.png){kind=icon}
.png){kind=icon}
Return
Artificial intelligence and machine learning in drug discovery: From lead discovery to clinical validation (2020-2025)
.png){kind=icon}
PRE
AI
V
D
DOI：10.1016/j.lddd.2026.100341.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Introduction: Artificial intelligence (AI) and machine learning (ML) are fundamentally changing the small-molecule drug discovery process, primarily by moving the workflow from being empirically screened to data-driven, designed ones. This narrative review consolidates peer-reviewed studies and industry-validated reports published between 1 January 2020 and 31 July 2025 from PubMed, Scopus, and Web of Science. Methods: This narrative review consolidates peer-reviewed studies and industry-validated reports retrieved from PubMed, Scopus, and Web of Science, covering the period from 1 January 2020-31 July 2025. The included sources report on AI and ML technologies applied to small-molecule drug discovery and their translational advancements. Results: AI methods such as deep docking, active learning, and multi-task learning have substantially shortened the time for virtual screening and hit identification, although benchmarking results often give a higher impression of real-world performance than is actually the case. By July 2025, a verification of clinical trial registrations and company disclosures indicates that more than twenty-nine publicly reported AI-driven therapeutic programs have advanced to human studies. Translational potential of AI-guided design is demonstrated by early examples such as DSP-1181 and the TNIK inhibitor rentosertib; however, wider clinical validation remains scarce and depends on prospective evidence. Discussion: Although the early successes of AI-guided drug design demonstrate the feasibility of AI-enabled discovery, the gap between benchmark performance and real-world outcomes continues to be a limitation. Furthermore, despite progress, broader clinical validation is still limited. Continued evaluation of AI-designed molecules in prospective clinical settings will be essential to verify their true therapeutic value. Conclusion: In spite of the limitations, the growing willingness of big pharmaceutical companies to incorporate the technology in their operations and the appearance of standard regulatory guidance for the use of AI in drug discovery suggest that AI-augmented discovery is no longer an experimental application but rather a maturing component of R&D pipelines.
Keywords:
ADMET prediction
Drug discovery
Virtual screening
Clinical validation
Foundation models
Deep learning
Artificial Intelligence in Drug Discovery
Machine Learning in Drug Discovery
Journal
L
IF:
1.6
Papers:
50
Citations:
0
Organization
No organization information available
Cited Papers