AI Driven Brain Tumour Detection using Integrated ML and NLP based Medical Diagnosis System

This study investigates whether an AI based diagnostic system that jointly analyses structured clinical indicators and natural language symptom descriptions can reliably support brain tumour detection. The methodology follows a multi stage pipeline: anonymised brain tumour datasets were collected, structured and text data were preprocessed, TF IDF features were generated for symptom narratives, and supervised models were trained, including Decision Tree and Random Forest classifiers for structured attributes and a Logistic Regression model for text input. The models were integrated into a three tier architecture with a Flask based web interface and evaluated on a held out test set using accuracy, precision, recall, confusion matrices, and ROC analysis. The integrated system achieved about 91.3% accuracy, with precision around 0.90 and recall near 0.89 for tumour prediction, indicating strong discrimination between tumour and non tumour cases in line with performance ranges reported for medical AI systems. The primary conclusion is that combining machine learning on structured indicators with NLP-based symptom analysis yields a practical, explainable decision support tool that can augment, but not replace, clinician judgment in brain tumour diagnosis.