Personalized Medicine currently focuses on the right treatment for the right patient, but its long-term goal includes personalized risk assessment and prevention. Current emphasis focuses on advances in genetic testing and biomarker to enhance patient care. The considerable data generated by such approaches and access to patient EHR’s has led to many statistically-based studies to predict disease risk and prognosis, e.g. the Gail model for breast cancer risk assessment; evaluation of BRCA mutation profiles; and expression level analysis of her2/neu for Herceptin response. Such correlative analysis has been used to enhance clinical-decision making but is limited in its potential for understanding mechanisms of risk and disease. We have extended these correlative approaches to include systems-based process modeling, extending from pre-disease risk to early detection, treatment and outcome, in an effort to develop models for testing and validation against both existing data and enhanced data collection. Development and testing of this approach in breast cancer will be presented. Risk Assessment We have begun to model risk assessment by including specific aspects of a patient’s physiological development to help identify risk factors that can lead to improved guidelines for risk prevention. This can also identify specific causes for risk and disease through the course of normal breast development. Risk from any specific factor is something that changes over a person’s lifetime and is not likely to be constant. Simple statistical correlation of a risk factor, i.e. do you smoke? do you smoke more than 1 pack of cigarettes per day? needs to evolve to show how risk from each factor varies over a person’s lifetime/stage of development. This is because the molecular processes underlying physiological change also vary. Methods Risk factors were identified from the literature (from RR=1.0 to >4.0) and compared with those in the Gail model and then a physiological model of breast development, from pre-menarche to menopause, was drafted. These included both those included in the Gail model as well as other biomarkers, e.g. breast-feeding history, radiation exposure, oral contraceptive use, etc. A data set that represented a combination of actual and simulated patients, with 1458 patients in each, was used for the analysis. Univariate analysis was performed and comparison between the Gail model results and our models was performed using ROC analysis. Subsequent refinement eliminated several variables from consideration in the model.