Chatbots are increasingly common and a variety of websites use these automated chat systems to field user questions and provide direction and information. With advances in AI, chatbot interactions are becoming more sophisticated and have the potential to provide useful and meaningful engagement with a target audience. The technology has the advantages of being autonomous, rapid, and convenient, but there is also the chance for miscommunication and errors if things go wrong.
Healthcare companies have been slow to adopt this technology, but the possible benefits in this arena are significant. For instance, a chatbot could help to schedule patient appointments, answer patient questions on a disease or treatment, or help patients to locate their nearest doctor or hospital.
However, the consequences of providing a patient with the wrong information about a disease or treatment could be serious. This, coupled with the highly regulated nature of healthcare, may underlie the lack of chatbots in the healthcare industry.
Medgadget asked John Fitzpatrick, vice president of media and engagement at precisioneffect, a healthcare branding company based in the U.S. and London, some questions about the potential for chatbots in the healthcare industry.
Conn Hastings, Medgadget: Can you tell us a little about chatbots and how they work?
John Fitzpatrick, precisioneffect: A chatbot provides a new way for users to interact with a brand. Site visitors can ask questions and get information by interacting with a computer program designed to simulate conversation. The machine learning-based service processes natural language through pre-built scripts that evaluate sentiment and learn how to recognize what users want.
There is no shortage of information online today that would overwhelm the most experienced web surfer. In reality, a lot of web users are often looking for something specific. Therefore, giving them direct and quick access to the information through an online chatbot can be a really valuable asset.
Simply put—users come to a website searching for info and the chatbot acts as a concierge by serving up relevant content.
Medgadget: How have developments in AI enhanced user experiences with chatbots, and chatbot reliability/accuracy?
John Fitzpatrick: To help answer this question, I asked UI Architect, Raphael Rafatpanah, to chime in.
Chatbots have evolved from only being able to match exact keywords and phrases to being able to match a user’s intent. They can understand the context of a user’s request and even “remember” the context of a single conversation. These and other advancements in natural language processing (NLP), combined with the scalability and cost effectiveness of cloud computing, opened the doors for businesses to use chatbots as a way to provide their customers with real-time, personalized experiences.
With powerful, affordable cloud services like Amazon Lex and Microsoft LUIS, businesses are able to focus on making their customers’ experiences great instead of having to also become NLP experts. Therefore, businesses are competing on a level playing field where experience is the true differentiator.
Medgadget: What are the advantages and potential applications of chatbots in the healthcare industry?
John Fitzpatrick: On a very practical level, companies can leverage chatbots to help with cutting down costs and demand for live customer service. It’s an efficient way to create more direct conversations with customers without the long hold lines.
From a content perspective, chatbots allow us to understand the customer mindset and develop content based on what exactly our customers might be looking for. In a similar way that SEO transformed content creation, chatbots can help bridge the gap between how our customers speak and how we communicate as marketers.
Medgadget: Can you discuss some of the risks associated with chatbots in healthcare? How can these be reduced?
John Fitzpatrick: The ability for chatbots to process natural language and learn to recognize what users want might be exactly what healthcare needs. With any regulated industry, the risk involved in real time has often scared brands away, in particular with social media.
However, by identifying early on what content can be used and ensuring we create business rules around that content, we can get medical legal teams comfortable with delivering real-time, approved content. Even more rewarding is that over time, your AI chatbot can get smarter and start to evaluate patient or caregiver sentiment.
Medgadget: Do you envisage that this type of AI communication technology will become more commonplace in healthcare?
John Fitzpatrick: Absolutely. One of the most interesting learnings from our chatbot experience has been collecting, mining, and interrogating data to understand what people are looking for, how they might ask for it, and whether or not our content today can provide it.
For example, one of our clients was new to market, and after launching the chatbot we learned that interest in refills was extremely high, but we didn’t have a ton of that content available yet; a key learning that led us to expedite that content to ensure our customers had that info at their fingertips.
The data mined can be tremendously helpful to content creators as an additional way of unlocking the customer mindset, unearthing new topical territories, and, most importantly, mimicking our audiences’ lexicon for all media.
Medgadget: What would pharma companies need to think about when deciding on including a chatbot in their communications?
John Fitzpatrick:
Start by answering a few critical questions:
- Which of your customers are you looking to support? (e.g., HCP, Patient)
- What are some of the most common customer service requests/inquiries that you receive?
- What type of support content exists to answer the most common customer service questions?
- What are the current channels available for customer support? (e.g., call center, phone, email, social, chat)
- What website(s)/platform will the chatbot be accessible from?
- How will the chatbot be monitored?
- What type of business rules do we need to create to ensure we are building a system that complies with all FDA regulations?
Link: precisioneffect