Table of Contents
How Machine Learning is Transforming Drug Discovery
In August 2020, eight months or so after the pandemic struck, the future looked rather gloomy for big pharma. After claiming over 750,000 lives and counting, COVID-19 wasn’t going anywhere, and the global rush to produce a vaccine was proving fruitless. With the death toll rising and economies in meltdown, scientists turned their collective attention to finding potent antivirals that could mitigate the life-threatening effects of the virus. As the clock ticked, they deployed machine learning-based models to speed up research and development (R&D), relying on complicated algorithms to implement these viral therapies.
Could machine learning save the planet?
By inputting data based on specific biomolecules to machine learning models, scientists hoped to identify patterns within that data and advance the search for effective COVID-19 drugs. Then, in February 2021, that hope resulted in a scientific breakthrough. Researchers at the Massachusetts Institute of Technology identified medications they could repurpose to fight the virus. Artificial intelligence helped them achieve this incredible feat. That’s just one of many examples of how machine learning has transformed drug discovery in the post-COVID era.
A (Brief) History of Machine Learning and Drug Development
Scientists have used machine learning in R&D for years before the pandemic, feeding copious data into computer systems to predict outcomes that aid drug development, testing, and repurposing. As machine learning models become better at identifying patterns between data sets, researchers have applied this technology to all stages of drug discovery, from analyzing data in clinical trials to product characterization and bioanalytical testing.
There are still challenges when machine learning-based models forecast the drug developmental process, especially when scientists misinterpret complicated algorithms or can’t validate the data that comes from them. Still, machine learning has revolutionized how big pharma executes its developmental pipelines — and how new drugs move from the lab to the pharmacy shelf.
As of 2021, 71 percent of scientists use machine learning for structure-based drug discovery and exploring new chemical entities, while 57 percent use it for imaging and data analysis. Seventy-nine percent of researchers believe machine learning use will increase within drug screening.
How Does Big Pharma Use Machine Learning for Drug Discovery?
Machine learning is an umbrella term that comprises various techniques such as deductive inference, statistical inference, and supervised learning. These methods, in the context of drug development, can find new medications, validate drug efficacy, predict drug-protein interactions, improve safety, speed up time to market, and even increase the bioactivity of molecules.
One of the biggest proponents of this technology is AstraZeneca who, alongside the University of Oxford, developed one of the most successful COVID-19 vaccines. The pharmaceutical company used machine learning for years, long before the pandemic swept through communities and became the greatest global health crisis of the early 21st century.
Across its R&D processes, AstraZeneca has relied on machine learning-based models to identify targets for novel drugs, gain a better understanding of the diseases they want to treat, design better clinical trials, drive personalized medicine techniques, and speed up drug development.
“Data science and AI have the potential to transform the way we discover and develop new medicines, turning yesterday’s science fiction into today’s reality with the aim of enabling the translation of innovative science into life-changing medicines,” says Jim Weatherall, the company’s vice president of data science and AI.
Machine Learning and Drug Discovery in Other Industries
It’s not just big pharma benefiting from machine learning for drug discovery. Scientists use this technology for R&D across the medicine, biotechnology, and pharmacology sectors, hoping to discover new drugs and therapies that will change and save lives. Take the skincare industry, for example, where researchers use machine learning-based models to develop products that tackle specific beauty and healthcare needs. Artificial intelligence can now determine how a new face cream or lipstick will look (and react to) the skin without scientists testing products on humans.
“At present, armies of data scientists are working on AI systems that can understand the human face. Once mastered, the ability to test out new looks and products will become exceptionally easy and realistic,” says Sciforce in an article that explores how machine learning is changing the beauty industry.
The Challenges of Machine Learning for Drug Discovery
Data quality is one of the biggest challenges associated with machine learning. Models are only as good as the data that comprise them, and scientists can encounter various difficulties when interpreting poor-quality data sets. Algorithms get smarter with every passing year, and big pharma companies are investing more money into technologies that facilitate the most advanced data-driven modeling techniques. However, obstacles persist:
“The challenges of applying machine learning lie primarily with the lack of interpretability and repeatability of machine learning-generated results, which may limit their application,” says Nature.com. “In all areas, systematic and comprehensive high-dimensional data still need to be generated.”
Another challenge for scientists is that of data availability. Machine learning models often require data from patient databases, including electronic health records (EHRs), which raises ethical and privacy issues. Meanwhile, ever-stringent data governance frameworks like HIPAA, GDPR, and CCPA make it more difficult for scientists to use raw data and generate valuable model inputs. Researchers need to comply with data privacy principles when using modeling techniques for drug discovery and testing.
Final Word
Life-saving COVID-19 treatments were born from drug discovery methodologies that incorporate machine learning-based models. This technology helps scientists identify patterns in data sets and accelerate R&D for the rapid implementation of viral therapies and other treatments, including those in the beauty industry. While challenges surrounding data quality and availability exist, machine learning has transformed drug discovery and will continue to advance healthcare outcomes.
Discover more global health-tech trends at intech.media.