> Detecting whom exactly the next victim is going to be isn't really going to change anything as far as researching a treatment or cure goes.

Your reasoning relies heavily on this statement, which is only true if occurrence is entirely random, which is in most cases not true. A condition can easily mask the cause of the condition and then you have a confounder(-s) that you have no way of controlling. If you can build multiple strata with high risk ratios, you can find baseline similarities and differences in those groups. Early detection is highly important in knowing these confounders in the first place and then controlling for; and as GP mentions allows for more targeted research in treatment. Without this we could easily spend all the research effort on the effect (symptom) of a condition without even approaching treatment of the cause, i.e. prevention.

A very similar thing has happened with the infamous atherosclerotic plaques. AFAIK (correct me if you are aware of any evidence) there is currently no mechanistic model of how these atherosclerotic plaques form. Yet we spend so much effort in lowering the symptomatic side of increased cholesterol/LDL (which has well-known positives) even if there are known metabolic pathways for LDL increase, based entirely on correlational studies, when LDL is not even close to being the best predictor of cardiovascular conditions. LDL just happens to be easy to measure in a blood test and easy to control with oral medication.