Identification Of Vitamin Deficiency and Recommendation of Rich Vitamin Food Using Machine Learning Techniques

Abstract

According to a WHO report, Insufficient and unbalanced food consumption, is to blame for 14% of deaths from gastrointestinal cancer. 11% of ischemic heart disease fatalities and 9% of heart attack deaths occur worldwide. Moreover, iodine deficiency affects 0.7 billion people worldwide, 0.2 billion due to an iron shortage (anemia), and 0.25 billion children, who among other things lack vitamins A through K. The primary goal of this initiative is to offer a personalised diet plan that takes into account each person's needs. The system in use uses a wide variety of info from numerous datasets..

To accomplish this goal, two datasets were prepared. The first data set was built on a variety of high and low vitamin levels, including those for vitamins A, B, C, D, E, and K. and the conditions of vitamins were further classified into normal and pathological conditions Labels were separated into the numerals 0 and 1, which correspond to normal and abnormal circumstances, respectively. By analyzing the interplay of different vitamins and their inadequacies, the second set of data was compiled. Then depending on the insufficiency of a particular vitamin, the appropriate nourishment was identified.

In this project,They included KNN, decision trees, random forests, logistic regression, and vote classifier, among other classifier algorithms. To enhance performance, an ensemble algorithm was implemented, combining several algorithms to create a new one. Each algorithm's accuracy was computed, and the most accurate one used for prediction. A Flask web application was used to present the predictions and suggest suitable food types for various combinations based on the diagnosed vitamin deficiency.