16 Examples of Successful Applications of Artificial Intelligence in Pharma Marketing Part 3

13. AI Search Engines and Chatbots in Pharma Market Research
Google made search results more relevant to the searchers’ intent when they incorporated AI into their search algorithm Rank Brain in 2015. Through natural language processing and semantic search, relationships between similar products can be identified even when the searcher may not be fully sure of the name of item they are seeking.
 
This has numerous applications we have been able apply within pharma. For example, market research knowledge. Many companies have masses of PowerPoint documents, the content of which is known by the team. But what happens when team members leave and their replacements want to find information? It is dependent on the existing team bringing the new members up to speed. That’s why knowledge can easily be lost with job transfers.
 
By creating a database of all your market research PowerPoint files and applying AI search to it, you effectively allow everyone on your team to have the entire database of knowledge at their fingertips. Plus, they can search it in multiple ways, even if they don’t know whether what they are seeking is in there. The AI natural language processing will be able to interpret what is required and identify the content they are seeking.
 
We can even layer chatbot technology on top of that, so the marketers don’t even need to type a search. They can simply ask the question like they would with Siri or Alexa.
 
14. AI Enhancing Customer Call Centres
There are so many ways AI can be integrated into pharma call centers.
 
A simple example would be replacing Interactive Voice Response (IVR) with AI to improve the customer experience. We are all familiar with the classic IVR in call centers. You get an automated voice asking the purpose of your call and offering options. These are simple rule-based algorithms, and thus limited to the set responses.
 
By adding AI, such as natural language processing and machine learning, instead of giving a set of choices that are recognized set key words, the system can understand the question and deliver the appropriate response or action.
 
In addition, compared with IVR which always provides the same output based on the same input, AI could potentially provide a different output for different people, depending on what the system has learned about the person and the probability of their needs.
 
We did a project a few years ago for a pharma company that wanted to use AI to capture all the data from their call center to predict topics and optimal answers for the call center teams around specific drugs. This is just the tip of the iceberg.  With all the call center data, a lot more information and insights could be gained. We could add sentiment analysis on live calls so the staff can detect early signs of annoyance before a human advisor can.
 
AI could also help with is duplication of tasks. This is a particular annoyance I find when speaking with banks. After going through security prompts and speaking with someone, if you need to be transferred, you get put through security all over again.  AI combined with RPA (robotic process automation) could eliminate this need for repetition of tasks in these systems by capturing, analyzing, cross-referencing, and sharing information across platforms and channels, all without being intrusive.
 

15. Faster Reimbursement
Using a combination of natural language processing and machine learning to assist in faster reimbursement creates a lot of value for pharma.
 
Once a drug is approved by FDA/EMA, it must be submitted to the different formularies in the US, or the different country payers if in Europe. The goals of these payers can be different depending on their specific responsibilities. For example, if a payer is responsible for total healthcare costs, they will examine both drug cost and impact of the drug (reduced hospitalizations, etc.) and showing a significant savings in hospitalization in favor of your drug would be beneficial. However, showing data on reduced hospitalizations to payer whose responsibility is solely around drug costs may not be the best use of that meeting.
 
To ensure financial success of a drug, one needs to understand and incorporate the real payer drivers (for each payer influencing the decisions) in your strategy for each stage of development and commercialization.
 
Using AI we can actually write the submission documents utilizing both the drivers for that formulary as well as the optimal language shown in previous submissions to work best for that formulary committee.
 


16. Key Opinion Leader (KOL) / Thought Leader (TL) Mapping
 
Traditional and AI approaches to identifying KOLs in a given therapy area use many of the same sources, including publications, conference abstracts, Sunshine Act data and patent applications.  The main difference is that with AI, the data is constantly updated, analyzed automatically and can identify things traditional approaches miss.
 
The AI approach uses public data to mao KOLs and TLs in a way similar to how the CIA maps terrorists and drug cartels.  Our clients who are doing this are able to use the data in different ways to address questions across the organization from Sales & Marketing to Clinical and Discovery. These are just some of their wins:
    •    Validated the strategic brand plan
    •    Pressure tested the clients view regarding who were the top influencers
    •    Identified blind spots in the MSL engagement strategy
    •    New early phase pre-patented opportunities were uncovered
    •    Rising star and fresh faces identified for recruitment
    •    Better publication planning
    •    Congress interaction planning and communication
    •    Rapid identification of optimum influencers for clinical trials and research collaborations
 
So, Will AI Take My Job?
 
I got asked this question many times at every conference I have spoken at this year. There certainly is a lot AI can do, and it will allow you to achieve a lot more than you previously could. However, there is a lot AI cannot do and subject matter experts are needed.
 
I think of it this way. if there is anything that is almost impossible for humans to do (like identify rare disease patients from photos online – see case study here: https://www.eularis.com/case_study/patient-identification-with-ai-in-rare-disease), that is something that AI would be able to help with. Or if there is something that takes a lot of time and effort that could be automated, that is something AI can help with. By utilizing AI where you can, you are freed up to do the higher-level strategic planning aspects of your role.