AI in Drug Development: Applications and Challenges

Abstract

The drug discovery process is a long and complicated one, involving multiple steps and issues, starting from target identification and ending with clinical development. The amount of chemical, biological, and clinical data produced by modern pharmaceutical companies is tremendous, and there is a continuous need to develop new approaches that could allow for efficient data mining and identification of relevant information. In this regard, artificial intelligence (AI), and specifically machine learning (ML) and deep learning (DL) seem to be an attractive choice for researchers for tackling the challenging task of data analysis. In silico artificial intelligence has been applied at all stages of the drug discovery and development pipeline, ranging from molecular target identification, virtual screening, molecular docking, quantitative structure-activity relationship, ADMET prediction, de novo design, ligand and lead optimisation, computer-assisted synthesis design, drug repurposing, and clinical trials. These approaches can facilitate the exploration of large data sets and the prioritisation of compounds for experimental assessment, thereby reducing the time and cost associated with traditional drug discovery. and expensive traditional drug discovery. However, despite the reported benefits and opportunities, there are still some limitations associated with AI-driven drug discovery, including dependence on large data sets with diverse characteristics, insufficient data for model training, data bias, overfitting, lack of interpretability, absence of independent evaluation, and regulatory issues. Therefore, it is essential to view AI technologies as an instrumental but supplementary part of decision-making in pharmaceutical research and development rather than a fully independent alternative to traditional hit- and lead-discovery strategies The manuscript is a Review article that focuses on the main areas of application of artificial intelligence for drug discovery and development, including benefits, drawbacks, ethical issues, and opportunities.

Introduction

The text reviews the role of artificial intelligence (AI) in drug discovery and pharmaceutical development, emphasizing how AI can make drug research faster, more efficient, and more data-driven while also highlighting important limitations and risks.

1. Why AI is needed in drug discovery

Traditional drug discovery is expensive, time-consuming, and highly uncertain, involving target identification, hit discovery, lead optimization, preclinical testing, and clinical trials. The pharmaceutical industry now generates enormous amounts of chemical, genomic, proteomic, and clinical data. AI and machine learning can process these large and complex datasets, identify hidden patterns, and help researchers prioritize promising candidates.

AI is particularly valuable because:

• Drug development has high costs and long timelines.
• Many drug candidates fail because of poor efficacy, toxicity, or pharmacokinetic properties.
• Chemical libraries contain enormous numbers of potential compounds.
• Researchers need better methods to reduce unnecessary experimental work.

2. Major applications of AI

AI is being applied across almost the entire drug-development pipeline:

• Target identification: AI analyzes genomic, proteomic, and biological data to identify disease-related drug targets and relationships between genes, proteins, and diseases.
• Virtual screening: Machine learning can screen thousands or millions of compounds and prioritize those most likely to be useful.
• Molecular docking: AI can improve prediction and ranking of how molecules may interact with protein targets.
• QSAR and molecular-property prediction: AI models can predict biological activity and chemical properties from molecular structures.
• ADMET prediction: AI helps predict absorption, distribution, metabolism, excretion, and toxicity, allowing unsuitable compounds to be rejected earlier.
• De novo drug design: Generative AI and related approaches can create entirely new molecules designed around desired biological and physicochemical properties.
• Lead optimization: AI can predict how structural modifications may affect potency, selectivity, solubility, toxicity, and other characteristics.
• Chemical synthesis: AI can predict reactions and design retrosynthetic pathways for producing new drug molecules.
• Drug repurposing: AI can identify existing drugs that may be useful for different diseases by analyzing drug–target and disease relationships.
• Clinical trials: AI can assist with patient selection, biomarker identification, recruitment, trial-site selection, and clinical-data analysis.
• Formulation and manufacturing: AI can optimize formulations and manufacturing processes while reducing experimental requirements.

3. Main benefits

The major advantages of AI in pharmaceutical R&D are:

1. Faster analysis of huge datasets.
2. Better compound prioritization, reducing the number of molecules requiring laboratory testing.
3. Earlier prediction of toxicity and ADMET properties.
4. More efficient molecular design and lead optimization.
5. Improved synthesis planning.
6. Identification of new uses for existing drugs.
7. More efficient clinical-trial planning and decision-making.

Overall, AI can reduce trial-and-error and help researchers focus resources on the most promising candidates.

4. Major challenges and limitations

The article stresses that AI is not automatically reliable. Its effectiveness depends heavily on the underlying data and methodology.

Important challenges include:

• Poor-quality or biased data can produce inaccurate predictions.
• Data scarcity is a major problem because experimentally validated pharmaceutical datasets are often limited.
• Overfitting can make models perform well on known data but poorly on new molecules.
• Black-box models may provide predictions without clear biological explanations.
• Chemical-space limitations mean models may perform poorly on compounds unlike those in their training datasets.
• Regulatory issues involve validation, reproducibility, transparency, and data provenance.
• Lack of multidisciplinary expertise can hinder effective implementation because successful AI drug discovery requires knowledge of biology, chemistry, pharmacology, statistics, and data science.

5. Ethical and regulatory concerns

AI-driven pharmaceutical research also raises concerns about data privacy, algorithmic bias, accountability, transparency, and reproducibility. Researchers need to document how datasets are collected, processed, and used to train models. Regulatory frameworks must balance technological innovation with drug quality, efficacy, reliability, and patient safety.

Conclusion

Artificial intelligence is becoming an essential part of the drug discovery and development process. Its role now extends from target discovery and virtual screening to QSAR modelling, ADMET prediction, de novo design, lead optimisation, synthesis planning, drug repositioning, and clinical development. The most significant advantage of AI lies in its ability to analyse extensive and multidimensional data sets while helping to make decisions based on the information. By prioritizing hits and predicting valuable properties, AI-driven technologies can help reduce trial-and-error through better prioritisation of candidates in pharmaceutical research while accelerating progression through the drug development pipeline. At the same time, one should not forget that AI is only a tool. The issues of data quality, insufficient training sets, overfitting, interpretability, and external validation limit its capacity to achieve better results. The most plausible scenario suggests that AI will not replace medicinal chemists, pharmacologists, biologists, and clinicians in the drug-discovery process but rather complement them, benefiting from cross-disciplinary collaboration. Thus, the implementation of artificial intelligence in pharmaceutical research and development will ultimately depend on how wisely the experts in the field can utilise these opportunities.

References

[1] Paul D, Sanap G, Shenoy S, Kalyane D, Kalia K, Tekade RK. Artificial intelligence in drug discovery and development. Drug Discovery Today. 2021;26(1):80–93. doi:10.1016/j.drudis.2020.10.010. [2] Vamathevan J, Clark D, Czodrowski P, et al. Applications of machine learning in drug discovery and development. Nature Reviews Drug Discovery. 2019 ;18:463–477. doi:10.1038/s41573-019-0024-5. [3] Morgan P, Brown DG, Lennard S, et al. Impact of a five-dimensional framework on R&D productivity at AstraZeneca. Nature Reviews Drug Discovery. 2018;17:167–181. doi:10.1038/nrd.2017.244. [4] Kim H, Kim E, Lee I. Artificial Intelligence in Drug Discovery: A Comprehensive Review of Data-driven and Machine Learning Approaches. Biotechnology and Bioprocess Engineering. 2021;26:895–930. doi:10.1007/s12257-020-0049-y. [5] Bender A, Cortés-Ciriano I. Artificial intelligence in drug discovery: what is realistic, what are illusions? Part 1: Ways to make an impact, and why we are not there yet. Drug Discovery Today. 2021;26(2):511–524. doi:10.1016/j.drudis.2020.12.009. [6] Bender A, Cortés-Ciriano I. Artificial intelligence in drug discovery: what is realistic, what are illusions? Part 2: A discussion of chemical and biological data. Drug Discovery Today. 2021;26(4):1040–1052. doi:10.1016/j.drudis.2020.11.037. [7] Chen H, Engkvist O, Wang Y, Olivecrona M, Blaschke T. The rise of deep learning in drug discovery. Drug Discovery Today. 2018;23(6):1241–1250. doi:10.1016/j.drudis.2018.01.039. [8] LeCun Y, Bengio Y, Hinton G. Deep learning. Nature. 2015;521:436–444. [9] Rifaioglu AS, Atas H, Martin MJ, et al. Recent applications of deep learning and machine intelligence on in silico drug discovery: methods, tools and databases. Briefings in Bioinformatics. 2018;20(5):1878–1912. doi:10.1093/bib/bby061. [10] Wu Z, Ramsundar B, Feinberg EN, et al. MoleculeNet: a benchmark for molecular machine learning. Chemical Science. 2018;9:513–530. doi:10.1039/C7SC02664A. [11] Ramsundar B, Kearnes S, Riley P, Webster D, Konerding DE, Pande V. Is multitask deep learning practical for pharma? Journal of Chemical Information and Modeling. 2017;57:2068–2076. [12] Ma J, Sheridan RP, Liaw A, Dahl GE, Svetnik V. Deep neural nets as a method for quantitative structure–activity relationships. Journal of Chemical Information and Modeling. 2015;55:263–274. [13] Mayr A, Klambauer G, Unterthiner T, Hochreiter S. DeepTox: Toxicity prediction using deep learning. Frontiers in Environmental Science. 2016;3:80. [14] Lenselink EB, Ten Dijke N, Bongers B, et al. Beyond the hype: deep neural networks outperform established methods using ChEMBL bioactivity benchmark sets. Journal of Cheminformatics. 2017;9:45. [15] Jeon J, Nim S, Teyra J, et al. A systematic approach to identify novel cancer drug targets using machine learning, inhibitor design and high-throughput screening. Genome Medicine. 2014;6:57. [16] Ferrero E, Dunham I, Sanseau P. In silico prediction of novel therapeutic targets using gene-disease association data. Journal of Translational Medicine. 2017;15:182. [17] Mamoshina P, Volosnikova M, Ozerov IV, Lane E, Skibina E, Cortese F, Zhavoronkov A. Machine learning on human muscle transcriptomic data for biomarker discovery and tissue-specific drug target identification. Frontiers in Genetics. 2018;9:242. doi:10.3389/fgene.2018.00242. [18] Koscielny G, An P, Carvalho-Silva D, et al. Open Targets: a platform for therapeutic target identification and validation. Nucleic Acids Research. 2017;45–D994. [19] Nelson MR, Tipney H, Kunkel SD, et al. The support of human genetic evidence for approved drug indications. Nature Genetics. 2015;47:856–860. [20] Gentile F, Agrawal V, Hsing M, Ton AT, Ban F, Norinder U, Gleave ME, Cherkasov A. Deep Docking: A Deep Learning Platform for Augmentation of Structure-Based Drug Discovery. ACS Central Science. 2020;6(6):939–949. doi:10.1021/acscentsci.0c00229. [21] Pereira JC, Caffarena ER, dos Santos CN. Boosting docking-based virtual screening with deep learning. Journal of Chemical Information and Modeling. 2016;56:2495–2506. [22] Schneider G. Automating drug discovery. Nature Reviews Drug Discovery. 2018;17:97–113. doi:10.1038/nrd.2017.232. [23] Olivecrona M, Blaschke T, Engkvist O, Chen H. Molecular de-novo design through deep reinforcement learning. Journal of Cheminformatics. 2017;9:48. doi:10.1186/s13321-017-0235-x. [24] Popova M, Isayev O, Tropsha A. Deep reinforcement learning for de novo drug design. Science Advances. 2018;4(7). [25] Kadurin A, Nikolenko S, Khrabrov K, Aliper A, Zhavoronkov A. druGAN: An advanced generative adversarial autoencoder model for de novo generation of new molecules with desired molecular properties in silico. Molecular Pharmaceutics. 2017;14:3098–3104. [26] Segler MHS, Preuss M, Waller MP. Planning chemical syntheses with deep neural networks and symbolic AI. Nature. 2018;555:604–610. doi:10.1038/nature25978. [27] Mak KK, Pichika MR. Artificial intelligence in drug development: present status and future prospects. Drug Discovery Today. 2019;24(3):773–780. [28] Jiménez-Luna J, Grisoni F, Schneider G. Drug discovery with explainable artificial intelligence. Nature Machine Intelligence. 2020;2(10):573–584. doi:10.1038/s42256-020-00236-4. [29] Li L, He X, Borgwardt K. Multi-target drug repositioning by bipartite block-wise sparse multi-task learning. BMC Systems Biology. 2018;12(Suppl 4):55. doi:10.1186/s12918-018-0569-7. [30] Liu Z, Roberts RA, Lal-Nag M, Chen X, Huang R, Tong W. AI-based language models powering drug discovery and development. Drug Discovery Today. 2021;26(11):2593–2607. [31] Liu X, Rivera SC, Moher D, et al. Reporting guidelines for clinical trial reports for interventions involving artificial intelligence: the CONSORT-AI extension. Nature Medicine. 2020;26:1364–1374. [32] Cruz Rivera S, Liu X, Chan AW, et al. Guidelines for clinical trial protocols for interventions involving artificial intelligence: the SPIRIT-AI extension. Nature Medicine. 2020;26:1351–1363. [33] Teodoro D, Naderi N, Yazdani A, et al. A scoping review of artificial intelligence applications in clinical trial risk assessment. npj Digital Medicine. 2025;8:486. [34] Mehta CH, Narayan R, Nayak UY. Computational modeling for formulation design. Drug Discovery Today. 2019;24(3):781–788. doi:10.1016/j.drudis.2018.11.018. [35] Rantanen J, Khinast J. The future of pharmaceutical manufacturing sciences. Journal of Pharmaceutical Sciences. 2015;104:3612–3638. [36] U.S. Food and Drug Administration. Artificial Intelligence for Drug Development. FDA, 2026. [37] Hutson M. Artificial intelligence faces reproducibility crisis. Science. 2018;359:725–726. [38] Srivastava N, Hinton G, Krizhevsky A, Sutskever I, Salakhutdinov RR. Dropout: a simple way to prevent neural networks from overfitting. Journal of Machine Learning Research. 2014;15:1929–1958. [39] Costello JC, Heiser LM, Georgii E, et al. A community effort to assess and improve drug sensitivity prediction algorithms. Nature Biotechnology. 2014;32:1202–1212. [40] Gaulton A, Bellis LJ, Bento AP, et al. ChEMBL: a large-scale bioactivity database for drug discovery. Nucleic Acids Research. 2012;40–D1

Copyright

Copyright © 2026 Saikat Biswas, Somenath Bhattacharya, Soumallya Chakraborty. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.