Successful enterprise healthcare management involves making realistic estimates of clinical and support staff performance expectations and assignment of resources, equipment and facilities to meet the anticipated demand for services. A continuous training program is needed to maintain the quality of physician’s clinical decision making and to keep staff current with changes in clinical care to meet their performance goals. These training programs require investments of both time and money, which can be disruptive and burdensome, when an extensive training program is delivered indiscriminately across an organization. Many enterprise personnel management systems provide for personalized learning plans, but few organizations customize training schedules and content to the individual experiences and capabilities of physician learners on a continuing basis. A better understanding of individual differences in physician knowledge retention is needed to promote a more efficient and affordable approach to matching training to physician needs. This study examines de-identified retrospective data from electronic health records to assess physician performance when treating several specific clinical conditions. Diagnosis and management standards were based on clinical practice guidelines and consultation with experts in the field. This enabled the team to validate a generalized forecast model of performance decay based on time since previous encounter for each condition. A measure of individual deviation from the expected forecast model was estimated at each encounter. Individual score divergence was then regressed on variables including characteristics of the physicians that included recent exposure to similar encounters, performance on last encounter, workload, location of service and specialty. Results show how these characteristics were statistically significant (P<0.0001) in explaining some of the variation in individual performance. In conclusion, this paper proposes how these variables, which are available from electronic healthcare records and physician credentialing systems, may be used to inform a training management system that can dynamically personalize the training program for the healthcare professional.