A Machine Learning Model for Malnutrition Detection in Preschool Children

The use of artificial intelligence techniques has introduced a new dimension to the health-care sector, particularly in diagnosing diseases in their early stages, allowing patients and health-care workers to take early corrective action. The study discusses a machine learning algorithm created to detect and classify malnutrition in preschool children. Malnutrition affects a significant number of children in various nations, many of whom are preschoolers. Early identification of malnutrition in a preschool child will allow parents to monitor the child's health. The model also proposes dietary plans based on the type of malnutrition diagnosed. Introduction -Between the ages of one and five, children start to develop new diet habits, often influenced by exposure to new tastes and market foods. It can lead to health issues like obesity or insufficient nutritional intake. Malnutrition can arise from food that does not meet a child’s nutritional needs or from an imbalanced diet. Early identification and treatment are crucial to avoid long-term health problems affecting the child’s overall development. A well-balanced diet is vital to prevent diseases and chronic health conditions. Parents try to provide adequate nutrition, but deficiencies in early development stages can cause stunted growth, weakened immunity, and other issues. Objectives- The primary aim is to develop a machine learning model that can detect and classify malnutrition in preschool children early. By predicting malnutrition, the model assists parents in taking preventive actions and managing their child’s health effectively. Additionally, the model proposes personalized dietary plans based on the diagnosed type of malnutrition. It also emphasizes providing a simple, questionnaire-based system for remote or rural parents, making healthcare advice easily accessible. Methods –The development of model was like Exploratory Data Analysis (EDA), attribute selection, model selection, model training, and evaluation. Initially, twenty-eight attributes were shortlisted based on clinical and nutritional research, and then reduced to twenty-one essential attributes. Various machine learning algorithms were tested using the WEKA tool, and Logistic Model Tree (LMT) was selected for its high accuracy. The model was trained with 70% of the dataset and tested with 30%. A two-phase model implementation was done to first detect malnutrition and then to classify the type (undernutrition or micronutrient deficiency). Results The Logistic Model Tree (LMT) achieved an accuracy of 91% in detecting malnutrition. A two-step diagnosis improved the detection of undernutrition and micronutrient deficiencies. Grouping attributes related to medical and dietary factors further increased the accuracy to 95% in the second phase. The model successfully classified preschool children as healthy, undernourished, over-nourished, or suffering from micronutrient deficiencies, and suggested appropriate dietary plans based on the diagnosis​. Conclusions The application of machine learning models in healthcare supports early intervention and better health outcomes. The developed model accurately predicts malnutrition in preschoolers and provides a user-friendly platform for parents, especially those in remote areas. By suggesting customized diet plans, the model helps parents manage their child's nutritional needs effectively, reducing the risk of long-term health problems.