This review shows where AI is already delivering real value in drug discovery: not just faster predictions, but experimentally validated hits, better ADMET screening, large-scale virtual screening, and even early human translation. It maps which AI advances are truly credible, which claims remain fragile, and where the field is genuinely moving from hype to practical therapeutic impact.
Abstract: The aim of this review is to evaluate the efficacy, accuracy, and utility of AI-driven methodologies in drug discovery, focusing on target identification, molecular generation, pharmacokinetic prediction, and clinical trial forecasting across diverse therapeutic areas. The review utilises 516 original studies with 5027003 total participants (topic deduplicated ΣN). The mapped evidence suggests that AI now has a substantive role across the drug discovery continuum, with prominent signals in target identification, ultra-large-scale virtual screening, generative molecular design, repurposing, and absorption, distribution, metabolism, excretion, and toxicity prediction. Particularly notable findings included 79% accuracy for forecasting Phase II clinical trial outcomes, an AUROC of 0.88 for blood-brain barrier permeability, 90.4% accuracy for cytochrome P450 inhibition, and a 68% hit rate in one AI-designed liver X receptor agonist program. The evidence map also indicates that the most credible advances are those coupled to biological validation, including organoid, zebrafish, murine, and early human studies, which supports a practical role for AI in prioritizing compounds, reducing attrition, and accelerating repurposing and lead optimization. At the same time, the literature remains methodologically heterogeneous, and several studies highlight that benchmark choice, data curation, interpretability, and realistic validation strongly influence apparent model performance. Overall, this scoping review supports a role for AI as an enabling and increasingly translational layer in drug discovery, while future research should prioritize standardized external validation, explainable multi-objective models, and prospective human-in-the-loop studies that test whether computational gains translate into reproducible therapeutic success.
Keywords: Artificial Intelligence; Drug Discovery; Machine Learning; Generative Chemistry; Lead Optimization; ADMET Prediction; De Novo Drug Design; Virtual Screening; Graph Neural Networks; Drug Repurposing
Review Stats
Final search date and database lock: 2026-04-12 19:48:22 CEST
Plan: Pro (expanded craft tokens; source: PubMed)
Source: PubMed
Total Abstracts/Papers: 2424
Downloaded Abstracts/Papers: 2424
Included original and non-original Abstracts/Papers (all): 1005
Included original Abstracts/Papers (Vote counting by direction of effect): 516
Reference Index (links used in paper): 171
Total participants (topic deduplicated ΣN): 5027003
Get access to the full paper
Unlock the full evidence map
The full evidence review, including the Introduction, Methods, Results, Discussion, Conclusion, figures, and complete reference index, opens after purchase or sign-in.
[20] Peptide-based drug discovery through artificial intelligence: towards an autonomous design of therapeutic peptides. — https://doi.org/10.1093/bib/bbae275
[119] Hit Identification Driven by Combining Artificial Intelligence and Computational Chemistry Methods: A PI5P4K-β Case Study. — https://doi.org/10.1021/acs.jcim.3c00543
[229] Predicting pharmacodynamic effects through early drug discovery with artificial intelligence-physiologically based pharmacokinetic (AI-PBPK) modelling. — https://doi.org/10.3389/fphar.2024.1330855
[241] BioPrint meets the AI age: development of artificial intelligence-based ADMET models for the drug-discovery platform SAFIRE. — https://doi.org/10.4155/fmc-2024-0007
[265] Ongoing Implementation and Prospective Validation of Artificial Intelligence/Machine Learning Tools at an African Drug Discovery Center. — https://doi.org/10.1021/acsmedchemlett.4c00243
[287] Cheaper faster drug development validated by the repositioning of drugs against neglected tropical diseases. — https://doi.org/10.1098/rsif.2014.1289
[290] Medicinal Chemists versus Machines Challenge: What Will It Take to Adopt and Advance Artificial Intelligence for Drug Discovery? — https://doi.org/10.1021/acs.jcim.0c00435
[303] Artificial intelligence-guided Approach for Efficient Virtual Screening of Hits Against Schistosoma Mansoni. — https://doi.org/10.4155/fmc-2023-0152
[366] Prediction of pharmacokinetic/pharmacodynamic properties of aldosterone synthase inhibitors at drug discovery stage using an artificial intelligence-physiologically based pharmacokinetic model. — https://doi.org/10.3389/fphar.2025.1578117
[368] Artificial Intelligence-Assisted Optimization of Antipigmentation Tyrosinase Inhibitors: Molecular Generation Based on a Low Activity Lead Compound. — https://doi.org/10.1021/acs.jmedchem.4c00091
[369] A new strategy to HER2-specific antibody discovery through artificial intelligence-powered phage display screening based on the Trastuzumab framework. — https://doi.org/10.1016/j.bbadis.2025.167772
[381] Utilizing AI for the Identification and Validation of Novel Therapeutic Targets and Repurposed Drugs for Endometriosis. — https://doi.org/10.1002/advs.202406565
[401] Artificial Intelligence-Enabled Quantitative Assessment and Intervention for Heart Inflammation Model Organoids. — https://doi.org/10.1002/anie.202503252
[451] TCMBank: bridges between the largest herbal medicines, chemical ingredients, target proteins, and associated diseases with intelligence text mining. — https://doi.org/10.1039/d3sc02139d
[468] Prediction of Clinical Trials Outcomes Based on Target Choice and Clinical Trial Design with Multi-Modal Artificial Intelligence. — https://doi.org/10.1002/cpt.3008
[489] Explainable artificial intelligence in the design of selective carbonic anhydrase I-II inhibitors via molecular fingerprinting. — https://doi.org/10.1002/jcc.27335
[493] VGAE-MCTS: A New Molecular Generative Model Combining the Variational Graph Auto-Encoder and Monte Carlo Tree Search. — https://doi.org/10.1021/acs.jcim.3c01220
[497] AlphaFold accelerates artificial intelligence powered drug discovery: efficient discovery of a novel CDK20 small molecule inhibitor. — https://doi.org/10.1039/d2sc05709c
[498] Artificial intelligence-based drug repurposing with electronic health record clinical corroboration: A case for ketamine as a potential treatment for amphetamine-type stimulant use disorder. — https://doi.org/10.1111/add.16715
[533] Identification of SARS-CoV-2 main protease inhibitors from FDA-approved drugs by artificial intelligence-supported activity prediction system. — https://doi.org/10.1080/07391102.2021.2024260
[539] AI-enabled drug prediction and gene network analysis reveal therapeutic use of vorinostat for Rett Syndrome in preclinical models. — https://doi.org/10.1038/s43856-025-00975-8
[541] Artificial intelligence-driven drug repositioning uncovers efavirenz as a modulator of α-synuclein propagation: Implications in Parkinson's disease. — https://doi.org/10.1016/j.biopha.2024.116442
[550] AI-driven discovery of antiretroviral drug bictegravir and etravirine as inhibitors against monkeypox and related poxviruses. — https://doi.org/10.1038/s42003-025-09129-x
[595] Artificial Intelligence-Based Quantitative Structure-Property Relationship Model for Predicting Human Intestinal Absorption of Compounds with Serotonergic Activity. — https://doi.org/10.1021/acs.molpharmaceut.2c01117
[597] Scientific hypothesis generation by large language models: laboratory validation in breast cancer treatment. — https://doi.org/10.1098/rsif.2024.0674
[603] IDentif.AI-Omicron: Harnessing an AI-Derived and Disease-Agnostic Platform to Pinpoint Combinatorial Therapies for Clinically Actionable Anti-SARS-CoV-2 Intervention. — https://doi.org/10.1021/acsnano.2c06366
[620] Artificial intelligence assisted identification of potential tau aggregation inhibitors: ligand- and structure-based virtual screening, in silico ADME, and molecular dynamics study. — https://doi.org/10.1007/s11030-023-10645-3
[641] Automated discovery of noncovalent inhibitors of SARS-CoV-2 main protease by consensus Deep Docking of 40 billion small molecules. — https://doi.org/10.1039/d1sc05579h
[693] ADMET Evaluation in Drug Discovery. 19. Reliable Prediction of Human Cytochrome P450 Inhibition Using Artificial Intelligence Approaches. — https://doi.org/10.1021/acs.jcim.9b00801
[707] Development of a GCN-based model to predict in vitro phototoxicity from the chemical structure and HOMO-LUMO gap. — https://doi.org/10.2131/jts.48.243
[713] Literature-Based Discovery Predicts Antihistamines Are a Promising Repurposed Adjuvant Therapy for Parkinson's Disease. — https://doi.org/10.3390/ijms241512339
[757] Digital pathology with artificial intelligence analysis provides insight to the efficacy of anti-fibrotic compounds in human 3D MASH model. — https://doi.org/10.1038/s41598-024-55438-2
[785] Integrating text mining with network models for successful target identification: validation in MASH-induced liver fibrosis. — https://doi.org/10.3389/fphar.2024.1442752
[792] ShennongAlpha: an AI-driven sharing and collaboration platform for intelligent curation, acquisition, and translation of natural medicinal material knowledge. — https://doi.org/10.1038/s41421-025-00776-2
[800] Exploration of Novel Chemical Spaces to Discover JAK1 Inhibitors: An Ensemble Docking-Guided Deep Learning Approach. — https://doi.org/10.1021/acsomega.5c10773
[811] An Integrated AI-PBPK Platform for Predicting Drug In Vivo Fate and Tissue Distribution in Human and Inter-Species Extrapolation. — https://doi.org/10.1002/cpt.3732
[822] AI-driven discovery of dual antiaging and anti-AD therapeutics via PROTAC target deconvolution of a super-enhancer-regulated axis. — https://doi.org/10.1126/sciadv.adz9283
[829] Identification of CXCR4 inhibitory activity in natural compounds using cheminformatics-guided machine learning algorithms. — https://doi.org/10.1093/intbio/zyaf004
[853] Critical Assessment of Artificial Intelligence Methods for Prediction of hERG Channel Inhibition in the "Big Data" Era. — https://doi.org/10.1021/acs.jcim.0c00884
[860] A deep learning and docking simulation-based virtual screening strategy enables the rapid identification of HIF-1α pathway activators from a marine natural product database. — https://doi.org/10.1080/07391102.2023.2194997
[861] Diffusing on Two Levels and Optimizing for Multiple Properties: A Novel Approach to Generating Molecules With Desirable Properties. — https://doi.org/10.1109/tcbb.2024.3434461
[867] 4-Hydroxy-2,5-dihydrothiazole derivatives as a new class of small-molecule antibiotics for MRSA: AI-integrated design, chemical synthesis and biological evaluation. — https://doi.org/10.1016/j.ejmech.2025.118266
[871] A Fusion Deep Learning Model for Predicting Adverse Drug Reactions Based on Multiple Drug Characteristics. — https://doi.org/10.3390/life15030436
[879] Human-augmented large language model-driven selection of glutathione peroxidase 4 as a candidate blood transcriptional biomarker for circulating erythroid cells. — https://doi.org/10.1038/s41598-024-73916-5
[882] Systematic generation and analysis of counterfactuals for compound activity predictions using multi-task models. — https://doi.org/10.1039/d4md00128a
[895] De novo design and bioactivity prediction of SARS-CoV-2 main protease inhibitors using recurrent neural network-based transfer learning. — https://doi.org/10.1186/s13065-021-00737-2
[919] Drug repurposing for reducing the risk of cataract extraction in patients with diabetes mellitus: integration of artificial intelligence-based drug prediction and clinical corroboration. — https://doi.org/10.3389/fphar.2023.1181711
[939] GeneCompass: deciphering universal gene regulatory mechanisms with a knowledge-informed cross-species foundation model. — https://doi.org/10.1038/s41422-024-01034-y
[942] Identification of essential genes associated with SARS-CoV-2 infection as potential drug target candidates with machine learning algorithms. — https://doi.org/10.1038/s41598-023-42127-9
[965] Use of Deep-Learning Assisted Assessment of Cardiac Parameters in Zebrafish to Discover Cyanidin Chloride as a Novel Keap1 Inhibitor Against Doxorubicin-Induced Cardiotoxicity. — https://doi.org/10.1002/advs.202301136
[968] PepExplainer: An explainable deep learning model for selection-based macrocyclic peptide bioactivity prediction and optimization. — https://doi.org/10.1016/j.ejmech.2024.116628
[972] Development of Novel Methods for QSAR Modeling by Machine Learning Repeatedly: A Case Study on Drug Distribution to Each Tissue. — https://doi.org/10.1021/acs.jcim.4c00046
[985] Deep learning-assisted high-content screening identifies isoliquiritigenin as an inhibitor of DNA double-strand breaks for preventing doxorubicin-induced cardiotoxicity. — https://doi.org/10.1186/s13062-023-00412-7
