ADVANCING THE UNDERSTANDING OF CLINICAL SEPSIS USING GENE EXPRESSION–DRIVEN MACHINE LEARNING TO IMPROVE PATIENT OUTCOMES