Machine Learning Methods in Non-Invasive Detection of Iron Deficiency

Anaemia is a condition in which there is a lack of red blood cells in the bloodstream. The goal of this study is to develop a model for predicting anaemia in human beings using Complete Blood Count results. Diagnosing anaemia usually involves examining the colour distribution of the palpebral conjunctiva; a blood test is not necessary for this process. Nevertheless, since individual colour perception varies, we try to mimic the physical inspection of the palpebral conjunctiva to identify anaemia. This way computers may automatically identify anaemia patients in a consolidated manner for a screening procedure. The approach employed in this study is separated into three phases: gathering the datasets, which is the eye (conjunctiva), pre-processing the picture, which includes segmentation of formican & palpebral area. We make use of a thresholding decision method that is based on the high hue rate (HHR) feature, which is taken from the HIS colour space. As such, diagnosing anaemia with low-cost, non-invasive techniques remains a difficulty. One can determine whether a patient has anaemia by looking at the pallor of their palms, fingers, nail beds and conjunctiva. In order to overcome the aforementioned issues, this paper presents a unique intelligent system that helps the automated detection of anaemia and is built on machine learning. As a result, we suggest an algorithm that adopts a more thorough methodology. It includes a wide range of features, two based on colour and one texture. To improve accuracy and dependability, it also used a more advanced classifier, like an artificial neural network (ANN) or support vector machine (SVM). Even though it takes more time and processing resources, this approach priorities precision and resilience, making it appropriate for applications where precise categorization is required.