Imagined conversation with a health chatbot
- Hi Viola, how can I help you?
- I have a mild fever, headache, I feel dizzy. What could it be?
- What is your temperature at the moment?
- For how long have you been feeling the headache?
- In the last 2-3 hours.
- Do you have any other symptoms?
- No, I don’t.
- You had the exact same symptoms last week and the week before. This time, I recommend you to visit your GP. Let me call your doctor’s office and make you an appointment that fits your schedule!
With the appearance of intelligent personal assistants (IPA) supported by machine learning such as Siri on iOS or Alexa for Amazon, the above scene does not sound like science fiction anymore. If we also consider how Cognitoys support the cognitive development of small children with the help of AI in a fun and gentle way, personal health assistants on our phones suddenly become definite responses for certain needs. They can make our lives more comfortable and they could pay attention to our very personal wishes through constant learning – which might also have some downsides.
Half a million people professed their love for Siri
Remember the movie entitled Her? The main protagonist, Joaquin Phoenix falls completely in love with the voice of a digital assistant capable of learning at astonishing pace as well as fulfilling his every need. Luckily, we are not here yet. But according to a recent report in the New Scientist, hundreds of thousands of people say ‘Good morning’ to Alexa every day, half a million people have professed their love for it, and more than 250,000 have proposed marriage to it. So we have to be careful, and already have to analyse the complicated relationship humans have with intelligent machines and all already existing forms of artificial intelligence. Preferably before they overflow our life. So let’s start with digital health chatbots.
Chatbots are the future of messaging
Chatbots are already everywhere. Okay-okay, you are right, not everywhere, but sometimes the supply cannot even appease the huge demand. For example, around the holidays in December 2016, Amazon Echo was such a ‘bestseller’ that the e-commerce giant’s main website and app has been sold out of the Echo voice-controlled speaker for weeks. These bots talk to their users through speakers, and their built-in microphones hear from across a room. When Echo hears the name “Alexa,” its LED ring lights up in the direction of the user to acknowledge that it is listening. It answers questions, plays music, orders Amazon products and tells jokes. Mind-blowing! Obviously there was a huge interest, which in turn shows a clear trend: chatbots will take over the place of messaging applications in many areas and they will also hit such market segments, which were previously unimaginable. Such as healthcare.
Digital personal assistants in the form of chatbots offer a lot more than simple messaging apps. They can be voice controlled, which makes it possible to use them also when you’re hands are full with laundry or you don’t want to put down your crying kiddo. They come in handy when you want to play music, order food or take notes. As they are voice controlled, I can imagine that after a while they will be part of the dynamics of family communication.
Why will health chatbots become a success story?
There are so many simple medical questions out there which do not require the full attention of a physician, but letting them remain unanswered leaves the concerned people nervous, confused and clueless. Imagine a young mother with a two weeks old baby without any experience. She will bombard you with tons of questions, if you ask her how she and the baby is. What is the temperature for a baby that requires intervention? What is the ideal temperature for the baby’s bath? How much sleep is enough? Is it better to leave a sock on the little one’s hands for a while after being born or not?
These are all relevant questions, but they usually do not need the response from a physician. A health visitor might be a better choice, but the young mother does not always have that possibility. Moreover, doing Google searches to find the right answers is also challenging as most patients do not know how to assess the quality of information they find online. Plus the Internet is full of fake news and misleading sites that want to sell something.
A more expanded primary care service cannot and should not fill this gap. Patients should not leave from work, travel to the GP’s office just in order to get a 10 seconds response for every simple question. At the same time, no symptom checker website can help in a reliable way, as they are not personalized enough or not driven by algorithms smart enough. An instant messaging health app cannot fill this gap effectively either, as we cannot expect that whenever we have questions, there is a physician available somewhere in the universe.
The solution is a digital health chatbot constantly supported and gradually taught by physicians.
Where does the development of digital assistants and intelligent chatbots stand at the moment?
Since the introduction of Siri, Apple’s voice assistant and ‘knowledge navigator’ in 2011, every tech giant came up with its own version. Just think about Microsoft’s Office Assistant and later Cortana, Google’s Google Home and Google Now or Amazon’s Alexa or Evi. Their evolution is a constant factor and they all ‘possess’ a certain personality with their own names – well, except perhaps Google’s assistant. The independent, non-profit research center, SRI International, the inventor of Siri is working on how to make the conversations with this digital platform more ‘human-like’ and how such machines might get a clue about the emotional background of human conversationalists.
The development obviously goes into the direction of making chatbot conversations more and more realistic, but we are definitely not there yet. Low-capacity AI such as these digital assistants are rather in their toddler phase. They are learning from the world around them by watching and listening, while they respond with their already existing knowledge repeating our prejudices and bias. And as you cannot expect a toddler to give you advice about your divorce, you cannot expect empathy from a digital assistant – not yet.
However, I’m always very curious to test where technology stands, so I did it with chatbots and digital assistants as well. I can surely say we are not there yet. I talked with chatbot Eugene when they claimed it passed the Turing test, although it didn’t. In my experience, even though I knew I was talking to a bot, it took time to catch it making mistakes and losing the image of being a human being.
Digital assistants in healthcare
The need for digital personal assistants and chatbots in healthcare are obvious and not new at all. The first “operative” bot in the healthcare sphere dates back to 50 years ago. Eliza was created to mimic a Rogerian psychologist that is a therapist who asks questions to the patient simply by rearranging what the patient himself said. Intelligent conversation is only emulated, allowing the chatbot to have some success in those days, when AI was just something seen in the sci-fi novels.
Currently, there are already certain medical fields, where start-ups and companies are offering intelligent personal assistants. For the visually impaired, Horus, OrCam, BeMyEyes and Aira all offer their solutions in order to create the possibility to live a more independent life. They are using various algorithms for describing the environment to the user, read out text, recognize faces and objects such as supermarket products or bank notes, or notify about obstacles. These algorithms are all able to learn over time. And that’s just the beginning.
Health chatbots are shaping up
The company, Kore.ai offers smart bots for healthcare facilities. The digital assistant can connect patients to the right contacts directly, give appointment details or make any changes. It lets the patients easily refill prescriptions or pay bills. It delivers lab, test or procedure outcomes or recommended next steps. Safedrugbot embodies a chat messaging service that offers assistant-like support to health professionals, doctors who need appropriate information about the use of drugs during breastfeeding. Izzy helps women track their period and serves as a birth control pill reminder.
Bots like HealthTap or Your.Md aim to help patients find a solution to the most common symptoms through AI. However, a chatbot never replaces an experienced doctor. The bot itself exhorts the user to book an appointment with a doctor for a diagnosis, and eventually for the prescription of a therapy. Florence might be a very useful chatbot for older patients. It is basically a “personal nurse,” and “she” can remind patients to take their pills. You just write the name of the medicine in chat, the number of times a day you must take it and at what time. Then, Florence sends you a message in chat every time you must take the pill.
Some institutions already recognize the potential in chatbots for patients and their services alike. The UK’s National Health Service (NHS) will start to use a chatbot app for dispensing medical advice for a trial period in 2017, with the aim of reducing the burden on its 111 non-emergency helpline. The NHS is developing the app with Babylon Health, one of the new breed of paid, doctor on-demand services.
The way forward
Healthcare chatbots will spread around in many areas of healthcare. In February 2017, Merck announced it will sponsor an open innovation challenge in conjunction with Amazon Web Services and New York-based innovation consultancy Luminary Labs. The goal: a call to entrepreneurs, techies, and industry types to create open solutions for these voice-tech tools to help people with chronic conditions. The first one is diabetes: the so-called Alexa Diabetes Challenge launches in April, and will concentrate on type 2 diabetes.
As it is highlighted in the above examples, digital health assistants in the form of chatbots are already able to answer basic questions about health management. In the future, this function will get even better as it will be constantly evaluated and it will learn from its own mistakes. Moreover, not only one chatbot will learn from the given mistake, but the entire fleet of chatbots in that given service. As these intelligent personal assistants process new information very fast, it will follow-up and learn when to direct the patient to a physician’s attention or call for help itself.
In the future, we will be able to share the data we measure about our health and the data in our medical records with these digital personal assistants, so it can help make the wisest decisions. Maybe, they will also be able to detect certain medical conditions based only on our voices. There is already extensive research going on concerning vocal biomarkers and their use in healthcare. And I’m sure software developers and tech companies will recognize the huge potential for the use of gamification in healthcare chatbots as well. For example, if you could collect points for eating healthy food or doing some sports and then you could receive certain rewards, patients would be more motivated to live healthier.
Where are the boundaries of our privacy and our control over technology?
Healthcare chatbots and smart tech-assistants in general raise serious ethical questions about the nature of humanity as well as its connection to technology. How much should these artificial intelligence-supported platforms know about us? Who else could have access to our private and very personal health information? How could we defend ourselves in this vulnerable situation? As it is often stressed, any device can be hacked, and I certainly do not want unauthorized people to have access to my data. How can we safeguard the inconceivably huge amount of data we generate through smart devices such as chatbots in the future?
The other area of concern is where and how to find the power balance between technology and humanity. Who will have the control? Can we ensure that artificial intelligence remains confined in our control mechanisms or do we let an uncontrollable ghost out? These are very difficult and extremely relevant questions, which need careful consideration from decision-makers, powerful tech companies, researchers as well as average people on the street. As technological development will not slow down, we have to catch up with the innovation happening around us and give reasonable responses in order to ensure our humanity.
So, bots are coming even if you are reluctant to embrace them. And the first time you have a conversation with a bot that is wise, humorous and learns your preferences in seconds, you will be amazed how hard it is to resent them.