Data-driven drug treatment: enhancing clinical decision-making with SalpPSO-optimized GraphSAGE.

Area of Science:

• Biomedical Informatics
• Artificial Intelligence in Healthcare
• Computational Pharmacology

Background:

• Adverse drug reactions pose significant risks to patient safety.
• Existing drug recommendation systems require enhanced accuracy in predicting drug-drug interactions.
• Integrating diverse clinical data is crucial for personalized medicine.

Purpose of the Study:

• To develop an innovative safe drug recommendation system.
• To optimize the GraphSAGE algorithm's hyperparameters using a hybrid SalpPSO approach.
• To improve the accuracy of drug-drug interaction prediction and personalize drug recommendations.

Main Methods:

• Constructed a knowledge graph integrating MIMIC-III, Drug Bank, and ICD-9 data.
• Employed the GraphSAGE algorithm for learning node embeddings within the knowledge graph.
• Utilized the Salp Swarm Optimization-based Particle Swarm Optimization (SalpPSO) for hyperparameter optimization of GraphSAGE.

Main Results:

• The SalpPSO-optimized GraphSAGE model demonstrated superior performance in predicting drug-drug interactions.
• Experimental analysis confirmed the efficiency and accuracy of the proposed safe recommendation system.
• The system effectively integrates patient, disease, and drug information for enhanced predictions.

Conclusions:

• The developed system offers a robust solution for minimizing adverse drug reactions.
• This approach aids healthcare professionals in making more informed and personalized drug treatment decisions.
• The integration of advanced AI techniques significantly advances patient safety in pharmacotherapy.