Disturbance in sleep has a significant contribution towards physical as well as mental health. Technology, and machine learning-based techniques used are not well-studied, and an accurate diagnosis cannot be found easily. In this study, it is identified whether an individual suffers from sleep apnea, insomnia, or no disorder by implementing various machine learning algorithms to a synthetic data. To correct data imbalance and improve performance, data augmentation and undersampling (SMOTE, or Synthetic Minority Oversampling Technique) were utilized. This work illustrates the potential for data augmentation to enhance machine learning models that are applied in medical diagnosis. It emphasizes how all research in the future needs to employ real-world medical datasets to validate these findings.