Background: Predicting disease progression or adverse health outcomes can be accomplished using insights gained from real-world observational data. Information collected routinely while providing patient care enables the development of risk models that identify patients with increased likelihood of disease or poor/costly health outcomes. Such models can improve patient management, especially in populations prone to such outcomes (e.g. type 2 or gestational diabetes, asthma, mental disorders). Here we describe the process of developing risk models, along with selected successful examples. Methods: Large nationally representative EMR, claims, and EMR-claims linked databases were processed to evaluate demographics, vital signs, diagnoses, diagnostic tests/results, procedures, insurance, and prescription details in disease-specific patient populations. Predictive models were developed and validated using multivariable regression techniques in these populations to identify key drivers of disease and health outcomes. Results: Risk models successfully predicted primary and secondary medication non-adherence and non-persistence in diabetic patients, with index medication type, history of non-adherence to other chronic condition medications, HbA1C level, and prior prescription fill patterns identified as key predictors. Similar models of medication non-adherence for schizophrenia and bipolar disorder patients identified age, substance abuse and concomitant psychiatric medications as key drivers, while models of hospitalization highlighted substance abuse or other psychosis diagnosis, concomitant use of psychiatric medications and history of non-adherence to antipsychotics as key drivers. Conclusion: Key variables that identify patients at high risk of adverse outcomes such as those identified in our examples can be easily ascertained during physician visits, and thus used to initiate disease management interventions. Building risk models into EMR or healthcare claims systems, and arming physicians or other healthcare decision-makers with the ability to facilitate their use at the point of care, has the potential to intercept disease progression and improve outcomes. Such analytics can help both providers and payers achieve effective disease management goals.