• KARRI VARSHITA
• K. VENKATESWARLU

Vitamin deficiencies are a growing concern in both developed and developing nations, contributing to long-term health conditions such as fatigue, neurological dysfunction, weakened immunity, and chronic fatigue. While clinical tests offer accurate diagnosis, they are expensive, slow, and not scalable in rural or under-resourced settings. In response, this paper presents an intelligent, machine learning–driven system that predicts likely vitamin deficiencies based on symptoms, lifestyle, and demographic factors. Built using an XGBoost classifier with class balancing via SMOTE, and model explanation using SHAP, the system ensures both predictive accuracy and interpretability. Once a deficiency is detected, the system suggests food items tailored to the user’s dietary preferences using data from USDA’s Food Data Central. The user-friendly interface, built using Streamlit, allows for real-time or batch predictions. The system achieved an F1-score of 91.8% and delivered highly relevant food recommendations verified against dietary standards. By providing a complete pipeline from diagnosis to diet the proposed solution empowers users to take proactive, affordable steps toward improved nutritional health.

Vitamin Deficiency, Machine Learning, Food Recommendation, Nutritional Dataset, XGBoost, SHAP, Preventive Healthcare, Symptom-Based Diagnosis, Streamlit Web App, SMOTE Balancing

"A Machine Learning-Based System for Vitamin Deficiency Detection and Personalized Food Recommendation ", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.12, Issue 7, page no.g344-g349, July-2025, Available :http://www.jetir.org/papers/JETIR2507651.pdf

"A Machine Learning-Based System for Vitamin Deficiency Detection and Personalized Food Recommendation ", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.12, Issue 7, page no. ppg344-g349, July-2025, Available at : http://www.jetir.org/papers/JETIR2507651.pdf