In a simple definition, Artificial Intelligence is a very smart robot that can “think” for itself. We’ve all heard of at least one robot movie where they are programmed to think and behave like humans.

Well, that is one form of artificial intelligence. It is made to try to mimic what humans would do, think, or say and it is all on its own.

Artificial Intelligence is fed by massive amounts of data to analyze. This helps businesses know many things about their clients, the demand and supply chain and so much more.

Imagine if we could have a robot programmed to reply simple and common medical inquires and quickly respond to their requests based on the intelligence previously generated by our call centers database. Robots can very rapidly learn and form these “thoughts” based on historical interactions. This is a very applicable example to real life where we can bring agility to physicians’ questions and improve their perception regarding the responsiveness of our call centers.

As a real-life example, you will find more details about the Automatically Gleaning Insights from Voice of Customer Using NLP and Machine Learning project.

The Need:The Janssen Call center gets thousands of questions coming in every month. Besides answering the questions, there is also a need to analyze the pool of questions coming in, to glean actionable insights from them – because these questions are the Voice of our Customer. Example: If a lot of clarifying questions are being asked about some part of a product label soon after a product launch, those questions need to be answered for the customer, but we also need to examine the root cause of the questions and if the product label needs to be updated to make it clearer.

This remediation begins with first identifying an emerging trend in the VOC, that is indicative of a problem or an unmet need. Historically, this was being done by subject matter experts reading each VOC and manually classifying it into a topic/bucket and identify potential trends. The process was manually intensive and took time and was typically done quarterly. These was a real need to automate this process and make it real time. Bottomline: Understanding emerging trends in VOC real-time is important for us, to becoming a proactive insights-driven organization.

The Solution:  The Data Sciences team developed a pipeline using natural language processing and machine learning methods to automatically cluster Voice of Customer questions into themes as well as identify new/emerging topics. Results are displayed in a dashboard, enabling the trending topics to be easily consumed as word clouds. This solution has been launched for Invega Sustenna®, Spravato™, Invokana®, Stelara®, Xarelto®, Erleada®, Darzalex®, Tremfya®, Symtuza™ and for all Payer-related VOC in US and the same is being built for LatAm!

All these topics, Natural Language Processing (NLP) and Machine Learning (ML) are part of Artificial Intelligence (AI) umbrella, along with many other different techniques that you will see more in details in the next newsletters! Stay tuned!

If you have questions, ideas with regards these topics, feel free to contact Ana Egawa ( or Francisco Pereira ( an get more information.