Bots in healthcare: interview with Thomas Schulz, Organiser of Botscamp

Digital health and mobile health apps have been a hype topic, ever since Apple’s App Store began the app-craze in 2008. The initial hype about mHealth has now cooled, which is good news in a way because it shows that mHealth has made the leap from hype to reality. The hype of healthcare apps has since been replaced by other hot topics. So, what is the “next big thing”? Artificial Intelligence (AI), Virtual Reality / Augmented Reality, Blockchain and Bots/Chatbots. We want to have a closer look at chatbots in healthcare. What are these so-called chatbots capable of doing? What is the current market status of chatbots in healthcare? And how might they be able to shape the healthcare industry?

The recent bot-hype began in April 2016 when Facebook announced at its F8 developer conference that the Facebook Messenger app would be able to feature bots from outside brands. Shortly after the conference, bots gained the reputation as being the “next big thing”, as fantasies led to a plethora of possible use-cases, including healthcare.

Chatbots have since been around for quite some time, dating back to Joseph Weizenbaum’s computer program “Eliza“ in 1966. Chatbots generally work in a non-complex way (think of a customer service phone menu). Until today, most chatbots can only answer simple questions, like basic product or service information. More advanced chatbots are connecting the communication interactivity with the ability to learn — Artificial Intelligence. AI capability is producing a new generation of chatbots which become smarter the more they are used by learning from conversations, remembering previous dialogues, and using that knowledge during real-time conversation. This ability to learn from past experience enhances the chatbot’s ability to interact more effectively every time it is used. More usage leads to more knowledge and to better results.

Messenger apps as a platform

The new generation of chatbot is an additional layer on-top of the previously established eco-system of messaging apps. Similar to the way mobile apps are built on a layer of operating systems, this additional layer to the existing eco-system of messenger apps enables better daily usage rates, improved consumer adoption, and far greater app downloads.

The top four messenger apps (Facebook Messenger, WhatsApp, Viber, WeChat) combine nearly as many users as half the planet earth: 3-billion! Jim Tomanchek, social media and mobile advertising executive states on that there are more monthly active users of messenger apps than social networking apps. Only the rise of messenger apps, which serve as a platform, made the rise for chatbots possible.

Chatbots in healthcare

In healthcare, healthbots are used in multiple ways, mostly to take over systematic and repetitive tasks like booking appointments, checking patients’ IDs, checking insurance information and coverage, asking for medical history and drug intake, and giving information about side effects and drug interactions. In hospitals, bots can provide menus for patients with special diets, follow-up with treatments and adherence for outpatients, and renew subscriptions.

Here are some examples of healthbots:


Your.MD is a “chatbot-only” mobile app. The bot checks your symptoms, asks for information and gets you help for your symptoms. Your.MD aims be a “Personal Health Assistant” in the user’s pocket, which is available all the time. The app is using an advanced Artificial Intelligence algorithm.

Babylon Health

Babylon health, again is a chat app which asks the user questions about their symptoms, presents a diagnosis, and gives quite detailed information on how to treat the sickness. In addition, the Babylon bot will book a consultation with a doctor, if needed. The app is based on self-learning AI technology.


Joy is a Facebook-chatbot for mental health that encourages users to interact, and acts as a gateway to healthcare professionals if need be (not intending to replace them).


Melody is a chatbot created by Chinese search giant Baidu. It helps healthcare professionals to take care of patients via text messages. Melody makes medical consultation possible from anywhere. The app’s goal is to find out whether a user should see a doctor personally or not. Melody is integrated in Baidu’s Doctor app, which already lets users make doctors’ appointments and find answers to health-related questions. Melody’s answers are based on various sources (textbooks, forums, research papers) and comes up with a hypothesis for treatment that a doctor checks and validates.


Eva is a chatbot for womens health. Users enter some data (e.g. last period, pain, hormonal contraception). The app will ask for so-called “mood bits” every day directly via the messenger. Based on this data, Eva can help women to understand their bodies and adapt their lives accordingly.

Use cases for healthbots as of today

Today’s healthbots (even the most advanced) can only take over certain tasks. If we look at healthbots from a patient-journey perspective, bots are great for booking and preparing physician / or hospital visits, and possibly some of the tasks post-consultation or discharge. All tasks that require patient interaction and intuition still have to be done by the physician. In other words, the current status is seemingly at an “assistant level”.

What else can they do? Are chatbots just another hype topic that will blow over soon? Or do healthbots have the potential to disrupt healthcare?

To find out we talked to an expert, Thomas Schulz. Thomas is a technology and healthcare expert and the organizer of the world’s first global online conference on chatbots.

Research2Guidance: Can you give us an overview of where chatbots are used in healthcare right now?

Thomas Schulz: Chatbots are already working successfully in healthcare; mostly in doctor-patient communication where the chat delivers a first-level exchange in the sense of a preliminary talk. Another field where bots are helping out is in building digital communities where health experts exchange with patients about certain topics, and chatbots act as a welcome and help desk. Chatbots can also help to respond to frequently asked medical questions, where the chatbot delivers information from a peer reviewed server, 24/7 support.

Here’s one good example of a healthbot; in the UK the NHS has just started a chatbot service based on Babylon’s chatbot app. The app, which caters for the London area, is an alternative to the local emergency number 111, and acts as a non-emergency helpline.

Another example is the building of a healthcare community via @ShailaBot on Twitter and Slack as a pilot project — a project that I have been involved in personally. The goal is to bring together healthcare experts to exchange ideas about publications, research, congresses, etc.

Research2Guidance: The promises of chatbots in healthcare are huge; bots are supposed to deliver healthcare services to reduce costs, improve the quality and effectiveness of healthcare and make healthcare knowledge accessible (and free) to anyone, anywhere, anytime. What is the reality behind these early-stage hopes and dreams?

Thomas Schulz: Like every bot project, there are always the “initial questions”: is there a common process which is causing a lot of unbillable person-hours? Or for which the repetition of asking a series of questions is causing “tiredness” among the custodian and listener, and thus reducing the quality of the interaction? This human fatigue is not on purpose, but is sort of an overused helpline effect. The healthbot jumps in right there — taking over the communication between the two parties, offering recommendations, encyclopaedia services, or supporting interactions between target groups. At the moment, these bots still work on a basic level, but with the enhancement of artificial intelligence in the next months, bots will serve as an automated valuable add-on to the daily work of experts. The purpose of the bots will be to provide an authentic and understandable health services rather than trying to be a humanized health machine service.

Research2Guidance: Some of the bots are quite advanced already. How quickly is technology developing? What are the typical use cases now? What will be possible in the near future?

Thomas Schulz: The growth of healthbots as a technology is exponential — changes pop up in months or in sometime days. Even though we know speed is not everything, it is very characteristic of the changes within the bot world. The typical use case is focusing on dialogues between patients and experts. The future will present smart dialogues with artificial intelligence, covering more than basic answers and questions by looking into, for example, an encyclopaedia. For instance, IBM Watson Health — this is an impressive cognitive system reinventing the interaction between humans and computer. In the area of oncology, IBM Watson, within no time, succeeded in searching through literature and EMRs (electronic medical records), which provides physicians with complete evidence-based treatment options, at speeds for which humans can presumably not exceed. Here, artificial intelligence turned unstructured data into recommendations. At the same time, the system is advancing incredibly thanks to the ability to continuously improve with every use. This is for sure, a glimpse of the future.

Research2Guidance: A lot of mobile health apps are not used. Chatbots are a more natural way of digital communication, nevertheless, still artificial. How is the user acceptance of chatbots?

Thomas Schulz: The weird feeling of talking to a bot will disappear, as with every technology update comes a time of user behavior change, and a growing acceptance by users. I do not consider this to become a natural way of communication, but an accepted information transfer in an easy, 24/7 and fascinating and maybe gamified discovery and learning process for target groups. Since the bot-hype is still young, evidence-based studies will show up this year and 2018, and will all be about user acceptance. Personally, I have found that the majority of users’ trust healthbots.

Research2Guidance: Let’s think of a typical patient journey in a physician’s office or in a hospital. From booking an appointment, to sitting in the waiting room, to checking vitals, to probing the patient, offering a diagnosis, writing a prescription, leaving the clinic, taking prescribed drugs. Where are the biggest potentials for bots in healthcare?

Thomas Schulz: In fact, in all of these steps, when visiting a physician or a hospital about an ambulant or stationary treatment, or visiting a physician’s office, healthbots are applicable. The biggest potential for healthbots will be leveraged when they are able to provide extensive smart data services and dialogues for different tasks, and connect to a conversational interface to replace  typing with voice.

Research2Guidance: And what are the barriers that bots are facing and may not be able to overcome any time soon?

Thomas Schulz: NLP (Natural Language Processing) shows up often in combination with the buzzword AI. Artificial intelligence is the next frontier for bots. So far, a chatbot has difficulties in e.g. recognizing irony, sayings, or different tones of voices in a chat. In comparison, the physician in a face-to-face talk won’t be able maybe to search and find all evidence available about a diagnosis, but is fully capable of having a decent and empathic talk with the patient. This kind of high-level and holistic communication is a field of constant development, could be done in two years.

Research2Guidance: The disadvantage of bots of being impersonal can become a big advantage when it comes to diseases like mental health or STD. People disclose information easier to a machine then to a human being. Do you have any experience with this phenomenon?

Thomas Schulz: The comfort of talking to a healthbot, anywhere, anytime and about everything without the stress of sitting in an examination room, or thinking about questions and answers, has opened up the phenomenon of digital trust. Not only in healthcare, but as a further evolution of behaviour learned in chats, via SMS, telephone helplines, we are now reaching the next level with these bots.

Research2Guidance: It would be quite easy to imagine that in a physician’s practice or hospital a bot takes over tasks like welcoming patients (as some bots / robots are already successfully doing in hotels), asking about symptoms, drug-use, even coming up with a diagnose. When it comes to treatment, it gets more complicated. We are entering a whole new field here. In most countries, regulations are not allowing bots or Artificial Intelligence to make official diagnoses. How about legal issues concerning health bots?

Thomas Schulz: The bot-hype is representing a new universe of legal challenges, and new laws need to be defined. There is much work to do, as the market moves fast. And yes, when a patient-doctor talk is asking for conclusive diagnosis, through to treatment, cautiousness is already an indispensable element, even without healthbots. Just think about the risk of providing a medical recommendation by a healthbot causing a person to take a wrong pill. There will be regulations in the future providing some kind of certified, approved, or cleared healthbot; like the FDA administration in the US, which has opened up the Digital Health framework to “apply evaluated rules”, much like they have done already for mobile health apps.

Research2Guidance: The WHO predicts a health workforce shortage of 13-million in the coming decades. Hopes are high that digital health (as well as healthbots) will be able to help to solve that problem. Do you think legal restrictions concerning bots will have to be loosened or will loosen any time soon?

Thomas Schulz: The shortage of health experts is an issue I know from my own experience in the Swiss healthcare landscape. At the same time, there are new discussions about bots coming up like: Will they take away jobs? Do they have to pay taxes? Will they even become holograms at homes, inevitably supporting the isolation of human beings? I agree that there are high hopes for healthbots to deliver true value to certain processes in daily healthcare; to guarantee basic services in cities and remote places in spite of challenges to “human resources”, to cope with the challenges of “profitability” connected to tariff systems like the DRG (Diagnosis Related Groups between hospitals / healthcare providers and insurers/cost units) or to have more time with the patient after the healthbot did all the basic work.

Research2Guidance: Thank you, Thomas, for your answers.


About Thomas Schulz: Thomas is a healthcare expert, digital nomad and founder of

About botscamp: is the only digital platform about healthbots, and all other bots, AI and machine learning. presents 14 speeches every 3 months and has an online conference which can be accessed without the hassle of booking tickets, paying for travel costs or using your valuable time to travel. provides a 360-overview about the bot-hype, creating a space for bot makers, women in bots, bot researchers, bot investors as well as digital nomads and co-workers. In the online conference we include our rapid prototyping skills and design thinking methods.

  1. Noa says:

    NLP in this context is Natural Language Processing, and not “Neuro Linguistic Programming”. The latter may *sound* like it might be related to AI, but it most definitely is not…

  2. Nice interview .. your should know that conversationHealth in Toronto has been working on bots for healthcare and in particular Pharma with us at Ogilvy for a while. We held the first healthbot conference in Toronto in April, London in June and NYC in September.