Chatbots for healthcare: AI assistants to the rescue

Chatbots are the simplest pathway for any provider group or healthcare business entity to introduce a new technology into their business processes.

By Steven Stumpf, EdD and Lizette Alvarez, MBA

Chatbots and Artificial Intelligence (AI) can seem complicated especially when trying to figure out how these tools can be applied in healthcare. Nevertheless, chatbots are the simplest pathway for any provider group or healthcare business entity to introduce the new technology into their business processes.

What are chatbots?

Anytime a customer engages with the “person standing by” on a website, and the customer is told to click on the “word balloon” in the corner (to escape wait time) it is highly likely the customer is going to talk to a chatbot. The chatbot is programmed to answer the most common questions for which the customer is seeking an answer. Chatbots replace the “live” customer service agent of past decades with an algorithm; a programme that can be quite simple or quite sophisticated.

“Having one person call one person takes a lot of human capital,” says Dr. Z. Brown of Northwell Health Solutions.

Apply chatbots to any frontline patient interaction

The applications to which chatbots are being assigned is growing and will continue to grow. Here are a few of the popular ideas for deploying chatbots:

1. Virtual assistants handle simple, repetitive tasks.

2. Chatbots interact with the patient/consumer using keystrokes (from laptop or phone) or voice.

3. Chatbots can address each inquiry as a new or familiar inquirers (patients).

4. Chatbots work with patients, healthcare centres, insurers and providers.

5. Programming can be fairly simple if one uses a platform based upon natural language processing (NLP).

6. Chatbot are not apps although a bot can point the inquiring party to an app.

7. Chatbots collect information the user and the provider want collected.

Chatbots are one component of my Healthcare Information Technology class. Chatbots are well known to my millennial students who prefer a passive interaction. Accessing information need not have a living breathing person on the other end. When the other end is occupied by an algorithm – like a chatbot – that is even better! A decade or longer of texting and social media has closed the era of “live” conversation. Besides, chatbots can simulate conversation faster than you can say “I have a question.”

Building a chatbot is like planning an important conversation. Suppose you are going to meet with a vendor that wants to talk about fees or referrals or anything else related to business. The information is important. The encounter must be engaging, warm and polite. You want to plan ahead. Identify topics you can expect to come up. Figure how you will respond.

Apply this to a patient encounter. The steps that have preceded every patient encounter for generations are completely predictable. Most can be handled with a chatbot instead of a human.

The chatbot approach covers the same tasks. The headers – intent, action and information - are terms that define steps in building a chatbot.

How to think like a chatbot

The INTENT is the predictable conversation, which the patient will initiate. The ACTION completes the predictable activities that recede and follow the patient visit. The INFORMATION is what gets collected, which describes, summarises and logs the patient visit.

What does every patient want (INTENT)? To easily arrange meetings. Less interaction with the office is better than more. The inquiries which typify many kinds of patient behaviour are dictated by custom. These are predictable encounters.

What does the front desk want (ACTION)? Let the chatbot respond to the patient intention. The chatbot can look up dates and find open time slots. It can confirm to the patient if there have been any changes in employment or insurance coverage. If there has, the bot can fetch this after the phone call and before the appointment; easier when the employer and insurer has AI capabilities. Otherwise the medical assistant will have to make a phone call!

When the medical visit is completed, how are the records populated (INFORMATION)? Information summaries are logged in the EHR. Data can also be texted to the patient’s wearable device or to an email.

“We’re surrounded by work that is being done over and over by humans that’s not necessarily the best use of their time,” says Greg Johnson, CEO LifeLink.

Can chatbots be warm and fuzzy?

A chatbot can be quite personable. It is all up to the programmer. Warm conversation is also predictable; even seasonal. Greetings change with generations. You may have noticed the standard “openers” or “closers” for commonplace conversations.

Popular warm greetings or sign offs include “I hope you are having a great” or “I hope all is well with you.” If you are going on a trip for business, you will almost receive the warm “safe travels.” Whatever happened to bon voyage?

These friendly greetings are easily programmed in a chatbot. We want the chatbot to be familiar, certainly polite, and even nice. Once we get used to engaging with a chatbot we also want it to do more for us! Looking for chatbot is a lot like shopping for any desirable product; shoes, briefcase, vehicle. Relying on chatbots can quickly become habit forming.

A local burger and fried chicken stop in Denver, U.S. implemented a chatbot this month to handle the drive-thru window. “The system takes a lot of friction out of interactions between customers and employees,” the owner said, noting that the AI was designed to sound like an amiable woman’s voice.

“The AI never gets offended and it will just keep talking to you in a very calm and friendly voice.” Employees that are commonly assigned these repetitive tasks may also appreciate the relief from the routine.

Can a Virtual Assistant (VA) become the perfect employee?

Conversational interaction distinguishes this interactive bot from the earliest “fetch-bots” Siri and Alexa and Google Assistant. The VA handles predictable repetitive questions. The VA can learn from the different ways in which a question is asked. The more the VA chatbot gets a question in a different form the better the bot can become at answering the question. The VA saves time and increases practice efficiency. And time is money.

A chatbot is an algorithm (commands) that can record data on the fly and make a searchable record. The record includes searchable indicators, e.g., timestamp, contact’s address/phone, call length, etc.What exactly are chatbots? Algorithms make up chatbots. Algorithms are programmes. The greater the number of programmes/algorithms the more complex the chatbot. At the least, chatbots should replace the most simplistic time-consuming redundant tasks.

Also, consider that hardware speed, frequency of use (beyond comprehension), and the growing preference for “more passive, text-based communication” places chatbots in position to replace time intensive, active communication when called upon to respond! Better to respond when we feel like it; when it is more convenient to the person being prevailed upon.

Chatbots leverage several basic laws of computing that guarantee their proliferation. Computing power and speed is close to becoming limitless. Moore’s Law held that CPU speed would double every two years. Once CPUs are the size of atoms speed will become irrelevant. All speeds will be immediate. Other considerations are the concepts or machine learning and deep learning. The machine learns on its own. This ability makes the chatbot astute. Independence is in the future. Alexa, Siri and Google Assistant learn from frequency of use. Each derives from a database that grows with each interaction. Speed and algorithms are part of the engines driving the AI Assistants.

Ready to welcome the new employee?

Integrating a chatbot into a new or current business model means the “programme” will function as a new employee; rapidly co-existing amongst other technologies is becoming commonplace in healthcare; such as mobile apps, remote clinical devices that collect and transmit digital health data, secure clouds and more. Chatbots are attractive to small businesses and the largest powerhouses around the world. Healthcare can deploy the chatbot within a half hour of time and receive the endless benefits other sectors have rapidly seen. Adopt one? It is too simple not to.

Innovative chatbot platform launched in UAE

Medcare, a UAE-based hospital group, recently announced the launch of its new “virtual health assistant” chatbots. The two new health assistants use advanced artificial intelligence (AI) algorithms to learn about the patient in order to personalise responses and give accurate information in real-time.

The health assistants are reportedly the only chatbots in the region that allow the patient to manage all aspects of their appointments online including booking, rescheduling and cancelling, permitted by real-time back-end database integration.

A distinctive feature of the chatbot technology is that it has been humanised to make the health assistants intuitive, patient-centric and patient-friendly. Patients will be able to “talk” to the health assistants and use this new platform throughout their entire healthcare journey. They will also be able to see lab reports online, check insurance coverage and locate the nearest Medcare facility in order to receive timely treatment.

 

RELATED FEATURES

Care Model Innovation and Technology Adoption: The Next Investment Big Bets for Health Systems in GCC

Inspiring Innovation in the Spotlight at Arab Health 2019

Innovations in Healthcare: Challenges and Role of Digital Transformation