Healthcare generates over 2,314 exabytes of data annually, yet the choice between Electronic Medical Records (EMR), Electronic Health Records (EHR), and Personal Health Records (PHR) fundamentally shapes how organizations can leverage this data for analytics. This article examines EMR vs EHR vs PHR through a data science lens, analyzing their distinct data architectures, interoperability capabilities, and machine learning applications. EMRs offer deep facility-specific insights but limited analytics scalability. EHRs enable population health management and advanced predictive modeling through integrated multi-source data. PHRs unlock patient-generated behavioral data essential for precision medicine but face data quality challenges. We explore how data volume, variety, and quality differ across these systems, their respective machine learning use cases, and strategic implementation approaches. Healthcare organizations must understand these analytics implications to build data infrastructures that support AI-driven clinical decision support, real-world evidence generation, and personalized care delivery. The convergence of EMR, EHR, and PHR through API-first architectures and federated learning represents the future of healthcare analytics, where choosing the right system today determines analytical capabilities for decades to come.