Abid Rahman and Mike DeLuca discuss developing the AI-powered chatbot.
The life sciences industry continues to embrace the use of AI within very specific circumstances. While the majority of the industry agrees that humans should be making diagnoses and medical decisions, many also believe that AI can be a very useful tool to make life easier for HCPs.
Eversana recently developed an AI-powered medical information chatbot, which allows customers to find answers to complex questions around the clock. Abid Rahman, Vice President, Innovation, and Mike DeLuca, Senior Vice President, Medical Affairs, spoke with Medical Device & Technology about developing this technology.
Tell us a bit about the need within the business to create this tool for your client?
Eversana: Medical Information (MI) is an important customer-facing function for pharmaceutical companies, which plays a critically important role in supporting the safe and effective use of companies’ products by providing timely, scientifically balanced and evidence-based information in response to unsolicited requests for medical and scientific information.
Traditional channels of communication via phone, email, and field sales rep-facilitated requests still make up the majority of MI request volume, but as technology has advanced the needs of healthcare professionals (HCPs) and patients have changed. Offering digital channels, such as medical chatbots, has become an important part of the customer engagement strategy to ensure communications are occurring via the customer’s preferred channel.
Embracing and leveraging advancing digital technology can improve the customer experience, allowing for self-service and access to medical information 24/7/365.
From a technical standpoint, what were some of the challenges your team faced in building and programming an AI-powered chatbot for the medical industry?
Eversana: There were a number of technical considerations and challenges our team looked at as we developed the approach to develop this tool. Here are some of the technical challenges and considerations we needed to handle.
First was the availability of training data. To train a conversational chatbot requires a lot of relevant data and it can be challenging to find the datasets that we have the correct data to use for AI training. The data also must be cleaned and tested for inaccuracies and bias. Our team spent a lot of time ensuring that as we collected data sets to build the product, it was both accurate and, most importantly, compliant to meet both customer and industry requirements.
Secondly and in the same vein, accuracy and reliability are paramount. Medical information requires a high level of accuracy and reliability. We needed to create a unique testing mechanism to achieve a high level of accuracy.
Next, a challenge we worked through is the complexity of medical language. Medical and pharmaceutical terminology can be intricate and context, dependent of the disease state and other factors. As we built the tool, we were focused on ensuring the chatbot could both understand and interpret complex medical jargon, as well as the context of questions from both healthcare providers (HCPs) or patients.
Another critical component we spent lots of time ensuring accuracy with was regulatory compliance and content approval. We needed to put in the appropriate perimeters to ensure legal and regulatory compliance standards were met. Requests for medical information needed to be unsolicited and responses must be non-misleading, truthful, accurate, non-promotional, scientifically balanced, and tailored to only address the specific question being asked.
And the final critical consideration was how the platform reviewed and supported adverse events and product complaints. This is a common function in medical information, and we needed a fast and efficient way that when these were received, they were rapidly distributed to the appropriate team members who could log and report them as needed per the client’s requirements.
It took a large collaborative effort to ensure all these boxes were ‘checked,’ but the final product turned out well.
Are there other use cases for this type of chatbot in other parts of the life sciences industry?
Eversana: Absolutely. Chatbots continue to be a growing tool in many sectors, including life sciences. And use cases can vary depending on the customer’s end goal. From providing self-service agents in many situations where around-the-clock service is needed in areas like patient support programs, education, call-center training, getting questions from key opinion leaders or experts and more, the opportunities are endless.
Regardless of what the application is for a chatbot, the critical component is to ensure the rigor to develop the tool meets regulatory standards while creating a seamless experience for the target audience that will use the tool.
How has your client team measured success? What stats are you able to share on the impact of the tool so far? Has it led to more requests from other companies?
Eversana: We are in the process of configuring, implementing, and going live with the medical chatbot for several clients.
For some clients, the success is just having digital channels in place for medical information and allowing for multi-channel engagement for their customers. Other clients want the medical chatbot in place to be able to provide 24/7/365 medical information support for one MI Contact Cetner is not available. Other clients are looking to build awareness of their medical information services and expand access for their customers by offering digital channels and allowing the customer to engage with them via their preferred channel. Additionally, some clients want to measure if the chatbot will reduce call volumes.
As a service provider of global medical information services, it is important for us to be able to provide a holistic and customer-centric approach to medical information. We believe that medical information is a critically important customer-facing function, and that medical information should be a center of excellence for customer engagement. Providing multi-channel support and digital solutions is essential to make this happen.
Are there any limitations to what chatbots can do?
Eversana: Like any technology, chatbots do come with some limitations. First, while they can provide a lot of information to either patients or HCPs, they cannot and should not be used for disease diagnosis at this time. They should never replace clinical expertise. Instead, we think of them as a compliment to potentially make the process of disease diagnosis more efficient.
Additionally, in the case of emergency situations, chatbots cannot be substituted for human judgment or care. People still need to get advice and treatment paths from trained medical experts.
Finally, while they can do a lot based on how they are programmed and developed, they can only solve for the information they are programmed to do. Extremely complex challenges will require a combination of human and technological collaboration, and this should always be the case.
From a regulatory standpoint, what if any compliance issues did you face while developing the chatbot? What strategies did you deploy to overcome them?
Eversana: The goal is for the medical chatbot to be easy to use for the customer and to be conversational in nature vs. very guided and limited.
However, requests for medical information need to be unsolicited and responses must be non-misleading, truthful, accurate, non-promotional, scientifically balanced, and tailored to only address the specific question being asked.
Also, since any questions can be asked within the chatbot it is possible that adverse events or product complaints may be reported.
Strategies deployed to address these potential challenges included ensuring appropriate disclaimers and attended use of the medical chatbot; working closely with each client on content and materials that would be available and ensure the most appropriate answer is returned for each question; leveraging AI to identify possible adverse events and product complaints; and also putting in processes for Mi Specialists review of transcripts of chatbot interactions.
How has generative AI played a role in the development of these types of tools?
Eversana: Generative AI has played a massive role in the development of tools like chatbots, including drastically reducing the time to deploy conversational chatbots. The technology is currently used to create user personas, augment training data, generate test data, create variations of responses with the right tone based on the user, and much more.
It is a powerful tool to have in a developer’s toolbox as they work to build out future chatbots. However, it must be developed and then reviewed with a careful lens from a true medical expert.
Any closing thoughts on the project?
Eversana: Medical Affairs should think beyond just having a medical plan and strategy because having a digital
strategy for customer engagement is becoming equally important. Now is the time to take a more customer-centric approach and embrace and leverage advancing digital technology to improve the customer experience.
Medical Portals and Medical Chatbots leveraging conversational AI are just a component of that overall strategy. At EVERSANA we are looking at how generative AI can also be leveraged in other areas throughout medical affairs to improve operational efficiency and improve customer experience.
Any additional thoughts on your end?
Eversana: Chatbots are here to stay, and their role is only going to increase over time for a few key reasons:
First, convenience Chatbots provide around-the-clock access to information and solve real-world challenges such as making an appointment with an HCP or sales rep, reporting adverse events, and much more.
Second, scale. Chatbots can quickly scale to handle large volumes of conversations simultaneously. What before may require multiple people reviewing content for accuracy now can be automated using machine learning and other tools, creating faster ways to bring mountains of data together that can be then output through a chatbot.
Third, chatbots allow tremendous personalization. As technology has rapidly evolved over the past few years, AI-based chatbots can provide personalized responses based on the user’s role, location, and a variety of other factors. Such a powerful tool to create a truly customized experience.
Fourth, by their ability to take common and routine tasks and automate them, chatbots have the ability to lower costs to organizations and drive efficiency.
The acceptance and reliance on digital solutions like chatbots have already seen enormous growth in the industry today, and the tools are helping people get answers immediately to unmet needs. They have the ability and power to generate valuable data that organizations can use to understand user preferences, pain points, and trends, leading to better customer service. In the end, they are a powerful tool that companies should consider as they work to connect with their stakeholders more effectively in a compliant manner, especially in the life sciences industry.
In this Q&A with Pharmaceutical Executive®, Andrew Hopkins, founder and CEO of Exscientia, reveals how artificial intelligence (AI) is currently being utilized in the pharma industry, predictions for where AI can be implemented in the future, and what pharma fears most about AI.