Digital Technology
May 15, 2017

Artificial Intelligence in medical science: Fact or fiction? Part 3 - How technology and data empower consumers

After discussing what I consider "the parents" of artificial intelligence (AI)— namely, technology (the father) and data (the mother) — in my last two blogs, here is the last section in the series, in which the focus will be on the child itself. Consumerization of technology that is leading to data explosion around us, if needs to be processed in full; needs a superpower brain that is nothing but “artificial intelligence."

Consider this: On average, a physician needs to keep himself up to date and relevant by analyzing 10 new drugs for a single therapy every week, not to mention hundreds of conferences each month and an estimated 140 billion papers over the course of an entire career. Tough job, right?

Studies show that AI systems built on extensive research content and experiential learning have had 80% concordant results with evidence-based medicine. In fact, for the remaining 20% of the cases, the machine actually identified additional abnormalities in the patient that the normal physician had missed. Results like this are proving that machine learning could be a game-changer in medicine because, unlike humans, computers don't get tired and have an infinite capacity for learning.

Let me share a live example that we have currently executed within regulatory space of life sciences pertaining to drug safety i.e. pharmacovigilance. The diverse nature and format of data (e.g., PDF files, scanned and handwritten reports, call center transcripts, etc.), made it difficult for the pharma company to monitor and track adverse-event reporting related to a drug. And, as experts from the life science space will readily attest, the severity of a single miss in drug monitoring cases entails huge penalties for the drug manufacturer. However, by leveraging a natural language processing engine sitting on top of an automated data integration platform, the system has acquired the intelligence to accurately classify an event as adverse to the tune of 85%. This touchless system now runs 24X7, constantly monitoring adverse-event reporting for the drug manufacturer at one-tenth the cost while sustaining far better coverage and accuracy levels.

So, what does this imply for the healthcare and life science industry? Like any other industry, the impact on human jobs in medical science will be hierarchical — radiologists, pathologist, and others who sit lower in the value chain will become redundant sooner rather than later; eventually, a machine that has gained expertise with deep learning will be able to perform the jobs of physicians at the same, if not at a higher, level of precision.

Of course, the precondition for any of this is the critical coupling of technology (father) and data (mother). Way back in the 20th century, AI had apparently "failed" as a technology, because it had not advanced to support high-level, complex computing and data was not abundantly available. Today, on the other hand, there is a very high probability — nearly a certainty — that AI will be able to do most of the heavy lifting in medical sciences in the future.

So, AI is hitting healthcare and life sciences for real!

I hope you enjoyed the three blog series. Like, comment, or share if you did!

About the author

Deepika Goel

Deepika Goel

Vice President, Head of Healthcare and Life Sciences at Analytics Practice

Follow Deepika Goel on LinkedIn