The document discusses using natural language processing (NLP) techniques like word2vec to analyze structured clinical data. Clinical encounters can be treated as "sentences" with vitals, labs, procedures, diagnoses, and prescriptions as "words". The author ingested clinical records into "sentences" and will use Spark's word2vec implementation on Hadoop to explore relationships between clinical concepts. The author is available for questions after demonstrating the approach on a dataset from a Kaggle diabetes prediction competition.