The Chatbots Are Coming!
A mainstay of customer engagement is now making its mark in pharma
If you keep abreast with political developments and have a twitter account, you might be under the impression that chatbots are the enemy. Yet despite attracting headlines for peddling propaganda on social media, chatbots are transforming customer experience. The global market is set to exceed $1.34bn by 2024, according to a new research report from Global Market Insights.
The mushrooming market is now disrupting pharma and the wider healthcare industry. As digital health becomes de rigueur, chatbots are leading the march. This article will explore what is happening on the ground and what the future might hold for this burgeoning area.
The applications are wide-ranging and potentially game-changing, as Vyom Bhuta, Global Partner, Commercial Life Sciences at Cognizant, explains.
Important use cases include companion diagnostics; chatbots could detect and diagnose disease by describing the symptom (Ada and Babylon are making waves in this space). Interactive discussion guides; step by step instructions for therapy administration or structural procedures. Customer service bots handling complaints, product and service inquiries, as well as adverse event reporting.
Coaching for patients and HCPs is another key application. Text or voice driven guidance can be provided for therapy and lifestyle plan reminders and motivation. HCPs could receive daily briefings, such as relevant continuing medical education events (CME) for a particular therapy or drug.
Educational information around the condition that a drug might be treating would also be an invaluable resource for HCPs and patients, while providing credibility to a brand.
Bhuta also singles out some specific benefits to US pharma operations, such as order management. As opposed to going to a website, conversational AI could streamline the process — capturing, tracking and ordering literature about products through a simple request. It could also ease the burden of benefit verification. When prescribing a therapy to a patient, HCPs have to make sure their insurance can cover that particular therapy of choice. This piles on a lot of admin. Conversational AI capability could help verify coverage copay and also get the patient enrolled into a financial assistance program, reducing the paperwork.
The sales rep is also set to benefit, says Mark Bard, Founder of the Digital Health Coalition. It could help support a typical scenario such as sitting in the car prepping for a meeting with a doctor. The sales rep could ask the chatbot, “Can you give me an update on Dr. Smith?”, and the bot will bring up all the latest information on Dr. Smith, for example detailing the events he attended in the past three months and the recent inquiries he has made to the call centre.
The Digital Health Coalition recently conducted a study with partner Klick Health to gauge HCPs enthusiasm for greater adoption of chatbots. The results are auspicious.
Within their own practices, they see chatbots as automating a lot of the admin drudgery such as appointment reminders and they also see it as invaluable educational tool for patients. It also scored high on health management such as nutrition and wellness tips. HCP’s eye a golden opportunity in the diabetes space where behavioural cues are critical.
The jury is still out on whether chatbots could and should carry out companion diagnostics, however. HCPs expressed concerns about the lack of human emotion involved in making a diagnosis.
Bard is more optimistic. “These are things that we'll figure out in the next decade when we get more comfortable saying, ‘I don't need emotion if I have a sore throat. Maybe I need emotion if I am confirming a cancer diagnosis’”, but signposting towards allergies seems perfectly acceptable.
A finely-tuned feedback loop
Pharma’s current processes for harvesting data leave much to be desired. That could all change with the introduction of chatbots, says Bhuta. “The feedback loop for pharma right now lags quite a bit. They spend a lot of money on market research but a lot of it is dated and they use Awareness, Trial and Usage studies (ATUs). It is a Q&A — it's factual based but it's subjective. It's what the doctors are thinking about as they're answering some of the questions.
Chatbots allow pharma to learn about their customers across the journey at the point of interaction or at the point of need. You can start segmenting the different types of customers and tailor the types of questions.
“And this feedback loop can really go down into what do we need to do to better educate. Because we're getting a lot of questions for this step in the patient journey or the customer journey. They can start to tailor programs of awareness, education, interventions, all the things that pharma normally does. But now they can do it based on almost real time intelligence. You can start uncovering patterns.”
The technology could also help to better coordinate sales reps on the ground with the marketing campaigns, says Bard. When a rep comes out of an appointment, they could feed back to the AI bot the key points the doctor spoke about and what they would like to receive next. This could result in automated trigger emails and better tailored content.
Bhuta is mindful of the compliance and transparency issues this raises but looks to the tech giants as a barometer. “We give consent to Facebook or Amazon so that we're getting a service and it's helping us and we're fine with giving that amount of information to Amazon or Google or whoever. It's a similar thing for pharma and life science companies; they can be transparent but if they provide value at the time of need, I don't think it's an issue because you know so there's value on both ends.”
Who is doing what?
- Novo Nordisk launched a chatbot called the Ozempic® App (in some countries under a different name; e.g. in Denmark under the name “1 gang Ugentligt” meaning “Once-weekly”). It is for patients with Type 2 diabetes. To ensure it was user-friendly and intuitive, the company asked patients what apps they regularly use and modelled its interface based on their responses. “This was crucial to ensuring the experience didn’t feel too pharma heavy and corporate but could be integrated into the patient’s everyday routines,” says Herluf Nis Thomsen, Global Product Lead, Ozempic Project, Novo Nordisk. The interaction is powered by a pre-determined set of algorithms that offers users a “decision tree” of multiple-choice questions and answers about the treatment. Novo Nordisk opted for a closed loop system as opposed to an open-ended machine learning approach as there were ethical concerns about steering users outside of a controlled environment. There is also a thicket of regulations to contend with. “The question is, would you leave Alexa or Siri with questions or leave them to answer anything which is related to your health and well-being?” Crucially, as Thomsen points out, the numerous questions and answers still creates a sense of personalisation.
- Sanofi launched a chatbot in France called Nina. It was introduced to help people with mild sleep disorders. It provided 24/7 responsiveness and recommended certain user users to switch to Night Shift, an Apple iOS function that shifts the colours of your display to warmer colours (less blue light) after sunset7.
- Sugarpod is a voice-enabled diabetes care plan from Wellpepper, which helps consumers record low blood sugar levels, mealtimes and tracking their medication adherence. And all of that can be made available based on consent back to their HCP so it is supporting the patient and the HCP.
- JNJ launched Andy, a virtual assistant chatbot powered by artificial intelligence (AI). Andy is designed to help guide U.S. consumers throughout their ACUVUEBrand Contact Lens journey — from those considering contact lenses for the first time to long-term wearers. The chatbot also provides intuitive coaching to help new wearers develop healthy contact lens habits.
Work in progress
Getting the organisation aligned to stand up this type of capability and to do it in a compliant way takes a lot of conversation internally, says Bhuta. Organisations also need to have robust mechanisms in place to be able to detect an adverse event through these channels. Safeguarding against such risks slows down development.
There are also foundational hurdles such as getting to grips with master data management. Companies haven’t yet mastered how to uniquely identify patients. Capturing patient interactions is still stymied by privacy laws, outmoded data management and data infrastructure.
Market regulations can be a huge bottleneck too. Companies are limited in what they can say about a product in certain markets. Companies can only say what the drug is approved for and not mention the drug itself. This makes marketing a chatbot around a drug a tall order. It also means that a lot of the more nuanced questions that patients would like to ask a chatbot such as potential side effects, are off limits.
Novo Nordisk encountered this problem when launching its Ozempic® Appin Denmark. The regulatory authorities said the app couldn’t include the product name in its logo, so it is called ‘Once-Weekly’ in Denmark. While this complies with the rule of not directly marketing to patients, it feels counter-intuitive, as a patient would search for the product name in the app store.
One way to circumvent the problem could be to launch a chatbot that provides medical education about a condition and is sponsored by a drug company — a form of indirect advertising, posits Bard. But what you can and can’t do is still very much a grey area that is still being mapped out.
There are also external barriers to contend with such as the deluge of information HCPs and patients are faced with. A lot of this information comes from third-parties and not pharma companies so it is hard to incorporate another information stream into people’s everyday routines.
Consumer trust is also a challenge, says Bard. “Consumers have a choice. Do you as a patient want services from a pharma company or do you want it from someone that you view as independent when it comes to disease education?”
Beyond the bot
If pharma companies are to truly integrate their services into the everyday lives of HCPs and patients, Bhuta believes the answer lies beyond launching their own chatbots. Pharma must go to where the customers are and support the “partnership ecosystem”.
Pharma should input some of their algorithms and make it an accessible service through the common AI tools that are already out there.
“If Amazon right now has a platform which is compliant and can open the service up to pharma, pharma should be providing its service so when a HCP or a patient talks into Alexa or the Google assistant, the information and the algorithms that power these conversational AI services could be powered by pharma.”
This will require a step change in the way pharma companies think about branding. They can continue to pour money into websites and various initiatives, but they must think about where they will have the most impact. They must be cognisant of where they want to make information available and the best approach is to understand where people are going to ask those questions, says Bhuta.
This is necessitating a more open-minded business model. Bhuta points to the creation of digital functions within pharma companies as a sign of progression. “Companies are thinking about how they can join hands with other companies and other institutions to create a more integrated ecosystem and bring digital solutions and data science to life.”