Developers and educators have explored many different ways to create "Virtual Patients" as a method to simulate a patient encounter. Some of these attempts have been educationally useful, yet no approach taken to date has satisfactorily replicated the Patient-Doctor encounter in a way that can be generalized nor have the best developments to date been readily author-able by regular medical educators. The best simulator to date is the human standardized patient actor, which has considerable disadvantages. The manner in which a virtual standardized patient can be designed requires a breakdown of the clinical encounter into components and a strategic approach to simulating each phase. These components are compared to find the optimal approach for each part of the medical encounter. The paper proposes a blend of an artificially intelligent statistical matching dialogue system with multiple choice state machine-based sub-conversations as a way in which one may richly simulate the interview and counseling phases of the clinical encounter. Also elucidated are the steps necessary for educator author-ability and approaches that will extract rich, objective assessment data. If such integration proves to be successful, the result will be a rich conversational clinical simulation that closely approximates Patient-Doctor encounters.