AI systems in clinical practice

Laboratory systems
Expert System for the interpretation of pulmonary function tests for patients with lung disease

PUFF was probably the first AI system to have been used in clinical practice.

developed by clinical domains keywords
Stanford University and Pacific Presbyterian Medical Center (Janice Aikins, John Kunz, Ted Shortliffe, Robert Fallat) Pulmonary functions Knowledge-based systems, Rules-based systems, knowledge representation, Pulmonary Consult
location commissioned status
Pacific Presbyterian Medical Center, San Francisco 1983 Commercial descendants of PUFF continue in use.
PUFF can diagnose the presence and severity of lung disease and produce reports for the patient's file.

PUFF was the first system developed using EMYCIN (Essential MYCIN, van Melle, 1979). It included the domain-independent features of MYCIN:
  • rule interpreter
  • explanation
  • knowledge acquisition.

    This provided a mechanism for representing domain-specific knowledge in the form of production rules, and for performing consultations in that domain. PUFF was originally written in Interlisp using the SUMEX-AIM computer facility and had to be rewritten in BASIC before it could be installed at PMC. PUFF does not require direct interaction with a physician, avoiding the human engineering problem.

    Knowledge Representation: there are 75 clinical paramenters (representing pulmonary function test results such as "total lung capacity" and "residual volume").

    Control Structure: PUFF is primarily a goal-directed, backward chaining system employing some 400 production rules. If it cannot conclude the value for a parameter, it asks the user. A pulmonary physiologist reviews the PUFF report, and if necessary modifies it on-line before printing it out. The report was found not to require modification in 85% of cases. Modifications, where made, often consisted of comments such as "This is consistant with last visit".

    The PUFF basic knowledge base was incorporated into the commercial "Pulmonary Consult" product. Several hundred copies were sold in the 1980s and used around the world.

  • references
    Aikins JS, Kunz JC, Shortliffe EH, Fallat RJ. PUFF: an expert system for interpretation of pulmonary function data. Comput Biomed Res. 1983 Jun;16(3):199-208.

    [PubMed]   []

    " The application of artificial intelligence techniques to real-world problems has produced promising research results, but seldom has a system become a useful tool in its domain of expertise. Notable exceptions are the DENDRAL and MOLGEN systems. This paper describes PUFF, a program that interprets lung function test data and has become a working tool in the pulmonary physiology lab of a large hospital. Elements of the problem that paved the way for its success are examined, as are significant limitations of the solution that warrant further study. "
    Kunz, J.C., R.J. Fallat, D.H. McClung, et. al., "Automated interpretation of pulmonary function test results". Proceedings of Computers in Critical Care and Pulmonary Medicine, IEEE Press, 1979. " "
    Snow, M.G., Fallat, R.J., Tyler, W.R., Hsu, S.P., "Pulmonary Consult: Concept to Application of an Expert System", Journal of Clinical Engineering 13:3, pp. 201- 205, 1988. " "

    contact links
    John Kunz
    Stanford University
     bullet  PUFF - Stanford Medical Informatics
    Stanford Medical Informatics

    Archive of AI systems in clinical practice previously administered by Enrico Coiera. Used with permission. Maintained and extended since 2001 by OpenClinical.

    Entry on archive: March 26 1993
    Last main update: November 26 2002
    Search this site


    Privacy policy User agreement Copyright Feedback

    Last modified:
    © Copyright OpenClinical 2002-2011