A method and system are disclosed for computerized training, monitoring, certification or re-certification of human raters in clinical trials. The invention provides an interactive computerized interview that can be compared scores obtained by human raters to monitor and train clinical raters on a continuing basis in order to reduce rater drift and variance during clinical trials. Remediation can be automatically flagged or provided to improve a deviating rater's performance. Computerized assessment of symptom severity is provided without the need for human clinical raters. The system and method enable screening of prospective patients for inclusion or exclusion from a clinical trial by automatically obtaining computerized ratings of symptom severity.