A lesson from a course on medical software explains how algorithms make judgment calls under uncertainty. The instructor traces signal detection theory back to its radar origins before applying it to cancer screening software, laying out the four possible outcomes of a detection decision: true positives, false negatives, false positives, and true negatives. From there the lecture defines sensitivity and specificity and shows how shifting a detection threshold trades one kind of error for the other. The core teaching tool is the Receiver Operating Characteristic curve, walked through step by step, with Area Under the Curve introduced as a single number for comparing how well different algorithms separate real signals from noise. Aimed at students building or evaluating medical detection software, the lecture stays concrete throughout, using screening as its running example rather than abstract statistics.