ARTIFICIAL NEURAL NETWORKS - A NEW DIMENSION IN THE EARLY DIAGNOSIS OF PROSTATE-CANCER

Artificial neural networks (ANN) are computer derived systems which ar e of increasing use in clinical medicine. The ANN uses multiple data i nputs for pattern recognition of a clinical event. We evaluated the Pr ostAssure (R) ANN for the detection of prostate cancer in men who have a serum prostate specific antigen (PSA) value of less than 4.0 mu L. The data inputs ape serum measurements of PSA, prostatatic acid, phosp hatase, creatine kinase, and the age of the patient, The ANN output is a probability index (PI), which when greater than 0.51 is suggestive of prostate cancer. The sensitivity (93%) and specificity (89%) of the PI in our study of 225 men tvas significantly better than the free/to tal PSA ratio (78% and 77%, respectively).