Artificial Intelligence (AI) is radically transforming the pharmaceutical industry. The revolution is still in its early stages, but already we are seeing the widespread use of automated algorithms to carry out tasks which traditionally have relied on human intelligence alone. The importance of AI in pharma has grown exponentially over the last couple of months, particularly driven by the need to develop a vaccine to stop the spread of the coronavirus and treat its symptoms.
However, the developments in the pharma sector were already part of a larger shift within the health care market, which is adopting AI tools at a growing rate. Research conducting prior to the COVID-19 outbreak by Tractica found that the global market for AI software in the health care industry will exceed $8.1 billion by 2025, up from just $463 million in 2018. No doubt, the current pandemic will only accelerate the sector’s growth.
The pharmaceutical industry is now under stress for new and better ways of managing the drug discovery process, but answers are at hand. Several large pharmaceutical companies have already started using AI and machine learning in order to make this process faster and more cost-effective.
Speeding up the reporting process
In the fight against COVID-19, time is of the essence. And one of the most time-consuming elements of the drug development process is the generation of Clinical Study Reports (CSRs). The job of testing new products and getting them to market with maximum efficiency, while satisfying regulatory demands, is a perennial challenge for pharmaceutical companies.
CSRs are important and necessary, but their labor-intensive nature requires the time of highly skilled medical professionals. A CSR can take weeks, and in some cases months, to complete. This results in a delay before potentially life-saving drugs can be made available in the field.
AI is a way to reduce the time taken up by CSRs in order to increase efficiency and bring drugs to market in a shorter time frame. This is now available in the form of software tools that can automate the writing of the more repetitive and mundane sections of a CSR, augmenting the efforts of experienced professionals and allowing them to focus on the more complex and demanding aspects of the job.
In order to automate certain elements of the CSR process, pharmaceutical companies are now starting to use AI and Natural Language Generation (NLG) as a way to speedily generate some of the text they need.
The process usually starts with a pharmaceutical company feeding the software a dataset in the form of a statistical table. Generating 20 pages of a report might take up to 30 tables. That’s a lot of data, and a lot of analysis. But with NLG, this can be done in a few seconds, saving a third of the time it takes to produce these pages. However, NLG is not just an automation tool. It goes beyond that, providing the medical writer with a virtual companion which will work in a collaborative way. For example, the medical writer can ask the machine to add more details, less details, reword a section and so on. This goes far beyond simple automation tools which often need a lot of rewriting from the medical writers themselves.
Another CSR issue is that having multiple medical writers working on a single report can result in multiple writing styles for a regulator to navigate. NLG facilitates a consistent writing style across the document.
The CSR process can sometimes require information that is only known by a medical writer. It might be that only that person can explain the true value of what is in the statistical tables. In this case they can program how the document is to be read, and how the facts fit together. This will in turn generate a set of rules to help understand new tables given to the machine. This means that fresh data can be put straight into the machine to generate the text.
With this level of intelligence at its disposal, a pharmaceutical company can present its findings to the regulatory authorities faster and get a drug to market sooner. Even a day saved here could represent a huge impact in the fight against COVID-19.
The future of drug discovery
Bill Gates, whose foundation is focusing its efforts to fight the virus, doesn’t think life will return to normal until there’s a viable vaccine that can stop its spread. Gates theorized in a blog post published In April that “it could be as little as 9 months or as long as two years” to develop a vaccine. But he importantly noted that even if it takes 18 months, that would still be the fastest that scientists have created a new vaccine.
Normally, it can take around nine years to get a product from the start of the discovery stage to the pharmacy, costing an average of $2.6 billion on the way. Much of the money spent before a drug can be launched and monetized is effectively wasted, since many possible options will be trailed and abandoned before the final hurdle of regulatory approval can be reached.
Modern AI techniques are well-suited to drug discovery because of their ability to identify patterns hidden in large, complex and varied volumes of data. Using these, scientists can analyze and weigh up thousands of possible outcomes from the drug testing process.
Machine learning algorithms can help link potential treatments to the precise biological causes of disease and manage this in a far more effective manner than the trial-and-error approach that has typified the traditional drug discovery process. A lake of information can be reduced to a small pool of genuine insight in a massively shortened timeline.
Spearheaded by the COVID-19 outbreak, we are entering an era when algorithms can analyze clinical data and estimate the journeys of simulated patients throughout the trial, accurately predicting the outcomes. Further investment and development can turn these AI-led methodologies into increasingly robust and trustworthy simulation tools for data-only clinical trials. By disrupting traditional testing methods, pharmaceutical firms could radically reduce the average trial cycle from years to months.
The pandemic underlines the need for pharmaceutical companies to use algorithms to reinvent the drug discovery process. Bringing drugs to the market faster is no longer just a commercial incentive, but a moral one as it has the potential to save thousands of lives.
AI can play a huge role in streamlining regulatory challenges, data management and ultimately in the discovery of remedies for rare diseases. AI can be a force for good, helping to improve healthcare and saving lives. The future of pharma is here.
