washuravele/med-access-ai-project

MED-ACCESS is an AI-powered healthcare platform designed to address the shortage of medical doctors in South Africa. The AI solution targets healthcare by analyzing diseases and monitoring patients # MED-ACCESS: Disease Prediction Using AI ### Project Information * Course: `DIPLOMA INFORMATION TECHNOLOGY` * Module: `Business Analysis 3.2 Project` * Institution: `Vaal University of Technology (VUT)` * Group Name: `WECODE-04` * Year: `2025` ### Overview MED-Access is an AI healthcare system designed to help address the shortage of doctors in `South Africa`. The system predicts diseases based on patient symptoms using `machine learning` classification algorithms. By quickly analyzing symptoms, the model provides potential diagnoses that: * Support doctors with faster decision-making * Reduce patient waiting times ### Methods * **Dataset**: Symptoms (0/1 binary values) as features, Disease as the target column. * **Preprocessing**: Cleaning, duplicate removal, encoding diseases into numbers * **Algorithms Tested**: * Decision Tree Classifier * Naïve Bayes * Logistic Regression * Random Forest Classifier * LinearSVC * KNeighborsClassifier * **Evaluation Metrics**: * Accuracy * Precision, Recall, F1-score * Confusion Matrix * ROC Curve & AUC * **Improving Accuracy**: * GridSearchCV ### Results * Naïve Bayes a gave the best accuracy with `0.97`. * Evaluation showed improvements after data cleaning, encoding, and hyperparameter tuning (GridSearchCV). ### Repository Structure *** project-root/ |--- data/ `raw and processed datasets` |--- src/ `scripts for cleaning, training, evaluating, prediction` |--- model/ `saved trained models` |--- reports/ `results` |--- notebooks/ `EDA, preprocessing, and model testing` |--- docs/ `documentation (report parts)` |--- poster/ `poster drafts` |--- presentation/ `slides, graphs for presentation` |--- README.md `project documentation` *** ### Future Work * **NLP**: Allow patients to describe symptoms in text. * **Speech Recognition & Synthesis**: Patients can talk to the system, and it can talk back like a real conversation. * **Chatbot**: An interactive assistant that asks a …