saving lives in the process Better DECISIONS through SCIENCE

agonizing over whether an ambiguous spot is a tumor. A parole board weighs the release of a potentially violent criminal. A technician at an airport worries over a set of ultrasound readings: do they suggest a deadly crack in an airplane’s wing? All these people are grappling with diagnostic decisions. In spite of incomplete or ambiguous evidence, they must determine whether or not a certain condition exists (or will occur). Such problems abound in health care, public safety, business, environment, justice, education, manufacturing, information processing, the military and government. And the stakes can be high. In many cases, a wrong verdict means that people will die. Perhaps surprisingly, the diagnostic decision-making process turns out to be essentially the same across fields. Hence, methods that improve the process in one industry can usually serve in others. At least two such methods are already available. Sadly, though, they remain unknown or unused in many realms. One increases accuracy, enhancing the odds that any given decision will be the correct one. The other improves the “utility” of a decision-making approach, ensuring that the number of true cases found does not come at the cost of an unreasonable number of false positive diagnoses (“false alarms”). These methods are statistical, but math phobics have nothing to fear; the basic logic is easy to grasp. No one is saying that diagnosticians Math-based aids for making decisions in medicine and industry could improve many diagnoses—often saving lives in the process Better DECISIONS through SCIENCE