Does artificial intelligence help clinicians to recognize atrophic gastritis with thyroid disease?

The association of ABG with thyroid disorders (TD) was first described about 40 years ago.

These older studies assessed the association between Pernicious Anemia (PA) and Thyroiditis on the basis of gastric and or thyroid auto-antibodies. Only recently systematic studies have focused on this frequently overlooked association.

A research article to be published on January 28, 2008 in the World Journal of Gastroenterology investigates the overlooked association. A study was performed on the data set of 29 input variables (concerning anagraphical, life style, family and clinical history, biochemical and histological aspects) of 253 ABG patients.

The biochemical and ultrasonographic data on the diagnosis of TD were not included in the data set. Of these ABG patients, 185 were female (median age 54 [17-83] years) and 123 had pernicious anemia. TD was present in 135 patients (53.4%), and 118 (46.6%) had a healthy thyroid gland. In all the patients the presence or absence of TD was evaluated by biochemistry, ultrasonography and endocrinological evaluation at a single tertiary centre, and its presence or absence was considered as target variable.

The sample of ABG patients was randomly subdivided several times in two equal and balanced samples of subjects with and without TD; one for the training phase (testing) and one for the prediction phase (testing).

To reduce the number of input variables, selecting those most informative to predict the output, the T&T system, the IS system, as well as the TWIST protocol were used. T&T is a data resampling technique, based on an evolutionary algorithm developed by the Semeion Research Center, the Genetic Doping Algorithm (GenD). The IS system is an evolutionary wrapper system, also based on the GenD, able to reduce the amount of data, while conserving the largest amount of information available in the data set. In the TWIST protocol, the T&T and IS systems run in a parallel way to reach a situable representation of variables and optimal sample size when dealing with complex and non-linear problems.

The results emerging from the study suggest artificial neural networks are able to predict, with a good accuracy, the presence of TD in ABG patients, by using clinical and gastric biochemical and histological variables. Indeed, the optimized ANN yielded an accuracy of 76%, correctly identifying 82% of ABG patients with TD, outperforming the previous ANN models as well as the traditional linear models.

This study did not want to emphasize that advanced statistical decision support systems should replace or substitute experienced clinicians, but to underline that these systems should be viewed as a potential decision aid to better address investigations to save costs and to use resources when effectively necessary.