Varun Ranganathan, MCOptom

Clinical Optometrist

An OCULAR Interface Exclusive

Synopsis:

By leveraging advanced algorithms and data analytics, AI accelerates drug discovery, optimises clinical trials, and enables personalised medicine tailored to individual patients. This blog highlights AI’s role in identifying biomarkers, monitoring diseases, and enhancing treatment adherence through real-world evidence. While the integration of AI offers immense promise, challenges like data privacy and regulatory compliance must be addressed. The blog envisions a future where AI-driven innovation transforms vision care, making preventable blindness a relic of the past.

Introduction

Drug discovery, the process of identifying and developing new medications, is a complex and time-consuming endeavour that traditionally relies on labour-intensive techniques, such as trial-and-error experimentation and screening. However, AI techniques such as machine learning (ML) algorithms and data analytics offer the potential to accelerate and improve this process by enabling more efficient and accurate analysis of substantial amounts of data.

The role of ML in drug development

The use of ML and Deep Learning (DL) algorithms is key in prediction of efficacy and toxicity of drug compounds. Based on analysis of vast amounts of data, these algorithms can identify trends in the data which can take human researchers months or years to find out. ¹ Discovering new bioactive compounds with minimum side-effects will be much faster and potentially cost saving. The algorithms can also be trained on large databases of known toxic compounds thereby preventing drug toxicity. ²

Similarly, the algorithms can also run simulations of how several drugs interact with each other and can alert potential side-effects on the drugs in people with co-morbidities. ³ This can enable us to develop customised treatment plans based on a person’s genetic profile and response to medications, which can reduce adverse reactions.

In ocular drug delivery, the main challenge is traversing the anterior and posterior barriers such as blood-retinal barrier, blood-aqueous barrier, Bruch’s membrane, retina, choroid, sclera, tear film, and conjunctiva. ⁴ Current drug administration methods are mostly intravitreal or into the anterior chamber for maximum efficiency. Invasive retrobulbar, subconjunctival, and intravitreal injections can be painful ⁵ and novel administration techniques are needed which need to be both minimally invasive with maximum efficiency. These highly efficient formulations have increased drug permeability, longer drug residency, increased drug solubility, or improved drug stability which are minimally invasive or non-invasive formulations and targeted drug delivery systems. ⁶

Challenges and limitations in AI-based drug discovery

One of the main challenges in using AI for drug discovery is ethical considerations and it is also based on the data acquired. ⁷ If the data is limited or biased, the results could be unfair or incomplete. Limited data can be a major barrier as the AI needs training on vast datasets for a better outcome. Data augmentation is a process by which new data samples or synthetic data can be created using pre-existing data. ⁸ This can improve the quantity of data and diversity of the available data. As with any clinical trials, the reason behind a decision will need to be explained and by using Explainable AI (XAI), bias and fairness in AI-based approaches and the decisions regarding choice of drugs and clinical trials can be addressed. ⁹

Data privacy and security is an ongoing challenge even in the human world. Since AI is trained and manages enormous amounts of data, the risk of data leaking is a concern. ¹⁰ Sensitive information regarding the drug formulations can be altered or misplaced and can lead to drastic consequences in patients. Auditing the algorithms regularly for bias with strict security protocols and data protection methods are highly crucial in protecting public’s interests.

Conclusion

Current practices in drug research and development do not extensively use AI methods mainly due to the concerns on its approach towards bias and equality. Successful clinical trials employing AI has a long way ahead, but it offers promising outcomes for novel drugs to treat eye conditions with greater efficacy.

References

1. Xu Y., Liu X., Cao X., Huang C., Liu E., Qian S., Liu X., Wu Y., Dong F., Qiu C.W., et al. Artificial intelligence: A powerful paradigm for scientific research. Innovation. 2021;2:100179. doi: 10.1016/j.xinn.2021.100179.
2. Santín E.P., Solana R.R., García M.G., Suárez M.D.M.G., Díaz G.D.B., Cabal M.D.C., Rojas J.M.M., Sánchez J.I.L. Toxicity prediction based on artificial intelligence: A multidisciplinary overview. Wiley Interdiscip. Rev. Comput. Mol. Sci. 2021;11:e1516. doi: 10.1002/wcms.1516.
3. Jang H.Y., Song J., Kim J.H., Lee H., Kim I.W., Moon B., Oh J.M. Machine learning-based quantitative prediction of drug exposure in drug-drug interactions using drug label information. npj Digit. Med. 2022;5:100. doi: 10.1038/s41746-022-00639-0.
4. Peng Chen, Hao Chen, Xinjie Zang, Min Chen, Haoran Jiang, Shasha Han, Xianggen Wu, Expression of Efflux Transporters in Human Ocular Tissues, Drug Metabolism and Disposition, Volume 41, Issue 11, 2013, Pages 1934-1948, ISSN 0090-9556, https://doi.org/10.1124/dmd.113.052704 (https://www.sciencedirect.com/science/article/pii/S0090955624115358)
5. Di Huang, Ying-Shan Chen, Ilva D. Rupenthal, Overcoming ocular drug delivery barriers through the use of physical forces, Advanced Drug Delivery Reviews, Volume 126, 2018, Pages 96-112, ISSN 0169-409X, https://doi.org/10.1016/j.addr.2017.09.008. (https://www.sciencedirect.com/science/article/pii/S0169409X17301898).
6. Cheng Peng, Liangju Kuang, Jiangyue Zhao, Amy E. Ross, Zhongqing Wang, Joseph B. Ciolino, Bibliometric and visualized analysis of ocular drug delivery from 2001 to 2020, Journal of Controlled Release, Volume 345, 2022, Pages 625-645, ISSN 0168-3659, https://doi.org/10.1016/j.jconrel.2022.03.031.(https://www.sciencedirect.com/science/article/pii/S0168365922001572).
7. Vamathevan J., Clark D., Czodrowski P., Dunham I., Ferran E., Lee G., Li B., Madabhushi A., Shah P., Spitzer M., et al. Applications of machine learning in drug discovery and development. Nat. Rev. Drug Discov. 2019;18:463–477. doi: 10.1038/s41573-019-0024-5.
8. Gracia Moisés, A., Vitoria Pascual, I., Imas González, J. J., & Ruiz Zamarreño, C. (2023). Data Augmentation Techniques for Machine Learning Applied to Optical Spectroscopy Datasets in Agrifood Applications: A Comprehensive Review. Sensors, 23(20), 8562. https://doi.org/10.3390/s23208562.
9. Sajid Ali, Tamer Abuhmed, Shaker El-Sappagh, Khan Muhammad, Jose M. Alonso-Moral, Roberto Confalonieri, Riccardo Guidotti, Javier Del Ser, Natalia Díaz-Rodríguez, Francisco Herrera, Explainable Artificial Intelligence (XAI): What we know and what is left to attain Trustworthy Artificial Intelligence, Information Fusion, Volume 99, 2023, 101805, ISSN 1566-2535, https://doi.org/10.1016/j.inffus.2023.101805.(https://www.sciencedirect.com/science/article/pii/S1566253523001148)
10. Kelly D. Martin, Johanna Zimmermann, Artificial intelligence and its implications for data privacy, Current Opinion in Psychology, Volume 58, 2024, 101829, ISSN 2352-250X, https://doi.org/10.1016/j.copsyc.2024.101829.(https://www.sciencedirect.com/science/article/pii/S2352250X24000423)