Related Experiment Videos
Clinical decision support systems for intensive care units: using artificial neural networks
M Frize1, C M Ennett, M Stevenson
1Department of Systems and Computer Engineering, Carleton University, Ottawa, ON, Canada. monique_frize@carleton.ca
Medical Engineering & Physics
|June 19, 2001
Summary
Artificial neural networks (ANNs) show promise for improving patient assessment in intensive care units (ICUs). Technical refinements enhance ANNs
Area of Science:
- Medical Informatics
- Artificial Intelligence in Medicine
- Computational Biology
Background:
- Artificial neural networks (ANNs) are increasingly applied to complex medical problems.
- The intensive care unit (ICU) environment presents unique challenges for patient outcome prediction and resource management.
Purpose of the Study:
- To evaluate the impact of various technical enhancements on the performance of ANNs in adult ICUs.
- To improve the accuracy of ANNs in estimating medical outcomes and resource utilization.
- To provide tools for medical and nursing staff to aid in patient assessment, diagnosis, and therapy selection.
Main Methods:
- Investigated the use of a weight-elimination cost function.
- Explored the utility of 'high' and 'low' nodes for input variables.
- Assessed the effect of varying the total number of input variables.
- Tested the influence of the constant predictor value on ANN performance.
Main Results:
- Specific technical approaches were evaluated for their effectiveness in optimizing ANN performance.
- The experiments aimed to quantify improvements in predicting patient status and resource needs.
- Findings provide insights into the practical application of ANNs in critical care settings.
Conclusions:
- The study demonstrates the potential of refined ANNs to enhance clinical decision-making in ICUs.
- These developments can support medical and nursing personnel in patient assessment, diagnosis, and treatment planning.
- Optimized ANNs offer a valuable tool for improving care delivery and resource management in critical care environments.