Healthcare providers face significant challenges with monitoring and managing patient data outside of clinics, particularly with insufficient resources and limited feedback on their patients' conditions. Effective management of these symptoms and exploration of larger bodies of data are vital for maintaining long-term quality of life and preventing late interventions. In this paper, we propose a framework for constructing personal health knowledge graphs from heterogeneous data sources. Our approach integrates clinical databases, relevant ontologies and standard healthcare guidelines to support alert generation, clinician interpretation and querying of patient data. Through a use case of monitoring Chronic Obstructive Pulmonary Disease (COPD) patients, we demonstrate that inference and reasoning on personal health knowledge graphs built with our framework can aid in patient monitoring and enhance the efficacy and accuracy of patient data queries.