Machine learning holds the potential to drive transformative advancements in drug discovery, healthcare decision support, and personalized medicine. Nevertheless, the complexity and lack of transparency in deep learning models create obstacles in comprehending their internal mechanics and building confidence in their predictions, thus impeding their widespread use in healthcare. Interpretable machine learning surfaces as a resolution to these obstacles, guaranteeing the reliability and credibility of models in critical scenarios by offering understandable explanations to human users. This dissertation aims to augment the interpretability of machine learning within healthcare contexts through the integration of domain-specific knowledge, thus ensuring that the interpretations are grounded firmly in clinical relevance. Aim 1 emphasizes domain knowledge synthesis and interpretation. We have developed BioSift for efficient extraction of domain-specific information from clinical trials. Additionally, TrialSieve was created for optimized meta-analysis of drug repurposing studies. We also offer a method to generate abstractive lay summaries for biomedical documents, aiding comprehension. Furthermore, CCS Explorer provides an end-to-end system for summarization and entity detection. Aim 2 focuses on enhancing interpretable model features through domain knowledge. We propose SeizFt, a feature-based approach for seizure detection, surpassing deep learning methods. We also present an interpretable model for risk assessment in pediatric acute leukemia, aligning with clinical guidelines. Aim 3 seeks to develop interpretable domain-aware representations from deep learning embeddings. We blend feature-based and deep-learning approaches with proposed algorithms that promote interpretability without sacrificing accuracy. Representative methods include SERF and NormIntSleep, demonstrating clinical interpretability in sleep staging.