Sequential versus concurrent computation of complex model systems for medical decision support.

Area of Science:

• Computational modeling in medicine
• Medical decision support systems

Background:

• Medical Decision Support Systems (MDSS) utilize mathematical models to predict patient responses to therapy adjustments.
• A prior framework combined respiratory mechanics, gas exchange, and cardiovascular dynamics models for comprehensive patient simulation.
• Concurrent computation of these submodels proved computationally expensive.

Purpose of the Study:

• To introduce a novel sequential computing approach for MDSS to reduce computational costs.
• To maintain prediction accuracy despite the shift from concurrent to sequential computation.

Main Methods:

• A sequential computing strategy was developed, calculating submodels individually.
• Interface signals were precalculated using reduced respiratory mechanics and cardiovascular dynamics models to simulate submodel interactions.
• The accuracy of the sequential approach was validated against the concurrent method.

Main Results:

• The sequential approach reduced computing costs by a factor of 17.
• Initial results showed a discrepancy of less than 2.5% compared to the concurrent approach.
• Improving precalculation of interface signals decreased simulation error to 2%.

Conclusions:

• The sequential computing approach offers a computationally efficient alternative for complex MDSS.
• This method maintains high accuracy in predicting patient responses to therapy alterations.
• Optimized precalculation strategies are key to minimizing simulation error in sequential MDSS.