Original article was published on Artificial Intelligence on Medium
Drug Discovery and Development:
In the Age of Big Data, Machine Learning, and Artificial Intelligence.
The discovery and development of new drugs are celebrated by the public, academia, and even our government, as they should. How could we not be overjoyed when we make successful strides in providing tangible treatment regimens, or better yet cures, for those afflicted with cancer, HIV, neurological disorders, et cetera. Not only are these strides critical for our current populations’ wellbeing, it is crucial for our future population.
An Underlying Issue
However, under the hood the pharmaceutical industry as a whole has hit a breaking point. In layman’s terms all the low-hanging and most profitable fruits have been taken and it’s becoming extremely difficult to keep pace with improvement. In business terms, the more industry improves the standard of care, the more difficult and costly it becomes to improve any further, so pharmaceutical companies end up putting substantial efforts to innovate, i.e. R&D cost, to just get diminishing incremental benefits and added value for patients, which results in diminishing overall return on investments. Overall, the juice ain’t worth the squeeze anymore.
Currently, the drug discovery process is flat out too expensive. It is estimated that bringing a new drug to market costs major pharmaceutical companies at least $4 billion, and up to $11 billion, and can take between 11–16 years. Moreover, 9 out 10 clinical drugs fail to make it to trials, and a lot more don’t even reach the US Food and Drug Administration (FDA) approval stage, driving the costs of drug discovery and development through the roof.
A Modern Solution
A way out in this situation is through a major transformation towards more efficient and less costly innovative models, or a completely redesigned R&D process. Thus, in order to make pharma sustainable in the challenging economic landscape of making R&D more productive and less costly the adoption of advanced automation and analytics processes powered by various machine learning and AI-driven algorithms can serve as a solution.
Machine learning can automate many of the processes, utilizing large amounts of data that has been collected over decades about chemistry and drug effectiveness. Research firm McKinsey estimates that big data and machine learning in pharmaceuticals could generate a value of up to $100 billion annually. As well as helping with costs and timeframes, the analyst firm believes that predictive modelling of biological processes and drugs could help identify new potential-candidate molecules with a high probability of being successfully developed. Furthermore, real-time monitoring and removing data silos would also mean that costly issues such as adverse events or unnecessary delays could be avoided.
Pharmaceutical companies can also use the insights from the patient information such as mutation profiles and patient metadata. This information helps the researchers to develop models and find statistical relationships between the attributes. This way, companies can design drugs that address the key mutations in the genetic sequences. Also, deep learning algorithms can find the probability of the development of disease in the human system.
The data science algorithms can also help to simulate how the drugs will act in the human body that takes away the long laboratory experimentations. With the advancements in data science facilitated drug discovery, it is now possible to improve the collection of historical data to assist in the drug development process. With a combination of genetics and drug-protein binding databases, it is possible to develop new innovations in this field. Furthermore, using data science, researchers can analyze and test the chemical compounds against a combination of different cells, genetic mutations etc. Usage of machine learning algorithms, researchers can develop models that compute the prediction from the given variables.
What the Future Holds
While several leading pharmaceutical companies are implementing machine learning into their Drug Discovery and Development process, it is still a novel ideal for the industry. However, a group of MIT researchers and eight pharmaceutical giants formed a consortium to aid the drug discovery and development process, focusing on machine learning. The Machine Learning for Pharmaceutical Discovery and Synthesis Consortium (MLPDS) includes the following companies: Amgen, BASF, Bayer, Lilly, Novartis, Pfizer, Sunovion, and WuXi.
Furthermore, in my research while writing this blog, I uncovered there are a multitude of Data Science applications in the realm of Drug Discovery and Development. As such, in conjunction with learning the technical skills required to succeed, much more research is required in determining what sub-field I would like to become a part of. Thus, I am thoroughly pleased that this industry has exponentially grown right before my eyes. The following list is a categorical breakdown of how Data Science is influencing and impacting Drug Discovery and Development. As I continue my journey of immersing myself in the world of Data Science I look to learn as much as possible from the Flatiron School, as well as in my free time, so that I can apply these methods to progress a field that can ultimately change the course of our entire civilization.
- Aggregate and synthesize information
- Understand mechanisms of disease
- Establish biomarkers
- Generate data and models
- Repurpose existing drugs
- Generate novel drug candidates
- Validate and optimize drug candidates
- Design drugs
- Design preclinical experiments
- Run preclinical experiments
- Design clinical trials
- Recruit for clinical trials
- Optimize clinical trials
- Publish data
- Analyze real world evidence