Artificial Intelligence (AI) is certainly a top trend in pharma market research at the moment, but we haven’t seen too many examples of it actually in action. We have therefore conducted research to enable us to offer real-life cases of AI in action and we’re sharing our findings in a three-part series including articles and podcasts. Our first article offers insight into Michael our chatbot and what he can bring to healthcare market research.
Michael the Chatbot
Our first case study looks at our AI powered chatbot in action which was developed by Adelphi’s in-house Data Scientist Dr. Tom Gardner. Our chatbot ‘Michael’ acts as a healthcare professional who performs a mediator role between the patient and the physician. Here’s how it works…
Typically, in research we would present the HCP with a patient profile or collect details of a patient from a patient record study or chart audit and then ask them what they would use and why all up front. In this alternative approach, Michael converses with a doctor to understand firstly, what is important to the physicians and secondly, what will lead him to prescribing Product X. So instead of just understanding whether the doctor will prescribe Product X or not, we set to understand what is behind that decision.
Michael in Action
Imagine that Michael is introducing Cindy, a Chronic Lymphocytic Leukaemia patient. He is able to provide a brief background regarding Cindy and then he prompts the doctor with a question to understand what is important to understand next. From this, we can see the doctor is curious about the response and symptoms.
The AI component identifies the information to feed back to the doctor and can even ask some pointed questions such as which symptoms are most important and why. Michael can then ask the doctor if they have enough to make a diagnosis or if they need more information.
To find out how Michael is programmed and how he works listen to Adelphi’s illuminate video podcast on use of chatbots in pharma market research.
Decision Drivers – Deriving the Path to Prescription
We have found that we gain more information in this manner than showing a patient profile. Firstly, we can understand what is important in driving the decision and secondly, we can understand what the path is to prescribing Product X. From there it’s evident if there is a shorter path and an obvious patient. Additionally, a decision tree such as this one could highlight results across patient profiles.
Ultimately, having the doctor immerse themselves in this discussion with Michael increases engagement levels in the research. For instance, the respondents think more holistically about all the factors they would take into consideration when making a prescribing decision and not the usual normative response they might deliver in a linear online survey.
Educating Michael
Michael is a quick learner, but there are some key considerations. We need to anticipate all of the questions which might be asked so that Michael can answer whatever the doctor may ask. Therefore, to ensure we had the right information to feed the chatbot we worked closely with KOLs.
Once Michael had been programmed, he needed to be trained, which required mock conversations to ensure that he circled back the right information. We anticipate that the time to train will get quicker as the chatbot learns to handle more situations and patient types.
Get Involved
If you’d like to get involved in the application of chatbots in your next project, contact us to find out more.
This article is the first part of a three-part series on the topic of artificial intelligence and how we can add value in pharma market research. Look out for our next article on prescriptive analytics which will be out soon!