Machine Learning-Based Approach towards Identification of Pharmaceutical Suspensions Exploiting Speckle Pattern...

Area of Science:

• Biomedical Engineering
• Medical Diagnostics
• Pharmaceutical Sciences

Background:

• Parenteral artificial nutrition (PAN) is essential for patient care but incorrect drug administration poses severe health risks.
• Distinguishing between similar-looking PAN drug suspensions using basic optical methods is challenging.
• Accurate, real-time identification of PAN drugs before injection is critical to patient safety.

Purpose of the Study:

• To develop and validate a novel method for precise classification of parenteral artificial nutrition (PAN) drug suspensions.
• To leverage speckle pattern (SP) imaging combined with artificial intelligence (AI) for pharmaceutical analysis.
• To establish a new optical sensing platform for identifying critical medical treatments.

Main Methods:

• Acquisition of speckle pattern (SP) images from various commercial pharmaceutical suspensions used for PAN.
• Extraction of statistical parameters from the acquired SP images.
• Training and evaluation of machine learning algorithms (Random Forest and Multi-Layer Perceptron Network) for drug classification.

Main Results:

• The combined approach of SP imaging and AI achieved accurate classification of PAN drug suspensions.
• Machine learning models demonstrated high performance in identifying different pharmaceutical formulations.
• The developed method offers a reliable solution for distinguishing between visually similar turbid liquid drugs.

Conclusions:

• Speckle pattern imaging coupled with artificial intelligence provides a powerful tool for identifying parenteral artificial nutrition (PAN) drugs.
• This novel optical sensing platform enhances patient safety by enabling accurate drug verification.
• The study presents the first application of this combined technique for the specific identification of PAN drugs.