Therapeutic management based on protocols, and clinical algorithms is increasing in significance as medicine tackles more serious illnesses and must do so in cost-efficient manner. Clinical algorithms are appealing as they provide a good overview of the clinical state of a patient, identify key decisions and the outcomes associated with those decisions. They provide guidance and improve the quality of care without taking the control away from the physicians. Despite these advantages clinical algorithm design, specification, and delivery remain problematic. This research proposal seeks to overcome these difficulties, while retaining the advantages of clinical algorithms. We propose to develop a general framework for the representation for clinical algorithms and supporting medical knowledge, to communicate these algorithms effectively by extracting and presenting protocols specific individual patient's condition, and to provide decision aids for protocol- based patient care. This framework will use knowledge about therapy in the context of specific conditions to create patient-specific algorithms refined from the general-purpose algorithms. Aspects not pertinent to the case will be removed. The derivation of the specialized algorithms from the general algorithm will be recorded and used to explain the logic of the specialized algorithms. Because the algorithms will be tailored to the patient context, they will be detailed, yet manageable. This will avoid the over-simplifications of current """"""""one-size-fits-all"""""""" algorithm. Mixed graphic and textual presentation techniques will be used to develop a user-friendly interface with explanation facilities integrated with patient information systems. We will test the framework by developing operational systems for hemodynamic resuscitation and management of circulatory shock in critically ill postoperative patients. We have chosen the domain of the management of circulatory failure in critically ill postoperative patients because: (a) circulatory failure is the significant medical problem affecting many postoperative patients where proper management can mean the difference between life and death, (b) we have a long track record of on-going activities in development of objective treatment methodologies in this area, (c) we have developed and tested a general clinical algorithm for management of high risk postoperative surgical patients, (d) our recent attempts to extend and refine the initial algorithm using traditional approaches have been frustrated by the complexity of the task. The proposed framework will support the creation of detailed therapeutic management algorithms and decision support systems to implement these algorithms in daily practice.
