DEVELOPING A MORE USABLE PHYSICIAN'S WORKSTATION

SOME PHYSICIAN INFORMATION REQUESTS, FOR EXAMPLE "WHAT TREATMENTS ARE SUITABLE FOR MY PATIENT WITH STAGE II BREAST CANCER?" CAN BE FILLED USING CONNECTIONS TO EXISTING MACHINE-READABLE SOURCES, SUCH AS PDQ, CANCERLIT AND THE TEXTBOOK PRINCIPLES & PRACTICE OF ONCOLOGY. WE CAN NOW MAKE SUCH CONNECTIONS USING A SPEECH-DRIVEN KNOWLEDGE-SERVER. THE CONNECTIONS MAKE USE OF INFORMATION OBTAINED FROM A COMPUTER-BASED PATIENT RECORD (CPR), AND THEY ANTICIPATE CERTAIN PHYSICIAN INFORMATION NEEDS. PHYSICIANS CAN USE SPOKEN COMMANDS TO SELECT FROM AND COMPLETE THESE CONNECTIONS. THE KNOWLEDGE-SERVER IS DESIGNED TO BE IMPLEMENTED USING AN "OPEN" STANDARD, AND TO WORK WITH OR WITHOUT SPEECH, AND WITH OR WITHOUT EXPLICIT INCORPORATION INTO A CPR. THE KNOWLEDGE-SERVER TAKES MAXIMUM ADVANTAGE OF UNIFIED MEDICAL LANGUAGE SYSTEM (UMLS) RESOURCES AND RELEVANT STANDARDS. TEMPLATES ARE USED TO REPRESENT THE ANTICIPATED INFORMATION NEEDS. MOST TEMPLATES ARE IMPLEMENTED AS PATHS THROUGH A TABLE OF CONTENTS (TOC) DERIVED FROM EACH SOURCE, AND EACH TOC IS MAINTAINED AS A UMLS ENHANCEMENT. OUR FOCUS ON THE APPROPRIATE USE OF INFORMATION FROM A CPR, AND ON ANTICIPATING PHYSICIAN INFORMATION NEEDS RESULTED FROM OUR STUDY OF TWO LONGITUDINAL PATIENT SCENARIOS. THE INFORMATION NEEDS GENERATED NY THESE SCENARIOS MOTIVATE AND SUPPORT OUR DESIGN.