Table 1 The comparative study of the current drug combination strategies and the applied dataset.
From: Harnessing machine learning to find synergistic combinations for FDA-approved cancer drugs
Author/year | Dataset used | Algorithm | Problem statement | Major contribution | Pros | Recommendation |
|---|---|---|---|---|---|---|
Sun et al.36 | DCDB 2.0 drug combination database42 | Semi-supervised learning | It exhibits an Anti-Cancer Synergy Ranking System (RACS) | They got a probability concordance of 0.78 | Synergy prediction and significantly minimize experimental prescreening of current pharmaceuticals for repurposing to cancer treatment | One prediction is confirmed in vivo in a zebrafish MCF7 xenograft model, suggesting great synergy and minimal toxicity. The approach was validated using A549 lung cancer cells |
Li et al.37 | Dream43 | Random Forest | The goal of synergistic drug combinations is to find new ways to treat cancer | Drug chemical structure, drug target network, and pharmacogenomics features all have an effect | Three of the 28 anti-cancer drug combinations identified by the prediction algorithm were effective | The use of a prediction model could assist in narrowing the search area and speed up the discovery of clinically effective synergistic medication combinations |
Xia et al.44 | NCI-ALMANAC45 | Deep Learning | Predicting the response of a selection of medication combinations in cell lines | Predicting tumor progression. While they achieve the greatest results using a combination of molecular feature types | Based on the model's expected combination, they rate the medicine combos for each cell line | They demonstrate promising results in predicting combinational medication response using deep learning |
Malyutina et al.39 | O'Neil46 | Elastic Net, Random Forest, Support Vector Machine | Combinations in cancer have been aided by high-throughput drug screening | Cross-design to medication combination sensitivity and synergy testing | S synergy score was created by comparing the dose–response of a drug combination to a single drug dose–response | Overall, they proved the effectiveness of combining cross-design with CSS sensitivity and S synergy rating |
Jiang et al.40 | O’Neil46 | Graph Convolutional Network | A combination of various networks to predict synergistic medication combinations | The GCN model for predicting synergistic medication combinations in specific cancer cell lines in this study. The GCN technique | Using a large heterogonous network, the GCN model was able to correctly predict cell line-specific synergistic drug combinations | Predicting and optimizing synergistic medication pairings in silico |
Liu et al.41 | O’Neil46 | Transformer boosted Deep Learning | Drug combinations have shown considerable promise in the treatment of cancer | An improved deep learning model for drug combination prediction that enhances performance and interpretability. Through gene–gene interaction, cell-line gene reliance, and genome-wide drug-target interaction | In comprehensive benchmark testing, TranSynergy beats the state-of-the-art approach | Identify biomarkers for precision medicine. For ovarian cancer, which has few therapy choices |
