The Healthbot Arms Race is Coming
No matter where you are in the world, there is never enough healthcare to meet demand. There are too many human bodies out there to be maintained, tended to and healed. But we’re living in an age of automation and AI. Couldn’t there be an answer in technology?
Havas LYNX’s Interactive Director Adam Emmott thinks so. “With healthcare systems being pushed to their limits, there is a need to understand and implement how intelligent, AI-driven chatbots could help ease some of the burden and perhaps improve data infrastructures,” he says. “As we move in to new ways of delivering better ‘beyond-the-pill services’ in healthcare, the chatbot is emerging as an interesting opportunity to support.”
With healthcare professionals stretched thin, live face-to-face time with a healthcare professional can be variable. “These live conversations can be poor quality, average quality, or amazing, and the reality is that not enough are amazing,” says Dr John Reeves, Chief Medical Officer at conversationHEALTH, a Toronto-based startup that builds chatbots for healthcare and adjacent sectors. “These one-to-one conversations cannot scale to meet real time customer demand,” he adds. He believes branded conversational AI can step in to plug this gap. “Chatbots allow brands to capture the very best conversations (and those that are on-brand message) and scale to millions of end customers. And they do so in an on-demand, personalised way.”
You may have already encountered services like Babylon, Sensely and your.md that demonstrate the power of tracking user healthcare data in meaningful ways, but also using that data to make meaningful diagnoses. “It is still early days,” admits Adam, “but with the use of machine learning and huge data lakes of usable medical information being collected, it should be inevitable that we could see much needed support from chatbots in healthcare provision.”
The end game of that thinking is hard for us to countenance in 2018: a world where you can chat to an AI about your symptoms and receive a diagnosis that you can trust. Adam imagines the power of good conversational AI in the mental health space. “Supported, meaningful, conversation obviously lends itself to provide support in talking therapies. When we are able to create truly sensitive, personal conversations available whenever a user needs it, we could be onto something incredibly powerful. Connect that with deep learning, we begin start something groundbreaking in the understanding of mental health, from provision to diagnosis.”
That’s getting ahead of ourselves though. Chatbots aren’t likely to replace your doctor anytime soon. “There isn’t really a strategy around AI and bots that I’ve seen any health system or pharma company put together,” says Ritesh Patel, Chief Digital Officer for Health and Wellness at Ogilvy. “You’ll see point solutions coming through, not necessarily a strategic view of using this technology to really change the delivery of care model. Entrepreneurs will go after the market faster than health systems will. They’ll see a need, see how much the area is worth and go for it.”
Pharmaceutical brands are taking the chatbot arms race seriously, as Dr John is witnessing. “Compared to the three previous big digital opportunities (web, apps, social), conversational solutions (messaging, chatbots, voice) are gaining traction at a faster rate,” he says. “Pharma sees massive potential in the use of AI across the enterprise, including marketing communications to reimagine the way they engage their customers in a more ‘modern’ way. I mean, texting and talking is how life is lived now. No one wants to be searching or navigating through web sites, articles, brochures, videos to resolve a specific need.”
Ritesh observes that there are two arms races among the pharma clients in this space. One of them is the race to be first to market. In this case, being first is valuable, he believes: “You’ll be the first chatbot people use and they’ll remember you’re the one. And when someone else comes along, unless they have something amazingly different or unique to offer you, it’s much less likely you’re going to use it.”
The second arms race is the technology and the ability for image recognition, AI and natural language processing. “I think it’ll drive a lot of these bots to become voice-activated bots,” says Ritesh. “Right now while it’s chat and you’re typing. I think you’re going to see an explosion of voice. Where now you see people saying ‘OK Google’ and ‘Hey Siri’, you may get a doctor saying ‘Hey Pfizer’ or ‘Hey Merck’.”
Chatbots from pharma brands will appear in doctors’ offices before they appear on patients’ phones more generally, Ritesh predicts, and he can imagine how they might make physicians’ lives significantly smoother. “The doctor doesn’t wait until the pharma rep arrives at 2pm on a Thursday to hold all their questions. The brand’s website is highly complex and tough to navigate to get the content you need. So if I’m doing my rounds and I need a bit more information about the drug I’m going to prescribe to you, I’ve got two ways I could do that - look it up on the website of the pharma company and try and find the dosing information or I could just whip out my phone and say ‘Hey Merck M.D., what’s the dosing information for this medication? I’ve got a 55-year-old female who’s taking this medication and this medication. Can I also give them this? And how many milligrams?’ So that instant access to bits of data that I need during my day is the value proposition for a chatbot for healthcare professionals.”
Naturally, this is great marketing for the pharma brand with the best bot. Whatever makes doctors’ lives easiest will benefit from doctors considering that brand’s products more often.
There’s also a compelling argument for chatbots’ use around clinical trials. Chat apps on a patient’s phone could make the collection of data more natural as well as reminding patients to take their drugs when they need to and report any side effects. Many jobs that once took up a nurse’s time can be automated.
“It used to be that I gave you a book and you wrote down everything that occurred on a daily basis,” says Ritesh. “Now the bot does that. Anyone who has a mobile phone can install it and you can log your day with this medicine you’re taking during the trial. It also checks in with you - a virtual nurse: ‘How are you feeling today? Did you have any adverse effects?’ Sometimes people don’t report them because they don’t think it’s a side effect.”
Alongside the pharma brands, we are likely to start seeing chatbots emerging from advocacy groups for specific conditions to raise awareness around these diseases. In the UK the Asthma Foundation recently launched a chatbot for living with asthma that was sponsored by Boehringer Ingelheim called Tabatha. It’s a Facebook Messenger bot created to educate and help people. It ran as a pilot earlier this year and Ritesh found it interesting how well it was received. “It’s a generational thing. The millennials are adopting these things in droves because it’s part of their whole workflow. They’re good with things like chat because of text and Snapchat and all these apps.”
The concept of a chatbot has been proliferating for years now and we see them all over the web in customer service roles and it is now easier than ever for health and pharma brands and their agencies to build them. Adam notes that services such as Dialogflow and Bluemix do “most of the heavy lifting” for developers these days. “The challenge, as always with digital, is making them good! We are in that awkward toddling phase like we have seen with all new tech. We may have small pockets of awesome in our agencies, however, making those function across all disciplines is the challenge. As the development tools get better, functionality will undoubtedly improve. Hopefully, within the next 18 months, we’ll start to see chatbot personalities coming from outside of the big players and offering real value in healthcare research.”
The decisions the industry makes now will be vital, setting the bar by which conversational health solutions are measured for years to come, until we reach that point when we have trustworthy AI doctors. “Ultimately, we will begin to see patients ‘converse’ with their digital twins, a bot that will understand who we are, what we like and what healthcare issues we have faced in the past,” says Adam. “When considering the build of these kinds of interactions, I often think of the base interactions as ‘starter yeast’! Whatever core interactions we create and define now will form the ‘personalities’ of our chatbots of the future. Now more than ever, we have to get these foundations correct.”
Ritesh realises the challenge here comes from brands’ motivations. “Most marketers want to tell you stuff. They don’t want to help you. So how do you fix it so that you’re having a conversation and that leads to you giving me something of value so that I keep coming back to your brand? That builds a brand.”
Bot agencies tend to compete on the tech side and it’s easy to get lost in functionality, but as Dr John recognises, “all tech does is allow anyone to create a so-so bot. The magic lies in the conversation strategy and execution within this highly regulated industry (and how technology can enable this).”
One of the main reasons healthcare has always lagged behind in the innovation space is that it’s important. It’s literally a matter of life or death. So naturally, innovation by health and pharma brands faces a lot of scrutiny.
“As soon as we open up this area of conversation though, we inevitably end up in rabbit hole of ‘how much can and should we know?’, ‘who owns my data?’, and ‘how the hell will we approve this?!’” laughs Adam. “The major challenge for pharma clients is the age-old issue of compliance. How do we create meaningful, compliant experiences for our customers that go beyond the linear multiple-choice pathways provided by most pharma chatbots?”
Giving medical advice via an AI? That’s scary for corporate lawyers. “If you give bad advice and somebody dies, then what? You have to be overly cautious. Anything that slaps of providing medical advice is going to be difficult,” says Ritesh.
But difficult doesn’t mean impossible. Dr John denounces the perception that pharma can’t play in the messaging space. The fear is that AI might generate conversations that might include incorrect or off-message information and put the client at legal risk. “The reality is that pharma solutions can leverage AI in a different way,” he says, “meaning that all conversations are pre-structured and pre-approved and AI is used only to understand the user questions and then serve up the right conversation (from an extensive library which is pre-approved).”
Another huge paranoia is the great bogeyman of 2018 - use of personal data. Health issues revolve around the most personal kind of data - age, location, lifestyle choices - and chatbots often end up on Facebook Messenger as a natural environment for them to operate in. “For healthcare, the question is in the access to usable data sets,” posits Adam. “How do we let private companies have access to personal information? Even with the current level of chatbot capability, they could help provide the much-needed improvements in infrastructure. Enter blockchain. Once we have patient data available on a robust, secure and universal platform then maybe we can start to think of a future with android doctors.”
The final challenge for health chatbots is a more subtle, human one. Do you trust a robot doctor? “On the patient side I think trust is going to be enormous,” says Ritesh. “I think that’s why some of these bots in healthcare will come from patient advocacy groups, which is a trusted resource in patients’ lives if they’re living with a disease. They’re more likely to trust it than something coming from a pharma brand.”
None of these challenges are insurmountable though. Pharma brands are undeterred. Ritesh has been watching this space and predicts that we’re on the brink of something. “In 2019 you’re going to see an explosion of healthcare professional bots from pharma companies. The advocacy groups will be a bit slower. It will reach healthcare professionals before consumers.”
Dr John and conversationHEALTH are actively working with eight global pharma companies. “All are aggressively moving into the conversation world, he says, “with two quickly moving to voice solutions for healthcare professionals. Our belief is that all brands will have conversational solution (just as all have websites). Why would a user want to search for answers when they can simply ask and get one immediately?”
