Tags

, , , ,

By  Padmanabhan (Paddy) Eangoor

In the past, machines were considered as tools that reduce human effort by performing tasks that humans are capable of, except thinking and learning. Machines, these days, come with cognitive skills as well, making them highly intelligent, if not to the level of humans. Artificial intelligence (AI) is currently being utilized in a variety of areas including self-driving cars and interpreting complex data. The former have the ability to maneuver in heavy traffic, react instantaneously to unpredictable pedestrians or drivers, and face a completely new environment on-road every time, all of which is possible because of the complex algorithm built into it for learning, a term called  machine learning. Machine learning has also made machine-mediated interpretation of complex data a reality.

Data like genomic information from millions of humans is really big. This big data gets even bigger with the addition of multiple dimensions like mutations in the genome, correlation with physiological conditions, associated diseases, etc. It takes hundreds of people, thousands of hours, and billions of dollars to interpret such big data and make predictions about the health condition of an individual. With the help of AI, these complications have been reduced significantly, simplifying the way to precision medicine. Projects like Human Longevity are already using AI to help individuals prevent diseases and lead a longer life.

Another way that AI may benefit human health is by identifying promising new drugs in a fraction of the time and with a fraction of the expense it takes today. The reason that pharmaceutical companies put a hefty price tag on these drugs is mostly due to the research and development costs, operation costs that incur over several years, and the risk associated with bringing a new drug to the market.

Atomwise, a drug discovery company, searched the database of previously approved drugs in a day using AI and came up with a drug that can block Ebola infectivity. A discovery effort that can take several years without AI can be done in a day with AI. BenvolentBio, a company that uses its deep learning AI software to go through thousands of publications and patents, identify patterns in how drugs work, and come up with a potential cure for life-threatening diseases every day, has produced 24 drug candidates in the last three years. There are a bunch of other drug discovery AI startups already in the market, and this number is expected to grow rapidly.

We never imagined that Apple’s Siri could answer the question “Why are fire trucks red?” If cracking a simple code in AI can allow Siri to answer this question (funnily), cracking a highly convoluted code in AI can give us answers in health care for which we don’t have questions yet. AI may propel drug discovery into a new era, but what are the downsides? Are there risks when we remove the human from the drug discovery equation?

Padmanabhan (Paddy) Eangoor is a Ph.D. candidate in the Department of Pharmaceutical Sciences at Mercer University. He currently works on techniques like SPE, ELISA, and HPLC-MS/MS in an analytical toxicology lab.