Voice disorders are a common health concern that significantly impacts individual quality of life. Early detection and intervention are crucial for effective treatment and management. However, existing detection methods face limitations in handling unbalanced data and limited feature integration. This paper presents a novel approach to voice pathology detection using a unique combination of acoustic features, including harmonic-to-noise ratio (HNR), energy, glottal features, mel-frequency cepstral coefficients (MFCCs) with their first and second derivatives, fundamental frequency, and patient information (age, gender, status). Feature selection was performed using a combination of Mutual Information and sequential backward selection (SBS), as well as SBS combined with the Spearman test. Random Forest, Logistic Regression, and Support Vector Machine (SVM) algorithms were employed for classification. The models were evaluated using the VOICED (VOice ICar fEDerico II) database. To address the class imbalance in the VOICED database, data augmentation was performed using the Saarbruecken voice database (SVD). Conducted expriments show that data augmentation significantly improved the performance of models. The Random Forest algorithm achieved the best results, with an accuracy of 86% and an F1-score of 88% on the balanced dataset. The results demonstrate the effectiveness of data augmentation in enhancing model performance. However, the feature selection methods employed did not improve the performance of most classifiers.