Clinical Decision Support Systems to Predict Drug-Drug Interaction Using Multilabel Long Short-Term Memory with an...

Area of Science:

• Pharmacology and Bioinformatics
• Artificial Intelligence in Healthcare
• Big Data Analytics

Background:

• Drug-drug interactions (DDIs) pose significant risks in healthcare, especially when multiple drugs are co-prescribed.
• Traditional machine learning (ML) methods for DDI prediction often struggle with generalization due to reliance on handcrafted features.
• Deep learning (DL) offers advanced capabilities by automatically learning features from complex data, improving prediction accuracy.

Purpose of the Study:

• To develop an advanced computational technique for accurate prediction of drug-drug interactions (DDIs).
• To enhance healthcare decision-making through superior DDI forecasting in big data environments.
• To improve the safety and efficacy of drug discovery and prescription processes.

Main Methods:

• A novel sparrow search optimization with deep learning-based DDI prediction (SSODL-DDIP) technique was developed.
• A multilabel long short-term memory with an autoencoder (MLSTM-AE) model was employed for DDI prediction.
• A lexicon-based approach was integrated to determine the severity of identified DDIs, with sparrow search optimization (SSO) enhancing MLSTM-AE performance.

Main Results:

• The SSODL-DDIP technique demonstrated promising performance in predicting unknown DDIs.
• Experimental simulations confirmed the effectiveness of the proposed method.
• The technique successfully identified drug relationships and properties from diverse data sources.

Conclusions:

• The SSODL-DDIP technique offers a robust and accurate approach for DDI prediction in big data settings.
• This method has the potential to significantly improve patient safety and guide clinical decision-making.
• The study highlights the efficacy of combining deep learning with optimization algorithms for complex biomedical challenges.