A Data-Driven Predictive Machine Learning Model for Efficiently Storing Temperature-Sensitive Medical Products, Such...

Area of Science:

• Pharmaceutical Science
• Computer Science
• Engineering

Background:

• Maintaining precise temperature control is critical for pharmaceutical storage.
• Frequent opening of pharmacy refrigerators can compromise internal temperatures, risking medication efficacy.
• Existing storage solutions lack intelligent monitoring for temperature deviations.

Purpose of the Study:

• To develop a machine learning model for predicting temperature changes in multi-chamber medical refrigerators.
• To provide real-time alerts for pharmacists regarding the remaining time before temperature limits are breached.
• To optimize the use of multi-chamber refrigerators by suggesting less critical compartments for access.

Main Methods:

• A multiple linear regression model was developed using training data from a thermoelectric cooler-based fridge.
• The model predicts the time required for a specific compartment to reach its upper temperature limit upon opening.
• Model performance was evaluated using the coefficient of determination (R²).

Main Results:

• The developed multiple linear regression model achieved a coefficient of determination (R²) of 77%.
• The model accurately predicts the time remaining before temperature excursions in individual fridge compartments.
• The system can guide pharmacists to access compartments with more stable temperature profiles.

Conclusions:

• The proposed machine learning model offers a viable solution for developing intelligent multi-chamber refrigerators.
• This technology enhances the efficient storage of highly sensitive medical products by preventing temperature fluctuations.
• The smart fridge system can significantly improve medication safety and storage compliance in pharmacy settings.