AI-assisted Drug Development is the Future

AI-assisted Drug Development is the Future
AI-assisted Drug Development is the Future


The U.S. drug development process for novel therapeutics targeting difficult-to-treat diseases takes an average of 10 to 12 years to complete. For drugs that go into development this year, that timeline leaves most patients battling severe illnesses without a lifeline. This includes the 33 percent of cancer patients who are not expected to live past five years post-diagnosis. The odds aren’t much better for those suffering from other, lesser-known illnesses, like severe acute pancreatitis, which has a 10-year life expectancy of just 70 percent. But it doesn’t have to be this way. A drug development approach that makes use of hybrid AI can de-risk drug development while simultaneously removing other barriers to success. In other words, it has the power to significantly reduce the drug development timeline, and ultimately, save more lives.

Why does drug development in the United States take so long?

Drug development is a lengthy process and for good reason: it’s meant to ensure that new drugs going onto the market are both safe and effective. Clinical trials alone often take between six to seven years to complete, in addition to years of preclinical testing, independent reviews, efficacy studies, etc. The total process can be delayed if mistakes are made during testing or clinical trials, it’s discovered that the drug can cause significant adverse events, or it’s determined that the drug may not be as effective as proposed. In many cases, these problems can also completely derail the drug development process, making it impossible to bring a drug to market, even if it has the potential to help many people. That’s because, in addition to being time-consuming, drug development is incredibly expensive. Cost estimates for drug development range from $340 million to $2.8 billion per drug. Not every drug developer has the ability or funding to start the process over again, or to go back and correct major errors. What stings more is that the research is often lost, and the hundreds, if not thousands of useful bits of data about drug combinations and interactions, adverse events, and efficacy are effectively sitting in filing cabinets in the basement, where no one can access them.

How AI can help break barriers and streamline drug development to save lives

Globally, more than 2.5 quintillion bytes of data are created every day. Although only a fraction of this is medical research data, the amount of unused medical research has still been amassed in the billions, if not trillions of bytes. Even if these data were available to researchers around the world through open data platforms, it could take years for human scientists to sift through it. Perhaps even more importantly, the drug development techniques, like traditional statistical modeling, to which regulators are accustomed have limitations that would further slowdown analysis and use of these data.

Inserting AI into these techniques adds value to and can meaningfully impact the translation of drugs to clinical success. That’s because AI can be trained to compare many points of data in mere minutes. In fact, it’s been suggested that AI is billions of times faster than humans at analyzing and categorizing data. In medical research and drug development, this means that AI can help researchers quickly determine whether certain medicinal compounds will work together or not. Moreover, AI can also determine how drug combinations will impact individual patients or groups of patients before a drug candidate is ever used in a clinical trial. That’s important because it removes one of the biggest barriers to successful, cost-effective, and timely drug development: risk. If a drug’s efficacy and potential for adverse events can be tested based on AI’s broad understanding of human biology and chemistry before launching human clinical trials, there is a possibility that more potentially helpful drug candidates can be saved from unnecessary failure. In turn, reducing errors, mistakes, and failures would have a drastic impact on the drug development timeline, greatly reducing it from the current 10+ years.

The million-dollar question: Can we trust drugs that have been developed by AI?

There’s a major misconception that AI is capable of replacing every job on the planet, or that it will eliminate the presence of humans in the workplace. And that misconception leads people to believe that computers alone will develop and carry out research. But that’s not the case, at least not in drug development, where scientists and researchers will always be at the core of progress and success. AI does not replace good science, good ideas, or the discerning eye and knowledge that come from researchers. And it can’t develop drugs on its own. But it can bring speed and agility to the research process that humans can’t accomplish on their own and, as a complementary tool to drug discovery and development, help scientists to leverage great ideas, and create broader access to important scientific data. We will always be trusting therapeutics that were developed by expert researchers. We’ll just know that they’re doing it more quickly, efficiently, and with fewer risks.

Conclusion

In the time it took you to read this article, roughly four people in the US died of cancer. That’s two people every three minutes. Increasing the speed and efficiency of drug discovery and development in this country isn’t just about the future of pharmaceuticals. It’s about the future of those people who are waiting for novel therapeutics that otherwise may never come. AI has the ability to help scientists do things that otherwise seem impossible, or even are impossible at this moment. As it gets better, smarter, and faster it will be the complementary research tool that helps scientists answer questions that could reduce the drug development timeline and ultimately save millions of lives.

Photo: metamorworks, Getty Images



Source link

Leave a Reply

Your email address will not be published. Required fields are marked *