A new method for prediction of the hospitalization period in ICU using neural networks

Introduction:APACHE (Acute Physiologic and Chronic Health Evaluation) score is a medical tool designed to measure the severity of disease for adult patients admitted to Intensive Care Units (ICU). However, it is designed based on the American patients’ data and is not well suited to be used for Iranian people. In addition, Iranian hospitals are not equipped with High Dependency Units which is required for original APACHE. Method: We aimed to design an intelligent version of APACHE system for recognition of patients’ hospitalization period in ICUs. The new system can be designed based on Iranian local data and updated locally. Intelligence means that the system has the ability to learn from its previous results and doesn’t need manual update. Results: In this study, this new system is introduced and the technical specifications are presented. It is based on neural networks. It can be trained and is capable of auto-learning. The results obtained from final implemented software show better performance than those obtained from non-local version. Conclusion: Using this method, the efficiency of the prediction has increased from 80% to 90%. Such results were compared with the APACHE outputs to show the superiority of the proposed method.