Assisting the diagnosis of Graves' hyperthyroidism with Bayesian-type and SOM-type neural networks by making use of a set of three routine tests and their correlation with free T4.

In our previous paper, we proposed a novel screening method that assists the diagnosis of Graves' hyperthyroidism via two types of neural networks by making use of routine test data. This method can be applied by non-specialists during physical check-ups at a low cost and is expected to lead to rapid referrals for examination and treatment by thyroid specialists, that is, to improve patients' QOL. In this report, the amount of female sample data was increased and routine test data (14 parameters) from 120 subjects with a known diagnosis (35 patients with Graves' hyperthyroidism and 85 healthy volunteers) were adopted as training data, before 171 individuals who had also undergone the same routine tests at the Tohoku University Hospital were screened by the network for Graves' hyperthyroidism. The present re-examination of the screening method showed its high screening ability with the set of parameters used (low serum creatinine was added to the established measures of elevated alkaline phosphatase and low total cholesterol that appear in the Graves' hyperthyroidism guidelines) and robustness due to the increase of the training sample data. It was also found that there is a strong correlation between the three parameters and serum free thyroxine (FT4) in Graves' hyperthyroidism, which supports the usefulness of our screening method. Copyright 2009 Elsevier Masson SAS. All rights reserved.