Clinical problem solving--the role of expert laboratory systems

Area of Science:

• Medical Informatics
• Clinical Diagnostics
• Artificial Intelligence in Medicine

Background:

• Clinicians face challenges diagnosing conditions outside their expertise.
• Interpreting complex laboratory data, such as haematological and biochemical results, requires specialized knowledge.

Purpose of the Study:

• To evaluate a laboratory-based computer diagnostic program's ability to assist clinicians.
• To assess the program's effectiveness in interpreting emergency room data and suggesting diagnoses.

Main Methods:

• Utilized a computer program employing Fuzzy Sets and pattern recognition.
• Trained the program on a 10-year database of haematological and biochemical disease responses from a teaching hospital.
• Analyzed case studies from the New England Journal of Medicine and British Medical Journal.

Main Results:

• The computer program frequently proposed diagnoses overlooked by clinicians.
• In some cases, as few as three routine investigations were sufficient for diagnosis.
• Diagnostic accuracy could be enhanced through a structured approach to laboratory investigations.

Conclusions:

• Computer systems recognizing disease patterns from laboratory data can significantly aid diagnosis.
• This technology offers a valuable tool for supporting clinical decision-making, especially in complex cases.
• Further refinement can improve the integration of computer-assisted diagnosis into emergency room workflows.