Can you describe the types of patients who were tested and got into your training test?  Were they selected because they had an unknown apparently infectious condition?  If your current algorithm appears to perform as well in the validation set as it did in the training set, and you use the threshold defined by the Youden Index, what are the sensitivity and specificity?  How often would a patient at UCSF meet criteria that pushed them over the threshold where you might recommend using the test?