Exploring antimalarial activity of drugs using weighted atomic vectors and artificial intelligence

Area of Science:

• Computational chemistry and pharmacology
• Drug discovery and development
• Bioinformatics and cheminformatics

Background:

• Malaria remains a critical global health challenge, with millions of deaths annually, necessitating novel therapeutic agents.
• The urgent need for potent antimalarial drugs drives research into innovative drug discovery methodologies.
• Machine learning (ML) offers a powerful computational approach to accelerate the identification and development of new antimalarial compounds.

Purpose of the Study:

• To predict the antimalarial activity of potential drug compounds.
• To evaluate the effectiveness of weighted atomic vectors in representing chemical structures for ML models.
• To compare the performance of various ML algorithms in predicting antimalarial efficacy.

Main Methods:

• Utilized weighted atomic vectors for the numerical representation of chemical compounds.
• Employed machine learning algorithms including Decision Tree, Bagging Regressor, and Ada Boost for activity prediction.
• Assessed model performance using R2, Mean Absolute Error (MAE), and Root Mean Squared Logarithmic Error (RMSLE) metrics.
• Performed statistical validation with Friedman and Wilcoxon Tests.

Main Results:

• The Ada Boost algorithm exhibited exceptional performance in predicting antimalarial activity.
• Ada Boost consistently outperformed other evaluated ML algorithms across diverse datasets.
• A maximum precision of 93% was achieved by the Ada Boost model, indicating high predictive accuracy.

Conclusions:

• The integration of weighted atomic vectors with machine learning presents a highly promising strategy for antimalarial drug discovery.
• Artificial intelligence significantly contributes to advancing antimalarial pharmaceutical research.
• This approach accelerates the identification of effective compounds, potentially reducing the burden of malaria.