Automation in healthcare market research is now a known quantity. Think of the capabilities today to screen and reach physician samples with seamless ease, or algorithms that vet responses for bias and fraud, or data that pours into dashboards in real-time. This is all automation at work.  

But what about artificial intelligence or, more aptly, machine learning, a field in which computers use data to provide insights? There’s a lot of hype these days around these new technologies, especially in the world of consumer packaged goods, but its applications to healthcare market research are less clear.

Let’s first start with a definition of terms. There are few better than Jeff Bezos’ pithy explanation of the significance of this revolution:

“Over the past decades computers have broadly automated tasks that programmers could describe with clear rules and algorithms. Modern machine learning techniques now allow us to do the same for tasks where describing the precise rules is much harder.”

In fact, such modern machine learning techniques find their origins in relatively ancient mathematical models like logistic regression (dating back 175 years and still widely used to estimate the probability of an event given historical data). More recent developments like neural networks in the 1940s are modeled on the brain so computers can learn and recognize patterns. Recurrent neural networks, the latest and perhaps most powerful iteration, are a neural network arranged as a directed circle capable of unsupervised learning.

Arguably, the story of AI, machine learning, and its true impact in the life sciences industry is two-fold: automating tasks of higher and higher executive function magnitude, and discovering hidden groupings and classifications in large amounts of data.

The applications of AI and machine learning to market research are more of the former—mostly about augmenting human abilities and taking away the drudgery of day-to-day tasks.

AI finds patterns in numbers. Since many things in our world can be rendered as numbers, its application is potentially very broad. The most high profile uses of it are speech recognition, image categorization, and predictive analytics. For example, Siri, Alexa, and Google Translate, finding pictures of cats on the Internet, or estimating house prices.

Predictive analytics and AI uses in this arena could prove very useful to market researchers.

Jeff Bezos describes the applications of AI at Amazon, and it’s mostly behind-the-scenes stuff:

“But much of what we do with machine learning happens beneath the surface. Machine learning drives our algorithms for demand forecasting, product search ranking, product and deals recommendations, merchandising placements, fraud detection, translations, and much more. Though less visible, much of the impact of machine learning will be of this type — quietly but meaningfully improving core operations.”

In healthcare market research, we are already applying machine learning techniques to detect survey responses that don’t meet our high standards of quality, either because the respondent paid insufficient attention, or in the rare cases they are not who they say they are.

Transcription of qualitative research interviews is currently done by humans, but we’re not far from a point where natural language processing products transcripts as good quality as humans. Similarly, machine translation will soon be as good as human translation for the written word.

The practice of market research is, fundamentally, about understanding the attitudes of our fellow humans towards our products. This requires a deeper understanding of context and meaning than is currently available through machines, at least for the foreseeable future.

Still, the vast amounts of data being generated by human activity on the Internet and the motivation to harness them effectively is driving the innovation and application of machine learning techniques.