PredictRx: AI based decision support tool for molecular screening for breast cancer drug recommendation

Area of Science:

• Computational biology
• Bioinformatics
• Drug discovery

Background:

• Breast cancer is a leading cause of mortality worldwide, necessitating novel therapeutic strategies.
• Identifying effective drug combinations is crucial for advancing breast cancer treatment.
• Artificial intelligence (AI) offers a powerful approach to analyze complex biological and chemical data.

Purpose of the Study:

• To develop PredictRx, an AI-driven decision support tool for healthcare practitioners.
• To aid in the analysis of drug combinations for breast cancer patients.
• To accelerate early-stage drug candidate identification and screening.

Main Methods:

• PredictRx utilizes molecular descriptors, physicochemical properties, and drug interaction datasets.
• The tool integrates six supervised and unsupervised machine learning techniques (Random Forest, SVM, Logistic Regression, K-Means, DBSCAN, Agglomerative Clustering).
• Model performance was evaluated using classification matrices and clustering metrics; the tool is a browser-accessible web application.

Main Results:

• Random Forest achieved the highest predictive performance accuracy (1).
• Agglomerative clustering demonstrated strong performance (Silhouette Score: 0.6946; Davies-Bouldin Index: 0.2457).
• Exploratory Data Analysis identified molecular weight, lipophilicity, and structural similarity as key factors in drug compatibility prediction.

Conclusions:

• AI-driven predictive modeling, as exemplified by PredictRx, significantly speeds up molecular screening for breast cancer drug discovery.
• The tool facilitates the identification of synergistic drug combinations, offering a scalable platform for AI-assisted pharmaceutical research.
• PredictRx integrates various AI techniques into an interpretable platform, enhancing drug compatibility and synergy analysis.