Artificial intelligence is being used by companies to speed up the drug discovery process.
TechRepublic's Karen Roby talked with Dr. Krishnan Nandabalan of InveniAI about the ways artificial intelligence (AI) is helping companies discover new medications. The following is an edited transcript of their interview.
Krishnan Nandabalan: AI could be considered as a force multiplier. If you look at the amount of data that the world has been gathering for the past few years, it kind of doubles every two years now, and if you take the medical field, especially on a daily basis, more than 5,000 publications are being added. It's humanly impossible to actually keep track of all of these things, and seeing which of them are relevant to you in your specific area of investigational research, and what is not. And what was being done manually until about five years ago cannot be done manually anymore. There's just too much data and information that's flowing through the system. So you can think of AI as a way to comprehensively analyze all the data that's available to you, whether it's clinical data, whether it's scientific data, whether it's patient-recorded data, whether it's data from hospitals, managed care, and all of these have an impact on the drug discovery and development process.
SEE: Special report: Managing AI and ML in the enterprise (free PDF) (TechRepublic)
So to go into a little more detail, right? So you can think of new mechanisms or new chemical entities and new biological entities that can be designed to address certain diseases where no cures exist. Or further still, even for drugs that are approved, you can start following the performance in the real world. Whether patients are complying with the drugs that have been prescribed, or how are they actually responding? Are they responding in an efficacious and safe manner? Because clinical trials, no matter how well designed, do not actually reflect the real world. And now with things like wearables, I'm talking about instruments like the Apple Watch or the Fitbit, we can monitor heart rate, pulse, and soon there will be more instruments. I can even monitor blood sugar and things like that, and sleep patterns.
All of this will actually help us understand how the human body is responding in a disease state to certain drugs and not responding to other drugs in a safe manner. And this will allow us to actually design safer and better drugs. Even if the improvement is incremental, that incremental improvement can have a huge impact on healthcare, especially the healthcare costs. For example, if we can reduce visits to the hospital even by a third, that'll have a huge impact on costs. We are seeing a direct impact of artificial intelligence throughout the healthcare system.
Karen Roby: What does the timeline look like?
Krishnan Nandabalan: An excellent question. We started using algorithms to essentially compare one drug to another in a given disease situation in 2006. If you think of any disease area, doctors already follow certain paradigms in prescribing certain drugs based on how the drug performs. What is the standard of care? What is the expected response to a particular drug? And so on. You see this in control of blood cholesterol, you see this in asthma, and you see this in various first-line or second-line treatment in cancers and so on. We have been doing this since 2005, and some time around 2011 we realized that we were not able to keep up with the amount of data that was flowing through the system. So, we essentially went to AI and machine learning as a solution to deal with the amount of data, and really see if we can automate things that should be automated because human beings are not designed to do the same things over and over again, and which don't need our intuition or expertise or experience. And so AI really comes in handy there.
If you want to think of AI in a most benign manner, it's a way to actually do things with the help of machines. What human beings have already mastered, right? So any process that is well-defined and you know the outcome of it, then you should be able to automate it and you should have an artificial system run it. And if you can do this in a measured way where you can measure the accuracy and monitor the efficiency, that's the best way of doing it. This is done in manufacturing, this is done in quality control. Now we have just taken it to the next level and applied it to the discovery process as well. What AI can't do right now is replicate the intuition and the imaginative leaps that the human mind can. And that's why we still need scientists and artists and so on, we don't have computers doing the same thing.
At least in terms of drug discovery, AI allows us to streamline data. [AI] allows us to essentially be comprehensive and not spend too much time just processing the data, but actually looking at the cleaned up data to do the analysis and research so that we can come up with better solutions. We made this move in 2015, and since then we have been implementing AI across the board, not just in discovery of new molecules but in clinical trial design and looking at, in terms of analysis of the results as well. And I would say that you would see most of our applications in rare and orphan diseases and in cancers, as well as in central nervous system disorders.
- IT leader's guide to deep learning (TechRepublic)
- Telemedicine, AI, and deep learning are revolutionizing healthcare (TechRepublic download)
- Managing AI and ML in the enterprise 2019: Tech leaders expect more difficulty than previous IT projects (TechRepublic Premium)
- What is AI? Everything you need to know about Artificial Intelligence (ZDNet)
- 6 ways to delete yourself from the internet (CNET)
- Artificial Intelligence: More must-read coverage (TechRepublic on Flipboard)