Neural networks are specialized artificial intelligence techniques that hav
e shown high efficiency in dealing with complex problems. Paradigms such as
backpropagation have been successfully applied in a number of biomedical a
pplications, but not in attempts to identify women at risk of postmenopausa
l osteoporotic complications. In this paper, several neural networks were t
rained using different combinations of biochemical variables as inputs. Bon
e densitometric measurements in Ward's triangle and in the spinal column we
re used as separate classification criteria (outputs) between slow and fast
bone mass losers. The most parsimonious model with the best performance in
cluded plasma concentrations of estrone, estradiol, osteocalcin, parathyrin
and urine concentrations of calcium and hydroxyproline (expressed as ratio
to creatinine excretion) as input neurons; ten neurons in a single hidden
layer; and one neuron in the output layer. Diagnostic efficiency was 76 % i
n Ward's triangle and 74 % in the spinal column; sensitivity was 70 and 81
%, and specificity was 77 and 65 %, respectively. Linear discriminant analy
sis showed a diagnostic efficiency of 66 % in Ward's triangle and 64 % in t
he spinal column, sensitivity was 55 and 86 %, and specificity was 75 and 1
3 %, respectively. We conclude that performance of the stepwise discriminan
t analysis was not superior to the neural networks.