No appointment necessary — The rise of bots in healthcare
Written by Sarah Watters
Innumerable health apps and wearable devices have flooded the digital health space in recent years. Recently, however, app downloads have been steadily decreasing. Applications from ecommerce to productivity to politics to news and entertainment are sprinting toward the latest “conversation UI”.
Technology behemoths like Facebook, Amazon, Microsoft, and Apple, have shifted towards a new interaction model, using chatbots. Chatbots provide an “app inside an app” experience, allowing discussions to happen within the context where they are most relevant — such as on a website. Forecasts by Juniper Research predict that by 2022, chatbots will save organizations $8 billion worldwide by taking over certain, particularly customer service-related, tasks, and improving experiences and outcomes.
Health bot. Medical chatbot. Chatbots for patients. Chatbots for health.
Health care appears to moving in the same direction as other industries, that is, from apps to chatbots. However, using a robotic chat agent to interact with patients is not a new idea. In fact, the world’s very first chatbot was developed over half a century ago. ELIZA, was designed to be a Rogerian psychotherapist who could chat with patients by reflecting on what the patient said, using predetermined rules and scripts to formulate a response.
Timely, accurate information.
Health care is different from other industries because when concerns or questions arise, sometimes people Just. Can’t. Wait.
Patients and caregivers are often left concerned, confused and clueless since they can’t get (accurate) answers to these questions in a timely manner, often beyond the walls of a doctor’s office. While some people do Google searches to find answers, it can be challenging to know if the information found online is accurate and relevant to the particular patient. Importantly, expanded primary care practices are also not an ideal alternative since patients’ lives shouldn’t be disrupted (e.g. having to leave work or kids) by having to go to a doctor’s office just to get few second response to a simple question.
Further, the World Health Organization cites a shortage of more than 7 million health care professionals worldwide. This makes it increasingly necessary for us to supplement traditional care models with technology in order to allow doctors to focus on the most critical issues and patients.
Chatbots offer a cost-effective solution to the difficult problem of decrease resources and increasing costs. According to Juniper Research, health care chatbots are expected to generate an average time savings of approximately four minutes per inquiry, yielding an average cost savings of roughly $0.50-$0.70 per interaction.
However, for patients, caregivers, and health care providers, the question is:
Are chatbots effective enough to efficiently engage and aid patients?
Use Cases
A flurry of use cases — many of which center around customer (i.e. patient) service activities — have the potential to be overtaken by chatbots. Manual collection of data (e.g. medical history or current medications), for example, which is typically done over the phone or in the doctor’s office prior, can be done automatically via chatbot. This means that patients won’t be stuck on hold or spend longer than they need to at their appointments.
Through a simple chat window, chatbots provide patients opportunity to:
- Find a doctor and make an appointment (e.g. Kore.ai)
- Chatbots can triage patients based on symptoms and symptom severity
E.g. help determine if a patient’s condition needs escalation to a nurse, physician, emergency room visit, or specialty practice. - Be monitored for improvements/deterioration in their health (e.g. GYANT, MedWhat, HealthTap)
Is the medication helping? Are there any side effects? Are symptoms becoming less severe? - Receive assistance with procedures recommended by their doctor (e.g. Sense.ly)
E.g. follow up on adherence to post-op care and address any concerns
What should the patient eat or avoid? What symptoms should they expect to experience? - Determine the dosage of a particular medicine and get reminders on when it is time to take a medicine (e.g. Florence Bot, Izzy)
Can I drive after taking a particular medicine? Can I take an over the counter pain medication for my headache while I am on this prescription? - Establish a repository of their historical health data (e.g. Fitmeal, doc.ai, Forksy )
What were my blood-sugar levels in the past week? Where can I get my wheelchair serviced? - Get answers to their questions about medical equipments, lab procedures etc. (e.g. Luma Health)
For patients living in remote areas, where access to health care is limited, mobile health and the ability to get information efficiently, as with chatbots, can make a marked difference in the quality of their lives. Chatbots may also prove to be especially helpful in chronic disease management. Currently, evaluations are being conducted with Type 2 Diabetes patients and for people who have mental health conditions.
As users and developers get more experience working with chatbots in health care, the roles that chatbots play are likely to evolve from handling easier to, eventually, more complex tasks.
Digital Trust
There is a lot of discussion around the nature of the interaction with a chatbot.
Can you tell that it’s not a human on the other side of the conversation?
Can you take comfort from the advice and information provided to you by a computer?
Providing trusted health information can be an intimidating task regardless of who is on the receiving end. Thomas Schulz, health care expert and founder of botscamp.co, expects that the weird feeling that we may have speaking to a chatbot will eventually dissipate as we adjust our behavior, as with most other new technologies to which we’ve adapted. It’s also possible that some people are predisposed to feeling more at ease talking to someone when they’re not face-to-face, and knowing there isn’t someone truly ‘listening’ on the other end is relieving rather than odd.
Schulz refers to the concept of digital trust, referring to our tendency to adopt and adapt to new technologies as was the case with SMS messaging, or even earlier, speaking by telephone. Over time, we’ll develop a comfort with chatbots. Schulz notes that evidence based studies are just beginning to surface now, in 2017, and in 2018 we’re likely to see greater user acceptance.
Building Rapport
Striking the delicate balance between engaging the user and acquiring the necessary information is perhaps the biggest challenge for chatbots. With less than a minute amount of time to grab their attention, chatbots need to be responsive and intuitive if they’re going to be adopted by the masses.
Natural language processing (NLP) tries to understand what a user is asking about, gradually learning and building up a profile of the user, their questions, moods, and habits, among other information.
When a user and chatbot first interact, the chatbot tries to identify the intent of a the comment or question with some degree of confidence. Then, depending on the confidence level, the chatbot will either ask a follow-up or address the question based on a taxonomy that links the intent of the question with the desired response.
More sophisticated chatbots apply deep learning to help the chatbot improve its responses to subsequent interactions. Importantly, to convey a human element, sentiment analysis is used to detect the user’s emotions, such as frustration.
Our expectations for technology, in general, are to provide us with information and connect us to others. In health care, we need technology to go one step further. Health care technology needs to be trustworthy and empathetic, and able to provide us hope and comfort when we are feeling unwell or concerned.
Convo Health
Last week, Cossette Health launched a new conversation platform which allows patients and health care professionals to communicate asynchronously. In addition, the conversational-based interface provides patients or health care professionals with informational resources (via text, images, or video), notifications, adherence/compliance capabilities, and feedback opportunities. Content can be organized and scheduled through an integratable content management system and environment-specific experiences can be developed through the platform.
Designed for health care, Convo Health’s features are tailored towards enabling improved patient experiences and outcomes. It’s messaging capabilities offer value across many different environments — from mental health services to new immigrants or First Nations, or support for the care providers themselves.
From an emotional (and sometimes financial) standpoint, health and health care, can be a high-stress area. Convo Health offers patients, their caregivers, family, and friends, a valuable resource by enabling greater access and ability to access information 24/7.
The platform is available in a mobile app, through SMS/iMessage, and via Facebook Messenger and its standardized interface is supported by all major browsers (including Internet Explorer). User data and information stored within the app is secure and adheres to the latest PIPEDA & HIPAA security standards. HL7 and FHIR API allows for the platform’s integration into existing EHR/HIE systems.
Looking forward
By now, most of us are comfortable asking our Alexa’s and Siri’s about the weather or to dim the lights. Medical chatbots have the potential to drive quality of care while at the same time increasing efficiency in delivery and it may be just a matter of time before we have the same degree of openness when speaking a medical chatbot. While it is still is early in the game, experts anticipate that widespread satisfaction and success with AI chatbots in health care may come sooner than we think.
