Hypertension-Related Drug Activity Identification Based on Novel Ensemble Method

Area of Science:

• Computational chemistry
• Bioinformatics
• Machine learning in drug discovery

Background:

• Hypertension is a major risk factor for cardiovascular and cerebrovascular diseases, necessitating effective prevention and treatment strategies.
• Identifying hypertension-related active compounds is crucial for developing targeted therapies and improving patient health outcomes.
• Current methods for identifying active compounds may lack the accuracy and efficiency required for comprehensive drug discovery.

Purpose of the Study:

• To propose a novel ensemble method using a Flexible Neural Tree (FNT) for identifying hypertension-related active compounds.
• To evaluate the performance of the proposed FNT-based ensemble method against single classifiers and traditional ensemble techniques.
• To enhance the accuracy and reliability of compound classification in hypertension research.

Main Methods:

• A novel ensemble method was developed, incorporating a Flexible Neural Tree (FNT) as a nonlinear ensemble learner.
• Nine base classifiers, including Multi-Grained Cascade Forest (gcForest), SVM, RF, AdaBoost, DT, GBDT, KNN, logical regression, and Naïve Bayes, were used.
• The classification outputs of the nine base classifiers served as input features for the FNT model.

Main Results:

• The proposed FNT-based ensemble method demonstrated superior performance compared to individual classifiers across various metrics (ROC curve, AUC, TPR, FRP, Precision, Specificity, F1-score).
• Comparative analysis showed that the FNT ensemble method achieved higher accuracy in identifying hypertension-related compounds than averaged and voting ensemble methods.
• The experimental data, sourced from literature on hypertension-related and unrelated compounds, validated the effectiveness of the proposed approach.

Conclusions:

• The developed Flexible Neural Tree (FNT) ensemble method offers a highly accurate and effective approach for identifying hypertension-related active compounds.
• This novel computational strategy significantly advances the field of drug discovery for hypertension by improving classification accuracy.
• The findings suggest that FNT-based ensemble methods hold considerable promise for future research in medicinal chemistry and pharmaceutical sciences.