همایش بین المللی هوش مصنوعی

صفحه اصلی / دومین همایش بین المللی هوش مصنوعی

An interpretable framework based on deep learning and the Internet of Things for predicting risk dynamics in chronic patients

نویسندگان :

Behnaz Pouriayevali¹ َAsghar Ehteshami²

1- دانشگاه علوم پزشکی اصفهان 2- دانشگاه علوم پزشکی اصفهان

کلمات کلیدی :

Internet of Things،LSTM،deep learning،E-health،Explainable Artificial Intelligence(XAI)،chronic patients

چکیده :

Introduction: Monitoring chronic diseases requires models that are able to understand temporal patterns of physiological data. Traditional machine learning models often face limitations in this field and act as a "black box". Objective: This study provides an interpretable framework based on the Internet of Things (IoT) and deep learning for predicting health risk in chronic patients. Methods: Time series data were collected from the reputable PhysioNet database. A long short-term memory (LSTM) model was designed and trained to predict critical risk in patients. The performance of this model was compared with two baseline models (artificial neural network and random forest). In addition, the SHAP (SHapley Additive exPlanations) method was used to interpret the model predictions and identify the most important risk factors. Results: The LSTM model performed significantly better than the baseline models, achieving 95% accuracy and an area under the ROC curve (AUC) of 0.97. SHAP analysis revealed that sleep quality and sudden changes in activity level were the most important predictors of risk in the patients studied. Conclusion: The proposed framework not only improves prediction accuracy but also helps increase trust and acceptance of smart health systems by providing clinically meaningful interpretations.