Learn how an AI-powered medication assistant can increase the reach of healthcare professionals, drive patient engagement, and increase medication adherence in populations.
Can a chatbot help people be healthier?
Studies show that humans prefer to talk to a person about healthcare issues.[i] So, next time someone tells you robots are soon to replace your doctor it’s probably not true. However, more people are warming up to the idea of letting technology assist them in their quest for healthier, automated lives. At EveryDose, we’ve deployed our Conversational AI bot, Maxwell, to assist thousands of patients on their journey to a healthier life. We wanted to share some of the insights we’ve gained from using Conversational AI in healthcare.
Artificial intelligence still conjures up a bit of fear in some people. But the reality is that this technology has already found a place in many of our daily routines. People use Siri, Alexa, or Google Home to ask about the weather, they use Google Duplex to handle dinner reservations, and they use chatbots for customer service questions on websites. All of these instances are examples of Conversational AI in daily life.[ii] As Conversational AI use grows, it’s important to know that not all chatbots are created equal. Some bots make life easier. Others might be considered annoying.[iii]
Let’s Talk About Conversational AI
The goal of this article is to outline how Conversational AI can increase medication adherence, but we also wanted to highlight general ideas about Conversational AI that can be applied to any industry. To start, let’s break down what “Conversational AI” means.
The major purpose of Conversational AI is to create personalized engagement at scale. The “conversation” aspect is key to accomplishing this. The ability to command language is what sets humans apart from other species. If you can read this sentence, then you should have no learning curve when it comes to Conversational AI. Here’s an example to explain what I mean. I’m an iPhone user, so if I picked up an Android phone it might take me some time to understand how to operate the new interface. Android phones have a different user experience, different buttons, apps, etc. than iPhones. However, there is almost no learning curve in switching between different Conversational AI platforms, because the medium is conversation. As long as I can speak to the voice-enabled device or type words into a chatbot, then I should be able to master any Conversational AI platform in no time. This is a huge advantage for certain demographics, like seniors, who did not grow up in the technology age. (Seniors might be more tech-savvy than I’m giving them credit for. It’s worth pointing out that senior smartphone ownership has been growing at a rapid pace).[iv]
The ability to use conversation at scale is extremely helpful for customer service. While some of us introverts may avoid talking to people when possible, most people still want to speak to “someone” when they encounter a problem. It’s usually quicker for a customer to talk to a “person” than trying to solve a problem on their own or reading an entire FAQ page. When it comes to certain industries, like healthcare, there are even more intricacies that require personalized help. For example, people typically have unique health problems, diagnoses, or medical issues. Thus, a one-size-fits-all approach (like an FAQ page) might not be effective. This is especially true when it comes to managing different types of diseases or combinations of prescriptions. Getting something wrong could have drastic consequences. Plus, we’ve all heard the joke that Web MD always leads you to a page suggesting that you are doomed. Hence, why trying to research your own medical issues can be problematic.
This is where Conversational AI comes in. Conversational AI can learn about what types of problems an individual faces through data and (you guessed it) conversation. It engages customers in a personalized way while also drawing data from all the other customer conversations that have occurred in the past. Like a human being, every time an AI bot works on a problem it gets smarter and gets better at problem-solving. It also learns about you as an individual and asks you unique questions based on your circumstances. Unlike a human, you can upload knowledge into the AI via data, so it can learn faster and retain more information than the smartest Homosapien. This is often referred to as training the AI algorithms. This requires a lot of data. (I won’t get into how Elon Musk wants to upload data to human brains).[v]
If you want to learn more about Conversational AI and how it works, you might enjoy this article from Amazon around their Alexa platform.[vi]
When and How Conversational AI Should be Used
Google the phrase “am I talking to a bot” and you’ll see a mountain of results. This indicates that people don’t like being tricked into thinking they are talking to a human when they are not. A lot of this stems from the nefarious bots that have been actively pushing information to social media users, disguised as fellow human beings.
EveryDose believes it’s important that people know when they are talking to a bot. That’s why our bot is represented by an illustrated mascot and we make it clear that his white lab coat does not make him a medical professional. (There’s a reason we didn’t name him “Dr.” Maxwell).
How much conversation is too much?
As mentioned previously, you could use Conversational AI for anything, but it doesn’t mean you should. I had an old boss who had a rule about emails saying, “if it’s longer than a few paragraphs, pick up the phone.” This can apply to conversational AI as well. If the customer is typing in too much information at a time, then they might be better served by picking up a phone. If it’s a voice-activated bot, it may be best used for asking a question or giving a command. Both are often only a few words. “Hey, Siri. Where is the closest pharmacy?”
Conversational AI can also be great at directing people to the right resources. In our own conversational AI platform, we like to point people to vetted, fact-based medical information when questions get more complex. We don’t intend for our chatbot to answer every question on its own. Part of what makes Maxwell great is that he is trained to find the right resources for patients. As technology continues to improve, and bots get better at mimicking human conversation, you will see higher usage of conversational AI. Oh, and like Siri, Maxwell can also direct you to the nearest pharmacy.
Replacing Humans?
Conversational AI may make a human’s job easier, but we aren’t ready to believe that robots are going to take over our jobs. However, they are a great resource when no humans are available. For example, a chatbot can be deployed 24/7. This means they can help customers at times when a human might not otherwise be available. Chatbots can also fill in the gaps when a company doesn’t have enough bandwidth for customer engagement. We are a great example. Patients can always ask their doctors questions when they are in the point-of-care or when the doctor is available by phone. However, patients often have simple questions about their medications at times when their doctor is not available. Our Conversational AI medication assistant steps in and helps doctors answer patient questions about specific medications. Maxwell even provides tailored information to help improve patient health literacy based on a patient’s specific medical conditions. Of course, a chatbot is no replacement for a doctor and is still limited as to what questions it can (and should) answer. The key here is that the chatbot should extend the reach of an employee, not replace them.
AI Invasion?
Many AI chatbot companies will tell you that chatbots increase sales and customer engagement. While this can be a reason to deploy conversational AI at a company, it can certainly backfire. An in-your-face chatbot can feel like a pushy salesperson, and not everyone wants a bot that feels like it’s invasive. Companies should offer a chatbot to their consumers when they want to use it, but it shouldn’t feel like the chatbot is forcing you to talk to it (no one wants a needy chatbot). More importantly, a chatbot is only going to be as good as the data you feed it. No different than a person, it needs to be taught what to do and say.
Using Conversational AI to Increase Medication Adherence
If you aren’t familiar with medication adherence, you should know that the World Health Organization estimates that 50% of people don’t adhere to the proper medication regimen.[vii] This is particularly important in patients with chronic diseases. Non-adherence significantly increases hospital visits every year and is a huge burden on healthcare spending.
Medication regimens can be confusing for patients and they may have simple questions about things like side effects or interactions. Every patient also has unique obstacles to adherence (i.e. reasons why they might not be taking their medication properly). Until these obstacles are determined it’s nearly impossible to initiate an effective intervention strategy for that patient.
Here at EveryDose, we decided that every patient deserves personalized interactions. The only way to achieve this was by using technology to scale our patient reach. That’s why we created the first Conversational AI ChatBot for Healthcare.
Our chatbot talks to patients, asks them questions, encourages them, alerts them to potential interactions regarding their medications, and, most importantly, he determines the unique obstacles or risks that might keep an individual from achieving medication adherence.
Once he identifies these obstacles he routes this information to the proper care team or healthcare provider. He also provides tailored interventions for the patient. The care team can also access all of this information in real-time using a web portal. This helps ensure that they reach out to the highest risk patients first and use person-to-person communications effectively. This part is important. Maxwell may be able to talk to a patient, but he doesn’t replace humans. He just makes their job easier!
Along with alerting the proper people of the potential adherence risks, Maxwell also actively works to educate patients on why their medications are important and how they can combat their struggles with adherence. Patients who receive educational components within their intervention strategies show increased adherence rates by up to 92%.[viii]
We hope that technology can amplify the human touch within healthcare.
Want to learn more about Conversational AI and Medication Adherence? Don’t hesitate to get in touch with us.
Download Our Free Medication App
[i] “Your Customers Still Want to Talk to a Human Being.” Harvard Business Review, 19 Dec. 2017, hbr.org/2017/07/your-customers-still-want-to-talk-to-a-human-being.
[ii] Velazco, Chris. “Google’s Reservation-Making AI Will Be Making Calls Soon.” Engadget, 13 May 2021, www.engadget.com/2018-06-27-google-duplex-assistant-public-testing.html.
[iii] Cantillon, Sarah. “A Chatbot Could Damage Your Brand.” The Drum, The Drum, 22 Feb. 2019, www.thedrum.com/opinion/2019/02/22/chatbot-could-damage-your-brand.
[iv] “Demographics of Mobile Device Ownership and Adoption in the United States.” Pew Research Center: Internet, Science & Tech, Pew Research Center, 23 Nov. 2021, www.pewresearch.org/internet/fact-sheet/mobile/.
[v] Metz, Rachel. “Elon Musk Hopes to Put a Computer Chip in Your Brain. Who Wants One? | CNN Business.” CNN, Cable News Network, 21 July 2019, www.cnn.com/2019/07/20/tech/elon-musk-neuralink-brain-chip-experts/index.html.
[vi] “What Is Conversational AI?” Amazon, developer.amazon.com/en-US/alexa/alexa-skills-kit/conversational-ai.
[vii] Mathes, Tim, et al. “50% Adherence of Patients Suffering Chronic Conditions–Where Is the Evidence?” German Medical Science : GMS e-Journal, German Medical Science GMS Publishing House, 2012, www.ncbi.nlm.nih.gov/pmc/articles/PMC3525884/.
[viii] “Evaluation of the Efficacy and Cost Effectiveness of Health Education Methods to Increase Medication Adherence among Adults with Asthma.” American Journal of Public Health, ajph.aphapublications.org/doi/10.2105/AJPH.80.12.1519.
