Biometric authentication encompasses various methods that verify an individual’s identity by using their physiological or behavioral characteristics. Photoplethysmography (PPG) is a commonly used biometric feature in this authentication process. PPG technology monitors changes on the skin surface resulting from heartbeats or blood circulation. Machine learning-based systems are frequently employed to process and analyze data for biometric authentication. In these systems, various features and distance metrics derived from PPG signals can be utilized to authenticate individuals. Feature extraction involves the process of extracting meaningful information from PPG signals and providing this information as input to machine learning algorithms. The use of distance metrics in biometric authentication assesses the similarity or dissimilarity between individuals based on similarity or distance measures. These metrics can be employed to measure the similarity or dissimilarity between the features used in the authentication process. In this study, 25 statistical features were extracted from PPG signals. Additionally, 9 more features were derived from the same signals using distance metrics. These two authentication systems were trained using machine learning methods and evaluated based on performance criteria. The system trained with features derived using distance metrics achieved an accuracy of 80.38% according to the ensemble performance evaluation criterion. A hybrid metric was created by averaging the Pearson, Hellinger, and Matusita distance metrics, which yielded the highest accuracy. The system trained using 25 statistical features also achieved an accuracy of 86.1% with the ensemble method.