DIABETES PREDICTION BASED ON BODY AND SYMPTOMS PARAMETERS

                                                                   ABSTRACT:

This study focuses on utilizing machine learning algorithms—Artificial Neural Networks (ANN), Long Short-Term Memory (LSTM), Random Forest, and XGBoost—to predict the onset of diabetes by analyzing two distinct datasets. The first dataset comprises clinical measures such as age, plasma glucose concentration, BMI, and family history, while the second dataset includes symptom-based features like polyuria, polydipsia, weight loss, and visual blurring. By leveraging these diverse datasets, the models aim to identify hidden patterns that indicate diabetes risk. The study evaluates model performance using key metrics, including accuracy, precision, recall, and F1-score, to ensure reliability. The results demonstrate that XGBoost and Random Forest exhibit strong predictive performance by effectively capturing complex relationships between clinical and symptom-based data. Meanwhile, ANN and LSTM offer additional insights into non-linear patterns, enhancing the model’s ability to detect subtle correlations within the data.

The primary goal of this research is to assist healthcare professionals in making early and accurate diabetes diagnoses, ultimately improving patient care and facilitating timely intervention. To enhance accessibility, the system will be integrated into a user-friendly web-based interface, enabling easy diabetes risk predictions. This approach fosters proactive healthcare strategies, empowering both medical practitioners and individuals to assess diabetes risk efficiently. Future work will focus on optimizing model performance through advanced hyperparameter tuning and integrating real-time data to improve predictive robustness.

Keywords: Diabetes Prediction, Machine Learning, ANN, LSTM, Random Forest, XGBoost, Healthcare, Predictive Models, Early Diagnosis, Clinical Data, Symptom-Based Data, Web-Based Diagnosis System.