VESTACKA NEURALNA MREZA ZA PREPOZNAVANJE b-TALASEMIJE MINOR I ANEMIJE UZROKOVANE NEDOSTATKOM GVOZDJA: MODELI VESTACKE INTELIGENCIJE: Artificial intelligence-driven diagnosis of b-thalassemia minor

Sažetak

Background and aims:

Iron deficiency anemia (IDA) and β-thalassemia minor (BTM) are the two most common causes of microcytic anemia, and though these conditions don’t share many symptoms, differential diagnosis via blood test is an a-time-consuming and expensive process. CBC can be used to diagnose anemia, but without advanced techniques, it cannot differentiate between iron deficiency anemia and BTM. This makes differential diagnosis of IDA and BTM a costly procedure, as it requires advanced techniques to differentiate between the two conditions.

This study aims to develop a model to differentiate IDA vs. BTM using an automated machine learning method using only CBC data. In this study, an  Artificial Neural Network-based system is recommended for differentiating the two states.

Materials&Methods:

This retrospective study included 396 individuals, consisting of 216 IDAs and 180 BTMs. The work was divided into three parts. The first section focused on the individual effects of hematological parameters on the differentiation of IDA and BTM. In the second part, traditional methods and discriminant indices used in diagnosis are discussed. In the third section, models developed using artificial neural networks (ANN) and decision trees are analyzed and compared with the methods used in the first two sections.

Results:

The study's conclusions are presented in three parts. The first part of the results suggests that MCV and RBC are the most effective predictors in discriminating between the two conditions.

The second part of the results suggests that the effects of discriminant indices on BTM and IDA differentiation were similar. However, using G&K and RDWI instead of other discriminant indexes for BTM and IDA greatly increases differentiation. The third section of the results reveal that machine learning models such as ANN are more powerful than traditional discriminant indexes.

Conclusion:

In conclusion, our results show that ANN method has higher performance than existing methods. Although other approaches have been effective, artificial intelligence can better predict the presence of various hemoglobin variants than traditional statistical approaches. This differentiation is important because it can have important medical implications on patient care, family planning, and health-related genetic counseling. The neural network model can also provide time-saving, less cost, and easier diagnosis.