Survival Analysis of Heart Failure Patients through Logic Mining in Discrete Hopfield Neural Network

Survival analysis of a disease is vital to ensure the treatment of patients appropriately. Artificial Intelligence, particularly the application of Artificial Neural Networks, is garnering substantial interest within the healthcare sector due to its promising potential to enhance patient outcomes. This includes its potential to improve the diagnosis and management of various diseases, notably heart failure. Medical practitioners or specialists are not able to diagnose a disease accurately when there are several numbers of attributes/ risk factors that make the disease life-threatening for a patient. Logic mining is one of the platforms to induce significant attributes through the logical rules from large data sets. Boolean 2 Satisfiability logic can be used to represent real-life data sets incorporated with the Hopfield Neural Network. This study proposes a novel approach to extracting logical rules from medical datasets using the Modified Boolean 2-Satisfiability Reverse Analysis (MB2SRA) technique. The performance of the extracted rules is evaluated using RMSE, MAE, and accuracy metrics. Our computational simulations demonstrate the efficacy of the proposed model in predicting patient survival post-heart failure.