WellnessWay: A Unified MERN Stack Web Platform Coupling Real-Time Geospatial Healthcare Discovery with Machine Learning-Driven Symptom Analysis

Fragmentation between healthcare facility finders and AI-based symptom analysers forces patients to switch between multiple tools during medical decision-making—an inefficiency that WellnessWay is engineered to resolve. This work proposes a full-stack intelligent healthcare web portal constructed on the MERN (MongoDB, Express.js, React.js, Node.js) architecture, capable of simultaneously directing users to the most suitable nearby hospital or pharmacy and delivering machine learning-driven preliminary health assessments within a single browser session. Four supervised classifiers—Naïve Bayes, Decision Tree, Convolutional Neural Network (CNN), and Random Forest—were trained and benchmarked on the Kaggle Symptom–Disease corpus (4,920 samples, 41 target classes, 132 binary features). Random Forest emerged as the top performer, recording 91.3% classification accuracy and a macro-averaged F1-score of 0.89 at a mean inference latency of 185 ms, surpassing both the CNN (88.6%) and simpler baselines. Geospatial queries powered by MongoDB's 2dsphere indexing and the Haversine proximity formula resolved within 220 ms on average—roughly 35– 37% faster than comparable REST-only healthcare locator systems. A complementary rolebased dashboard lets hospital and pharmacy staff push live updates on bed occupancy, doctor rosters, medicine stock, and service availability, ensuring that end-users always consult current operational data. Post-deployment feedback from 30 trial participants showed that 86% preferred the co-located experience over separate single-purpose tools. Rigorous mathematical derivations, a fully parameterised CNN layer table, and comprehensive performance benchmarks are presented to support reproducibility and underline the practical readiness of the proposed framework.