Design an Optimized Deep Learning Model for Liver Disease Prediction

Introduction: The incidence of liver disease resulting from excessive alcohol consumption, inhalation of polluted gases, drug use, contaminated food, and food packaging is rapidly increasing. As this disease often presents no symptoms or complications in its early stages, timely diagnosis can be challenging. However, if diagnosed early, treatment can be administered more easily and at a lower cost. In more advanced stages, liver diseases can progress to cirrhosis and liver cancer, leading to significantly more difficult treatment options and, in some cases, resulting in the patient's death . Method: Considering the challenges involved, early diagnosis of liver disease is crucial for determining treatment solutions, duration of treatment, and recovery. With advancements in machine learning and deep learning technologies, which can analyze and learn from complex and large datasets, these tools can be employed for the early prediction of liver disease. In this study, we utilized a dataset of liver disease patients from India. After data preprocessing, a deep learning model optimized by the metaheuristic algorithm of the Lizard Search was proposed to enhance prediction accuracy by leveraging the advantages of metaheuristic algorithms and to assist in timely disease diagnosis . Results: Our proposed method employs a deep neural network optimized using the metaheuristic RSA algorithm. This model achieved promising results in diagnosing liver disease, with an approximate accuracy of 96.9%, a reliability of around 97.2%, and an F1 score close to 96.7%. This approach demonstrates high efficiency in the early and accurate detection of liver diseases, contributing to improved treatment processes . Conclusion: The results demonstrated that combining a feedforward neural network with metaheuristic algorithms significantly improved the accuracy and reliability of liver disease diagnosis. The proposed model enhanced prediction accuracy and diagnostic validity compared to previous methods. These advancements can effectively facilitate the early detection of liver diseases, leading to improved treatment processes and a reduction in complications associated with disease progression .