Sepsis-III is a widely accepted medical framework introduced in 2016 to precisely define sepsis. It employs a clinical approach that emphasizes organ dysfunction resulting from an exaggerated response to infection. When organ dysfunction is severe and co-occurs with an elevated risk of mortality, it may be classified as septic shock. High-risk sepsis-III patients pose a challenging problem in accurately predicting mortality, necessitating the development of precise tools for timely intervention. This paper performs a comprehensive comparative analysis of machine learning techniques, including Logistic Regression, XGBoost Classifier, LGBM Classifier, and Deep Learning techniques like Artificial Neural Networks, primarily aimed at predicting mortality in sepsis-III patients. Leveraging the “Refined MIMIC-III 30-day mortality prediction of sepsis-3 patients” dataset from IEEE Dataport, the study systematically evaluates these models, with an emphasis on the deep learning model, Artificial Neural Network. Key metrics, such as accuracy and AUC Score, are utilized for a comprehensive assessment for the identification of the most effective model for mortality prediction. The LGBM classifier model gives the best predictive results providing accuracy 0.9 and AUC score 0.76. The significant contributions involve enhanced mortality prediction in sepsis-III patients, offering a comprehensive comparison of machine learning techniques. The work systematically evaluates models using key metrics, including accuracy and AUC score, to identify optimal predictive performance. The future refinement of study focuses on the ANN-based methods by combining PCA with hyperparameter tuning. This study provides a feasible approach for the betterment of sepsis healthcare systems.