The Future of Bots: Will They Ever Be Prescribed for Healing?

Be sure to check out part 1 and part 2 of this 3-part series.

Imagine this: you’ve had a long day at work and you want to vent and decompress. So you get online with a chatbot, that gives you actionable advice and practical solutions. Think prototype ELIZA bot, but totally refined, redefined, and updated. No, really — it’ll even send an empathetic emoji or two your way.

Though chatbots are still far from this reality (mainly because there’s little scientific evidence), it still begs the questions about future chatbots:

1. Can a chatbot be therapeutic?
2. And can a provider prescribe a chatbot as a way to help patients in their treatment plan?

With healthcare always changing, a chatbot’s role will likely be changing with it (we already discussed how they’re effectively used today both in everyday life and in healthcare).

For this last article of our series, we have two case studies to unpackage how can chatbots may (or may not) actually have therapeutic benefit.

Case Study #1: Woebot

Woebot is the first to cross the line from being a cool chatbot project with a hypothesis to one with a proven health impact. Here’s what it looks like:

Meet Woebot. Isn’t he more approachable — and dare I say, cuter — than ELIZA?

Woebot is a chatbot that is available on Facebook to help patients with depression to reinforce and automate cognitive behavioral therapy.

Patients can talk with Woebot at any hour, and it answers questions and provides material to help patients based on their mood.

Woebot’s creators even completed a study to see whether it was effective at helping patients — and to much surprise, it actually was. It’s pretty promising to see the bot demonstrate benefits.

HelloJoy is mental health chatbot similar to Woebot, but its effects haven’t been validated yet.

HelloJoy’s messaging. On point.

While Woebot has a strong case for success, most providers still have mixed opinions about letting it take complete control of the patient’s entire therapy regime, thus calling for a hybrid approach.

The hybrid approach could look something like this:

-The provider allows the chatbot to take over parts of the conversation a provider is comfortable automating.

-The provider still has the flexibility to jump in when needed — like if the bot feels the conversation warrants attention.

In fact, some tech enthusiasts see this approach as a viable solution to the growing shortage of health care providers.

Why does this work well?

See, medicine is driven in protocol. If a patient meets certain conditions, then the doctor gives an instruction to direct the patient’s care. Like giving aspirin after a heart attack.

Actions like giving instructions are pretty easy to automate.

It may make more sense for a providers to task chatbots with that.

Case Study #2: Ava

We can walk through another scenario where a chatbot could have therapeutic value.

Let’s say you’re an OB and you have a triage nurse who manages your incoming calls, mainly of which are pregnant patients who are close to (or already in) labor. But asking pregnant patients to call to when they think they’re in labor is challenging for patients and triage nurses alike.

Pregnant patients are typically experiencing higher than normal levels of stress, and a worried patient may have to call in and explain herself each time to a new triage nurse because of simple scheduling logistics.

This can be a pretty inefficient experience for both parties.

Enter Ava, the pregnancy chatbot.

Patients have 24-hour access to Ava.

When a patient is having contractions, she can tell Ava. Ava asks how long they’ve been going on, how far apart they are, how long they last, among other things to gauge the urgency of the situation..

Ava can then do first-level triaging tell the patient whether she should stay home or not.

The patient’s provider could also teach Ava to reach out to to the patients 6 hours after the initial interaction to see how the patient’s doing. The provider can also watch the entire interaction happen at any point in time.

Let’s say twelve more hours go by, and Ava receives another message from the patient.

This time, Ava knows the patient had contractions earlier and asks only relevant questions based on previous answers (i.e. “Have your contractions gotten closer together?” and “Did your water break?”)

If Ava thinks the patient is nearing active labor based on what the practitioner teaches her, Ava can give the patient care and delivery prep-instructions while also notifying the provider simultaneously.

Why does Ava work in this case? Because patients want immediate answers and streamlined communication that requires little effort. While doctors can monitor their patients status in real-time.

So, About Those Chatbots…

…Aren’t they cool?

Here’s a recap of what we learned in this series:

• What chatbots are
• What chatbots do in healthcare
• What therapeutic value chatbots have

Now, what do you think? Are chatbots the next big thing in healthcare? Do you think they can help patients get better? Or should they only be used as appointment reminders?

Do you think chatbots are taking over the world?

Tell us below.

And if you liked this series, leave this post a big ♡.

Written by Tashfeen Ekram, MD.
Tashfeen is a radiologist, self-taught coder, healthcare innovator and Co-Founder of Luma Health. Contact him on Twitter at @tashfeenekramMD.

Sources:
Business Insider, Vela, Joy, X.ai, TechCrunch