HealthBOT2018 Learnings: Conversation Is King

Posted by Kenzie Kline from Ogilvy CommonHealth Worldwide ― North America on September 20, 2018


Last week I attended the HealthBOT2018 conference curated by both Ogilvy and Conversation Health. Speakers from pharma companies, chat bot services, and UX experts shared some of the opportunities and challenges that come with this exciting, developing technology. I must admit, I went into the conference wondering how bots could be used effectively in an industry with so many constraints and limitations. I was familiar with success stories like the robot ticket lawyer1 in the consumer world, but my prior experience on IBM Watson Health led me to believe that only large organizations with a steady cash flow could bring AI solutions into healthcare. I’m happy to report that there are an increasing number of niche, medical bot solutions. It turns out that healthcare and chat bot services are more complementary than I initially thought.


Shwen Gwee of Novartis shared five points that illustrate why this technology can work even within a notoriously regulated industry:

1. Many aspects of healthcare are already driven by decision trees that bots require:
coders can look to pre-established treatment guidelines that have been professionally and scientifically validated.


2. Many pre-approved responses/scripts already exist in some form:
coders don’t have to reinvent the wheel when mapping potential conversations.


3. Experienced doctors rely on pattern recognition:
teaching bots to recognize the same patterns can help patients make informed decisions about whether or not to seek care.


4. Healthcare systems are over-stretched and have limited availability:
chat bots can give you their full attention 24/7/365.


5. Right now, even the humans in healthcare sound like robots:
many call centers set up for patient assistance get feedback that they sound “cold” or “robotic.”


High-quality conversations in healthcare have limitations, they’re episodic and hard to pin down. It’s common to wait two weeks to get a doctor’s appointment only to sit down and have their attention for 15 minutes. In this paradigm, many people are left with more questions than answers. The healthcare system needs more intelligence to help get these people the information they need to successfully manage their conditions. There are three potential solutions to this intelligence issue: throw more people at the problem, replace people with automation to solve the problem, or enhance and augment HCP conversations with the use of AI. The first two options would require major financial or IT implementation overhauls, both of which typically scare off procurement leads.


However, AI augmentation through bot technology can be implemented in a relatively seamless manner. One of the biggest advantages of a bot solution is that it can be easily integrated into the apps we already have installed, like text and messenger. Think about it, when was the last time you downloaded a new app? If you’re like the majority of Americans, you download a whopping zero apps a month on average.2 By integrating with the communication platforms patients and HCPs already use, conversational AI doesn’t have to go against the grain of our baseline technology behaviors. Bots are also extremely effective in cost reduction, especially when triaging care. Here are some statistics that should give us all pause:

• 70% of doctor’s visits are avoidable3

• 71% of trips to the ER are non-urgent4

• 31% of urgent care visits could be solved with a phone call5


It’s wasteful, but people who seek unnecessary medical care are doing so with sincere intentions. Andrew Le from Buoy Health shared that patients’ perception of their illness and its severity is the No. 1 factor that drives them to seek care. It’s not easy to decide what to do, especially when a quick Google search for a headache can quickly lead to a WebMD diagnosis of brain cancer. Even Google acknowledges that this process can be frightening and difficult to navigate.6 This means that to prevent unnecessary medical intervention we must first change a patient’s perception of the symptoms they’re experiencing. That’s where the augmented intelligence of a chat bot can help. Although we were lucky enough to hear from leaders at Healthtap, The Neon Project, Conversa, Infermedic, and Buoy Health, I’ll focus solely on Buoy for the sake of brevity.


Buoy’s bot interviews patients over the course of two to three minutes about their symptoms. When the interview is completed, Buoy gives patients a series of likely conditions based on how past respondents were ultimately diagnosed. The service always makes sure to clarify that it can’t make a formal diagnosis and doesn’t offer medical advice, to avoid any legal complications. Ultimately, it’s a tool that patients can use to help make decisions about the type of care they should seek. If a patient is using Buoy through their employer and decides to seek care, the service will actually refer them to an appropriate, curated service provider. The results were incredibly encouraging – Buoy was able to de-escalate 35.4% of ER visits, 40.7% of urgent care visits, and 23.8% of primary care visits.


All of the services previously mentioned have provided similar results with triage, appropriate escalation, and general health literacy. The current caveat is that most of these bots address symptomatic patients or help with one overarching disease state, rather than representing a brand. While bots certainly have the capability to serve pharmaceutical companies and their individual brands, many in the industry aren’t comfortable with the risk (yet). However, I’m optimistic for the future of pharmaceutical bots. In two years 85% of all customer service conversations are predicted to be facilitated by AI conversations.7 As pharmaceutical marketers, it will fall upon us to create case studies, become bot-literate, and educate our clients on the decreasing risks and increasing benefits of working with bots to help improve patient outcomes. I highly recommend doing some research on the solutions provided by Healthtap, The Neon Project, Conversa, Infermedic, and Buoy Health to get started.