Coronary artery disease risk assessment from unstructured electronic health records using text mining

•Developed a rule-based text mining system to identify Framingham risk factors required for predicting coronary artery disease.•Discussion on performance of the text mining system.•Applied a systematic approach to understand missing data and subsequently chose appropriate imputation strategies.•Analysis of 10-year Framingham risk scores for coronary artery disease development in a cohort of diabetic patients.•Challenging issues in applying text mining in a clinical setting.