Endless hours of laborious work, billions of dollars spent testing hundreds of drugs, and extremely high rates of error riddle the contemporary market for creating a new therapeutic compound. The barrier to entry for most scientists is too high, as they are not certain that their new molecular discovery is translatable to improvements in clinical outcomes. However, recent advancements in technology and computer science have developed tools to pre-screen molecular compounds in an artificial (in silico) environment. reducing the costs of generating ineffective compounds, and producing a significantly higher degree of small-molecular compounds to target novel pathways for translational medicine.
Generally, the structure for clinical advancements and therapeutic pipelining follows a 20+ year timeline from when a molecule receptor is first discovered to the actual prescription of a drug or a trial. Typically, a scientist discovers a novel biological property related to a specific condition, which goes through a series of different animal model tests prior to beginning experimentation in a human. In addition to these different models, compounds have to be generated in new and unique ways to fully target the health condition of a patient. In this process, new therapeutics are pitted against previously identified pathways and truly testing if a compound is needed, compared to the other products on the market. Finally, once they have made these compounds and tested them in a clinical trial, the issue becomes how generalizable the trial is, how the results from this trial can be used to alter the specific treatment courses for different patients, and which criteria are used to delineate between different treatment styles. However, advancements in artificial intelligence (AI) and machine learning (ML) have been used to identify a number of core challenges in biotechnology: In silico testing of molecular compounds, Generating more streamlined assessments and augmentations of existing data and finally, combining genetic and clinical data to generate precise medical modeling [1].
With respect to in silico testing of molecular compounds, different AI and ML tools have been deployed with hopes of creating a virtual environment to test compounds prior to clinical trials. Companies like Atomwise have been created to root out treatments based on a database of molecular structures. This company focuses on testing unknown combinations of safe and existing medicines to find out if the drug is likely to combat the disease. This modeling was even used during the screening for Ebola medications, as it only took the company a few days to find two compounds effective at reducing the infectivity of Ebola [2]. In addition to this, the company has been using this same approach to commercialize and market to basic science researchers around the country. Rather than painstakingly derive the molecules and animal models appropriate to test a new compound, they can readily send the biological and biochemical information to these AI/ML researchers who can identify the effectiveness of a novel medicine using their algorithms [3]. With their algorithm for in silico testing, they can screen billions of different compounds set against hundreds of different biological receptors to get extremely accurate models of compound receptor interaction. However, the only limitation on this model is that they are generally developing small molecule drugs, which are typically used for pharmacological purposes but evade the newer classes of molecules like biologics and monoclonal antibodies which might have more functionality down the line.
Outside of generating new molecules, there are also groups dedicated to altering drugs already on the market that can be used for patients with other conditions. Other research groups or companies are dedicated to repurposing generic medications for specific rare conditions. RebootRx, a company dedicated to innovating and testing existing drugs on the market for cancer patients. This approach yields itself to becoming a non-profit startup dedicated to improving affordable and available cancer treatments. This has been supported in the literature, as it is perceived as one of the most practical and pragmatic actions that could be taken to innovate cancer treatment in the short term [4]. This approach uses AI and ML to look at the already existing clinical trial data as well as post-market evaluations of certain drugs that are already on the market for different oncological conditions. The goal of this is to optimize treatment outcomes for those with complex cancers and to create more functional opportunities to make truly precise medicine treatment courses and models [5, 6].
Overall, AI is rapidly transforming the landscape of translational medicine. It is changing the pipelining to discovery, minimizing the cost of research, maximizing the testing, and increasing the efficacy of therapeutics on the market. Using the tools of AI, we may truly be able to have individually tailored medicine.
References:
1) Shah, P., Kendall, F., Khozin, S., Goosen, R., Hu, J., Laramie, J., Ringel, M., & Schork, N. (2019). Artificial intelligence and machine learning in clinical development: A translational perspective. Npj Digital Medicine, 2(1), 69. https://doi.org/10.1038/s41746-019-0148-3
2) Meskó, B., & Görög, M. (2020). A short guide for medical professionals in the era of artificial intelligence. Npj Digital Medicine, 3(1), 1–8. https://doi.org/10.1038/s41746-020-00333-z
3) Chan, H. C. S., Shan, H., Dahoun, T., Vogel, H., & Yuan, S. (2019). Advancing Drug Discovery via Artificial Intelligence. Trends in Pharmacological Sciences, 40(8), 592–604. https://doi.org/10.1016/j.tips.2019.06.004
4) A call for pragmatism in cancer research. (2018). Nature Reviews Clinical Oncology, 15(4), 193–193. https://doi.org/10.1038/nrclinonc.2018.41
5) Ryan, L., Hay, M., Huentelman, M. J., Duarte, A., Rundek, T., Levin, B., Soldan, A., Pettigrew, C., Mehl, M. R., & Barnes, C. A. (2019). Precision Aging: Applying Precision Medicine to the Field of Cognitive Aging. Frontiers in Aging Neuroscience, 11. https://doi.org/10.3389/fnagi.2019.00128
6) Uddin, M., Wang, Y., & Woodbury-Smith, M. (2019). Artificial intelligence for precision medicine in neurodevelopmental disorders. Npj Digital Medicine, 2(1), 1–10. https://doi.org/10.1038/s41746-019-0191-0