Abstract

This paper introduces a generative artificial intelligence platform to enhance early-stage drug discovery in rare diseases, with conventional methods that are limited to smaller data volumes, costly, and lengthy development cycles. The suggested system combines molecular generation, based on Variational Autoencoders (VAEs), Generative Adversarial Networks (GANs), and Transformer-based models, with the optimization of reinforcement learning and in silico screening in a single computational pipeline. This architecture models drug discovery as a constrained optimization problem, diverting unguided sampling to learned probabilistic modeling to make more likely to find viable therapeutic candidates.The experimental analysis of high-performance cloud infrastructure with a synthetic dataset of 10 million records shows that it works well, with molecular validity at 92.5, novelty at 88.3 and uniqueness at 85.7. Such findings affirm the production of chemically sound and diverse compounds and not just memorization of training data. Candidate quality was also further optimized by reinforcement learning which improved average scores of binding affinity between 0.55 and 0.73 with an acceptable safety profile, such as a Lipinski compliance rate of 84.6.The efficiency of operations was greatly improved as shown by a drop of 12,000 to 3,500 candidates screened and a drop of 18 to 7 weeks to time-to-hit. Additionally, the framework made the cost per successful hit lower to 85,000 as compared to 240,000. Altogether, the paper reveals the possibility of using generative AI to make early-stage drug discovery of rare diseases a more focused, extensive, and resource-saving process.

References

• T. K. Mahato, “Pharmaceutical Innovation and Integrative Research: Bridging Drug Discovery, Biotechnology, and Life Sciences for Next-Generation Therapies,” Biopress Journal of Computational Life Sciences (BJCLS), vol. 1, no. 09, pp. 14–17, 2025, Accessed: Mar. 15, 2026.
• P. Kumar, “Advances in developing novel therapeutics, strategies, approaches, and use of emerging techniques,” Protein Misfolding in Neurodegenerative Diseases, vol. 4, no. 1, pp. 291–318, 2025
• J. A. Katakowski and J. C. López, “Drug development for neglected ultra-rare diseases of no commercial interest: Challenges and opportunities,” Drug Discovery Today, vol. 30, no. 4, p. 104346, Apr. 2025
• M. Schlander, K. Hernandez-Villafuerte, C.-Y. Cheng, J. Mestre-Ferrandiz, and M. Baumann, “How Much Does It Cost to Research and Develop a New Drug? A Systematic Review and Assessment,” PharmacoEconomics, vol. 39, no. 11, Aug. 2021
• E. Tambuyzer et al., “Publisher Correction: Therapies for rare diseases: therapeutic modalities, progress and challenges ahead,” Nature Reviews Drug Discovery, vol. 19, no. 4, pp. 291–291, Jan. 2020
• A. Chaudhary and V. Kumar, “Rare diseases: a comprehensive literature review and future directions,” Journal of Rare Diseases, vol. 4, no. 1, Jul. 2025
• R. J. Mead, N. Shan, H. J. Reiser, F. Marshall, and P. J. Shaw, “Amyotrophic lateral sclerosis: a neurodegenerative disorder poised for successful therapeutic translation,” Nature Reviews Drug Discovery, vol. 22, no. 22, Dec. 2022, doi: https://doi.org/10.1038/s41573-022-00612-2.
• N. Singh, P. Vayer, S. Tanwar, J.-L. Poyet, K. Tsaioun, and B. O. Villoutreix, “Drug discovery and development: introduction to the general public and patient groups,” Frontiers in drug discovery, vol. 3, no. 9, May 2023
• H. Cai et al., “Artificial Intelligence-Assisted Optimisation of Antipigmentation Tyrosinase Inhibitors: Molecular Generation Based on a Low Activity Lead Compound,” Journal of Medicinal Chemistry, vol. 67, no. 9, pp. 7260–7275, Sep. 2024
• B. Ahmad, K. Ouahada, and H. Hamam, “Machine learning for drug-target interaction prediction: A comprehensive review of models, challenges, and computational strategies,” Computational and Structural Biotechnology Journal, vol. 31, no. 4, pp. 316–345, Jan. 2026
• Z. Xu et al., “A generative AI-discovered TNIK inhibitor for idiopathic pulmonary fibrosis: a randomised phase 2a trial,” Nature Medicine, vol. 5, no. 8, Jun. 2025
• A. Gangwal et al., “Generative artificial intelligence in drug discovery: basic framework, recent advances, challenges, and opportunities,” Frontiers in Pharmacology, vol. 15, no. 9, Feb. 2024
• T. Khater et al., “Generative artificial intelligence-based models optimisation towards molecule design enhancement,” Journal of Cheminformatics, vol. 17, no. 1, Aug. 2025
• P. Tiwari, R. Pal, M. J. Chaudhary, and R. Nath, “Artificial intelligence revolutionising drug development: Exploring opportunities and challenges,” Drug Development Research, vol. 84, no. 8, Sep. 2023
• S. K. Bhattamisra, P. Banerjee, P. Gupta, J. Mayuren, S. Patra, and M. Candasamy, “Artificial Intelligence in Pharmaceutical and Healthcare Research,” Big Data and Cognitive Computing, vol. 7, no. 1, p. 10, Jan. 2023
• Z. Chen, X. Liu, W. Hogan, E. Shenkman, and J. Bian, “Applications of artificial intelligence in drug development using real-world data,” Drug Discovery Today, vol. 26, no. 5, pp. 1256–1264, May 2021
• F. Rajaei et al., “AI-based Computational Methods in Early Drug Discovery and Post Market Drug Assessment: A Survey,” IEEE/ACM Transactions on Computational Biology and Bioinformatics, vol. 6, no. 3, pp. 1–20, Jan. 2024
• Y. Chen and W. Xue, “Machine Learning for Molecular Generation: A Comprehensive Review,” ACS Chemical Neuroscience, vol. 17, no. 4, pp. 666–680, Feb. 2026
• B. P. Munson et al., “De novo generation of multi-target compounds using deep generative chemistry,” Nature Communications, vol. 15, no. 1, p. 3636, May 2024
• K. Khamrayev, “AI-Driven Drug Discovery: Accelerating the Path to New Therapies,” 2025 3rd International Conference on IoT, Communication and Automation Technology (ICICAT), vol. 12, no. 3, pp. 1–8, Dec. 2025
• W. Gao, S. Luo, and C. W. Coley, “Generative AI for navigating synthesizable chemical space,” Proceedings of the National Academy of Sciences of the United States of America, vol. 122, no. 41, p. e2415665122, Summer 2025.
• C. Chakraborty, M. Bhattacharya, S.-S. Lee, Z.-H. Wen, and Y.-H. Lo, “The changing scenario of drug discovery using artificial intelligence (AI) to deep learning (DL): Recent advancement, success stories, collaborations, and challenges,” Molecular Therapy — Nucleic Acids, vol. 14, no. 1, pp. 102295–102295, Aug. 2024