With the advent of whole-genome DNA sequencing technologies, tailoring of medical treatment to individual patients based on their genetic makeup has become the vanguard of modern medicine. One such area that can benefit from individualized medicine is that of brain and other Central Nervous System (CNS) cancers. The prognosis of malignant brain cancers is among the worst due to the heterogeneity and complexity of these tumors and their micro-environment. We present a framework that combines data mining and machine learning techniques with semantic approaches for building a clinically-relevant knowledge base of brain cancer profiles. We construct clusters of patients based on the similarity of their profiles using the k-means clustering algorithm and extract relevant molecular attributes of these clusters to classify instances of the clusters. We create a semantic model with ontologies, rule checking and reasoning, to enable rational therapeutic regimen selection. Finally, we lay the foundation to incorporate this framework into a digital twin architecture of a patient.