Development of a computational recommendation system for parenteral nutrition prescription: An algorithm study

Area of Science:

• Clinical Pharmacy
• Health Informatics
• Nutritional Support

Background:

• Growing demand for multichamber bags in parenteral support for chronic intestinal failure.
• Need to preserve aseptic pharmacy capacity.
• Complexity of potential product combinations (over 3000) for multichamber bag regimens plus additional IV fluids.

Purpose of the Study:

• To develop and test a computer-based selection algorithm (PNMatch) for multichamber bag-based parenteral nutrition regimens.
• To assess the algorithm's efficiency and accuracy in matching complex prescriptions.
• To evaluate the potential for improving parenteral support prescribing.

Main Methods:

• Development of PNMatch algorithm using a k-nearest neighbor strategy.
• Staged testing against clinician prescribing and existing prescriptions.
• Evaluation of algorithm performance on parenteral nutrition formulation requests.

Main Results:

• Successful processing of 20 parenteral nutrition formulation requests.
• Achieved a 90% exact match rate in prescription selection.
• Demonstrated a mean selection time of 116 seconds and high agreement (Cohen kappa 0.8) with existing formulations.

Conclusions:

• First successful development and testing of a computer algorithm for multichamber bag regimens in chronic intestinal failure.
• PNMatch shows potential to enhance efficiency in parenteral support prescribing.
• Algorithm may reduce variability in complex intravenous fluid regimen selection.