami warning systems, or billion dollar BioWatch systems is both the obvious—that they make a lot of false alarms—and the repercussions of those false alarms. Our brains cause us to turn them off, tune them out, or get more stressed worrying about them.
It’s estimated that 21 million American households have disable their home smoke detectors at some point. In the case of tsunami alarms, even the areas hardest hit by the 2004 Boxing Day tsunami soon stoned their new tsunami alarms to death because they produced so many false alarms. It seems from the LA Times article, local responders have learned to tune out the false BioWatch alarms—not too reassuring in the case there is a real alarm. And that leads to the third repercussion of false alarms: they cause a lot of stress, because for all the listener knows, they could be a real alarm, and if so they shouldn’t be ignored.
Dan Blumstein, a behavioral ecologist who also serves as an editor for ETVOL, has shown this in his favorite mammal, the marmot. Dan observes individual marmots and their behaviors in groups and finds that some marmots, which he calls, “Cool Hand Lucies” only make alarm calls when there is a clear and present danger. But other marmots, the “Nervous Nellies” make calls at the slightest hint of a problem—a snapping twig, or a sneezing biologist. Unfortunately, the other marmots can’t simply ignore Nervous Nelly, because she actually might be correct once in a while, and if they don’t look up from foraging, they might get caught by the real hawk or coyote that she was calling about. In this way, despite the frequent references in the Times article, false alarms are not like the Boy Who Cried Wolf.
False alarms are a function of the sensitivity of the alarms, and over-sensitivity can be buffered by resampling. In resampling for false alarms, you need both repetition and a different way of sampling. The Times article shows that even when some false BioWatch alarms were resampled four times they still gave false positives, which suggests that conducting the same test over and over is not enough to separate out a false signal (say, a harmless agent that appears or acts similar to the toxic agent).
The problem with some alarm systems on the flip side of sensitivity is their specificity. DHS health practitioners I’ve talked with confidentially tell me they also deplore the BioWatch system for being too fixed on a set list of “select agents” that have been deemed the most likely to be used in a bio-attack. Yet a good toxicologist knows that virtually any chemical can be harmful or harmless depending on the dosage—an overdose of water, otherwise known as drowning, will kill you as surely as an anthrax attack. Terry Taylor of the International Council for the Life Sciences (www.ilcs.org
) advocates a “full spectrum” view of bio-defense which acknowledges that there will always be some risk from biological agents, but the risk due to a deliberate terrorist attack is way out at the far end of a spectrum that includes naturally occurring chemicals, disease agents emerging through mutations, and laboratory accidents. Taking a full spectrum view comes closer to mirroring the natural adaptable capacities of biological organisms to deal with risk, something I’ve been exploring with Taylor, Blumstein and another ETVOL editor, Dominic Johnson (www.learningfromtheoctopus.com
In fact, organisms in nature show us that you can have alarm systems that are both wide spectrum and avoid false alarms. Animals prior to the Boxing Day tsunami did an exemplary job of warning of the coming tidal wave. Dogs refused to go for morning walks on the beach. Domestic and feral stock animals headed uphill. Elephants even tried to herd tourists away from shorelines. Unfortunately, too many humans didn’t know how to interpret these signs. As far as we know, these animals don’t have a specific tsunami warning gene. Rather they have all developed a multiply redundant resampling system that allows them to accurately sense dangerous changes in their environment without running uphill in a panic at a single tremulous seismic wave that might set off a human tsunami alarm. They have developed this system genetically--where they may have the capacity to sense a combination of vibrations, a change in air pressure, or magnetic fields—as well as environmentally, where a lifetime of careful observations of the world give each individual an ability to make sense of the myriad minute changes that add up to an approaching tsunami.
There is only so much we humans can do with our innate genetic abilities, but we find again and again that we can become better observers of the world and its changes. We can do this as individuals, by reversing the trend towards more and more screen time, and getting ourselves back into that ancient human tradition of natural history observation. And we can do this collectively through strong networks of observers. Such networks can’t perform the same function as a very sensitive chemical alarm, but they do provide multiple redundant ways to track potential biological agents across the risk spectrum.