Clinical decision support systems (CDSS) are designed to assist physicians and other health professionals with numerous clinical tasks, such as establishing a diagnosis or determining the appropriate course of therapy. While much research has been conducted in the field, with various degrees of success, little emphasis has been placed on unified knowledge representation with uncertainty and learning capabilities of a diagnostic system. Non-axiomatic logic (NAL), an Artificial General Intelligence project designed to realize general-purpose logic, provides consistent format for representing and reasoning-learning with knowledge that has diverse degrees of uncertainty. This paper reviews the methods used in design of CDSS, and proposes a CDSS framework based on NAL.