Early detection of health-related events through a cost-effective predictive model is crucial in guiding treatment decisions and preventive measures. Machine learning techniques have been widely utilized in healthcare research. This study explores the effectiveness of a neuro-fuzzy inference system in diagnosing heart failure based on three key variables: age (in years), ejection fraction (percentage), and serum creatinine levels (mg/dL). The model is developed using a publicly available dataset comprising medical records of 299 heart failure patients from the Kaggle database. The dataset is divided into training and testing sets. A first-order Sugeno inference system with six layers is employed for model development. The Kolmogorov-Smirnov test is utilized to assess the normality of the estimated output values, while the Mann-Whitney test assesses whether the mean of estimated values significantly differs between patients who survived and those who did not. The model’s performance is measured using the area under the receiver operating characteristic curve (AUC), correlation coefficient, mean squared error (MSE), and mean squared logarithmic error (MSLE). Additionally, the extracted data and corresponding estimated outputs are analyzed individually. The development and performance assessment of the model are conducted using MATLAB and Python. The results indicate an AUC of 0.880, a correlation coefficient of 0.649, an MSE of 0.147, and an MSLE of 0.069. This study demonstrates the potential of a neuro-fuzzy inference system for predicting heart failure risk.