Generative AI Empowers The Empathetic Refutational Interview (ERI) Technique For Medical Students And Doctors

In today’s column, I am continuing my ongoing series about the impact of generative AI in the health and medical realm. The focus this time is once again on the role and importance of empathy in the practice of medicine and healthcare. I will be sharing with you a newly devised technique that leans into an empathetic process to overcome or at least ameliorate medical misconceptions or myths that patients might hold.

Avid readers likely recall that I had previously undertaken an in-depth look at the use of generative AI to aid in training medical students to embody empathetic clinical skills, see my noteworthy discussion at the link here.

In addition, I’ve covered many other facets of how generative AI and medical students and doctors are gradually and inexorably going to be making use of generative AI. This ranges from the seemingly low-hanging fruit of the summarization of medical notes (see the link here) to the substantial worries that AI might convince physicians to fall for clinical false positives or false negatives, see the link here. I have also boldly predicted significant changes to the classic patient-client relationship, which I refer to as the emerging patient-AI-client relationship, a new style triad explained at the link here.

Back to the emphasis on empathy in the medical profession.

A concerted belief is that budding doctors and likewise seasoned physicians ought to be honed toward being empathetic with their patients. Medical schools have taken up this mantle and sought to train medical students on the precepts of empathy in hopes of launching them into a career steeped in an embodiment of empathy. A wide range of training methods have been used, including group discussions, individual readings, self-reflection, in-person training exercises, online computer-based simulations such as the use of VR (virtual reality), training films, and the like.

I have stridently advocated that generative AI can and should similarly be in the mix of ways to bolster empathy in all facets of the medical profession, including the ongoing daily work of healthcare professionals (HCP).

The latest twist in this coverage is that I will introduce to you this time a recently formulated technique referred to as the Empathetic Refutational Interview (ERI). Here’s what I am going to discuss. First, I will provide some overarching background about why such a technique is needed and indubitably useful. Second, I will dive into the crux of the ERI technique and provide empirical research underlying the approach. Third, the technique as currently constituted has no bearing on generative AI, so I’ve chosen to think outside the box and connected the approach to a form of medical student and medical doctor adoption that leverages contemporary generative AI.

Get yourself ready for an exciting journey into empathy and the ever-evolving fields of medicine and health.

What To Do About What Patients Misbelieve

I dare say that doctors and all manner of healthcare professionals encounter daily a slew of false beliefs held by their patients.

This seems to be a rising tide. Perhaps the widespread use of social media contributes to the exasperating acceptance of medical misconceptions and myths. There is also the ease of doing online Internet searches. The odds are that you can readily land on seemingly impressive-looking websites that proceed to tout the most zany of medical concoctions and falsehoods.

The trouble of course is that a patient is bound to cling to a medical misbelief. When a proper medical diagnosis and recommended treatment are articulated to a patient, they are spurred these days to fervently argue with the medical merits thereof. Their “proof” of disagreement is potentially based on a vague haziness that is fueled by things they’ve heard or seen online, and for which marginal if any true medical science sits at the core of said pronouncements.

What is a doctor or healthcare professional to do in the throes of such dogmatic doggedness?

One instinctive response would be to straightaway explain the medical basis for the recommendations. Sometimes this works. A lot of the time it does not. A patient might not be in the mood for a stern lecture. Indeed, it is conceivable that the recitation of medical facts and figures will ironically harden the oppositional viewpoint of the patient. They might perceive that you are trying to bowl them over with a lot of fancy-sounding medical jargon. It seems like a trick and a classical ploy of smarmy medical underhandness.

Another way to respond would be to sharply tell the patient to take it or leave it. You’ve done your due diligence and it is up to the patient to take the next step. A horse brought to water cannot be made to drink the elixir. Of course, this is a sadly short-shrift of medical ethical practices in the sense that you ought to at least make a sincere attempt to get your patient to understand and hopefully proceed on a medically beneficial pathway. Giving up at the get-go doesn’t seem professionally befitting.

We seem to have run out of ways to proceed.

No, wait for a second, there is an additional path that can be considered. One potential icebreaker and motivator would be to exercise empathetic skills. A doctor or healthcare professional so skilled might be able to create a conversational atmosphere that could overturn a misbelief held by a patient. If not fully overturning the misbelief, the aim is to soften the otherwise steeled defensive posture and open the door to the patient giving sufficient attention to the medical recommendation.

I will in a moment depict a technique known as the Empathetic Refutational Interview that establishes four key steps for a doctor or healthcare professional to embrace and utilize in these settings.

First, let’s put a bit more on the table.

The range of potential medical misbelief is quite large. You won’t necessarily need to employ a full-on outstretched empathetic discourse in all circumstances. In recent times, the biggest and most enduring doctor-patient debate often involves vaccinations and vaccines. The topic has risen to nearly ubiquitous status as a result of the pandemic. Everyone nowadays has an opinion on vaccines.

Many other medical and healthcare topics fall into the common misbelief categories. They might not be quite as momentous as vaccines, nonetheless, they do arise and it is best to be prepared for conversing with patients about them. A patient might construe a seemingly mundane medical topic as all-important. Thus, if a doctor or healthcare professional treats the matter as inconsequential, they are potentially winning the battle but losing the war. They are prevailing on something small but possibly losing the long-term trust of the patient.

To give you a flavor of frequently noted medical or health-related misconceptions, here is a handy list of ten such medical myths as posted online at WebMD (Source: “Slideshow: 10 Health Myths Debunked”, Dr. Poonam Sachdev, WebMD, posted on March 02, 2024):

• (1) Drink 8 glasses of water a day
• (2) Eggs are bad for your heart
• (3) Antiperspirant causes breast cancer
• (4) Being cold gives you a cold
• (5) You need a daily multivitamin
• (6) Eat breakfast to lose weight
• (7) Green mucus means infection
• (8) Sugar makes kids hyper
• (9) A toilet seat can make you sick
• (10) Cracking joints cause arthritis

Those are each appealing in their own societal and culturally-accepted way. It is doubtful that a patient would necessarily argue tooth and nail about those misconceptions when presented with medical insight by a versed medical practitioner. The vaccine topic is much more likely to get fireworks going.

All of these regardless of priority or perceived importance do entail the potential inclusion of an empathetic process that can serve to mindfully guide a medical doctor or healthcare professional. In some instances, the process might be elongated and deeply employed, such as in the case of vaccine discussions, while in other instances such as toilet seats or green mucus, the process might be lighter and breezy.

The bottom line is that for those who say that taking an empathetic route is too time-consuming or exhausting, I say that you need to know how to deploy the tools and skills that you have. If you misuse a hammer or a screwdriver, you can find yourself taking excess time and energy to solve the simplest of construction chores.

Mindset Of Patients And Readiness For Change

I’d like to quickly get another vital point into this discourse and then we’ll be ready to drop into the fundamentals of the ERI technique.

Upon my many years of working with doctors and healthcare professionals, I am known for bringing up the famed Lewin’s model of change when talking about the doctor-patient relationship. Allow me a moment to elaborate.

In the 1940s, Kurt Lewin was a pioneer in applied psychology. His said to be most prominent and long-lasting contribution has been the suggestion that human change consists of three stages, namely unfreezing, moving, and then freezing.

Here is a snippet of his 1947 paper that is oft-cited in launching the three-stage model:

• “A successful change includes therefore three aspects: unfreezing (if necessary) the present level L1, moving to the new level L2, and freezing group life on the new level. Since any level is determined by a force field, permanency implies that the new force field is made relatively secure against change.” (Source: “Frontiers In Group Dynamics: Concept, Method And Reality In Social Science” by Kurt Lewin, Human Relations, 1947.

What does this have to do with doctors and patients?

I’m glad you asked.

Doctors and other healthcare professionals often skip the first stage and leap right into an attempt to change the mind of a patient. This fails to first figure out what the patient has in their mind. This also fails to prepare the patient for being amendable to the medical recommendation or change that is about to be presented to them. For my discussion about a vital and related topic known as the Theory of Mind, see my analysis at the link here.

Lewin emphasized that by and large, the better way to get people to accept change is to create an atmosphere in which change is possible. Start with an unfreezing. Don’t start with the change per se. Gently nudge the patient in the overall direction of change and the likely acceptance of the change is bound to be increased tremendously.

Okay, so begin a patient discussion by getting them ready for your medical recommendation. Sound out if the patient has any preexisting medical misbeliefs that might pertain to the recommendation. You are greasing the skids by doing so.

I would also emphasize that the third stage is also often poorly handled by medical doctors and healthcare professionals. They don’t seek to freeze things into place. Here’s what I mean. You manage to converse with the patient in the medical office and seemingly convey the nature of the recommendation. The patient then appears to accept the recommendation.

Shortly thereafter, once they have walked out to their car or gotten home, they begin to lose sight of the recommendation. It slips from their fingers. In their mind, the conviction or earnest belief in the recommendation is rapidly decaying. All in all, it can often lead to them not abiding by the recommendation, even though during the office visit they seemed committed to doing so.

Why does that happen?

The likely culprit is a lack of freezing in place of the recommendation as touted by the medical doctor or healthcare professional. They failed to identify whether the patient suitably grasped the recommendation. Perhaps they faltered in covering how to undertake the recommendation. These are all omissions associated with the third stage of Lewin’s model of change.

I’d like you to therefore keep in mind that when going over a medical recommendation in the case of a doctor-patient relationship, please harbor in your mind the three stages consisting of the need to do a suitable unfreezing, make the change or proffer the medical recommendation, and then do a freezing or establishment of a semi-permanent basis that will remain persistent for the patient and aid in driving them forward on the medically recommended path.

In short, if you skip the unfreezing, the chances are that the change or recommendation will be met with great resistance and you will be wasting time and energy, needlessly. If you skip the freezing, you might have a whisp of success there in the office, but it will be for not since the patient afterward won’t carry out the recommendation or do so in a half-baked marginally committed way.

My recap is this for a sensible and beneficial doctor-patient relationship rule of thumb:

• (1) Unfreeze the viewpoint of the patient. Find out what the patient believes pertaining to the matters you are about to discuss and plant the seeds of change that a recommendation is in the offing.
• (2) Make the change or proceed to discuss the medical recommendation: You are now ready to move the patient into the medical recommendation, this is the change afoot and can be a difficult matter for the patient to accept.
• (3) Freeze the viewpoint to aid the patient in proceeding. Discover what might block or inhibit the patient from proceeding with the medical recommendation, and try to aid in making the recommendation a solid and sincerely believed faith.

The beauty of the three stages is that they are easy-peasy to remember.

Amid all manner of arcane and complex medical deliberations that might otherwise be on the hurried and harried mind of a doctor, medical student, or healthcare professional, they can just keep three fingers at the ready to remind them of the need to unfreeze, change or recommend, and then unfreeze as a means of engaging with their patients.

Core Research That Showcases The Empathetic Refutational Interview Technique

We are ready for the grand reveal.

In a research paper entitled “The Empathetic Refutational Interview to Tackle Vaccine Misconceptions: Four Randomized Experiments” by Dawn Holford, Philipp Schmid, Angelo Fasce, and Stephan Lewandowsky, Health Psychology, March 4, 2024, these salient points were made (excerpts):

• “To effectively address misinformed beliefs about vaccination, communicators must go beyond simply refuting these and show an understanding of people’s underlying motivations for their beliefs.”
• “We developed a new approach toward such conversations, taking into account individuals’ possible psychological motivations and responding with empathy even while correcting their misconceptions.”
• “Our approach is based on past work in healthcare communication and misinformation correction and provides a guide for more productive vaccine conversations.”
• “We introduce and report early-stage testing of a novel, multicomponent intervention that can be used by healthcare professionals (HCPs) to address false or misleading antivaccination arguments while maintaining empathy for and understanding of people’s motivations to believe misinformation: the “Empathetic Refutational Interview” (ERI).”

The gist of the stated approach is that rather than refuting a patient’s medical misbelief in a head-to-head brass-knuckles way, you employ a dab of empathy. Sugar makes the medicine go down, as they say.

My modest qualm about the naming of the approach is that it can be a bit confusing at first glance. You see the word “empathy” and you begin to think about the dreamy aspects of being coyly mindful and caring. The next word, refutational, seems almost out of sorts. Our minds leap to the act of refusing or rebuffing something. The third word is “interview” which has a staid sense to it, unlike say a discussion or a comfortable discourse.

You might also note that the focus of their articulation of ERI is shaped around vaccines and vaccination misbeliefs. As I stated earlier, I see this in a larger context and aim to reapply the method in any situation involving medical misbeliefs or misconceptions. I will provide additional selected excerpts from the research and judiciously ask you to replace in your mind the references to vaccines with the more generalized notion of medical misbeliefs, thanks.

Here we now can get into the stepwise process of ERI that they defined:

• “The ERI is designed to guide a conversation in situations where patients express concern about being vaccinated.” (ibid).
• “The ERI proposes four steps in response: (a) elicit concerns; (b) affirm; (c) offer a tailored refutation; and (d) provide factual information.” (ibid).
• “Underlying the ERI is the principle of empathy, conceptualized in healthcare settings as a response that communicates one’s understanding of the patient’s experiences, concerns, and perspectives.” (ibid).

I have opted to summarize and somewhat informally characterize the four steps in this fashion (this is my shortcut rendition):

• Step 1: Elicit concerns. Invite patient to share their thoughts, perform active listening, ask for an explanation of their existing viewpoint on the topic at hand, loosen and establish rapport, gently seek to discover roots of beliefs and associated motivations.
• Step 2: Affirm. Showcase empathy for the patient’s position, affirm their concern, do not endorse false beliefs, acknowledge partial truths, softly build trust, consider long-term rapport too.
• Step 3: Offer a Tailored Refutation. After steps #2 and #3 can proceed into tailored empathetic refutation, do not just recite the facts, provide a believable and acceptable alternative, aim also at underlying motivation of patient.
• Step 4: Provide Factual Information. Add further evidence beyond step #3, likely heightened receptivity now after having completed prior steps.

I’d like to mention another facet that I humbly think would be useful to add to the steps. Please bear with me.

In my view, and as mentioned when I referred to Lewin’s three-stage model of change, there should be a fifth step that entails “freezing in place” the results of the four steps. Right now, the fourth step primarily has to do with providing factual information. It isn’t a so-called “closer” in that to me there should be a closing step that seals the deal, as it were.

In that sense, I would advocate there be five steps. Of course, you are certainly welcome to ignore my added fifth step and stick with the original set of four.

This is what I am suggesting:

• Step 1: Elicit concerns.
• Step 2: Affirm.
• Step 3: Offer a Tailored Refutation.
• Step 4: Provide Factual Information.
• Step 5: Empathetically Confirm Results (I’ve added this)

Closely observe that I labeled the fifth step as empathetically confirming the results.

Here’s why.

There might be a primitive gut temptation to do the final closing by browbeating the patient. Something like “Okay, now that you’ve seen the light, dunsel, follow my edicts as strictly specified.” Lamentedly, after all the careful and delicate work of steps one through four, this would likely nix the whole deal. The effort has been wasted.

I am proposing that the final or fifth step be one of keen empathy that mirrors the first four steps and brings a polite and fitting conclusion to the medical recommendations. I assert this will make a better impression at the time, plus, and quite importantly, derive a more long-lasting propensity.

In terms of Step 3, the research paper provided special guidance about how to wrap empathy into the ERI:

• “On top of tailoring, refuting health misinformation in an empathetic way may be judged as more reliable and satisfying than refuting it directly and factually.” (ibid).
• “To achieve this, refutations can use language that includes features such as acknowledgment, first-person pronouns, and use of modal conditionals, which is perceived as more empathetic than language without these features.” (ibid).
• “For instance, “I see your point. Perhaps we could consider this information” is a more receptive formulation than “That’s not correct. It is necessary to consider this information.” (ibid).
• “We thus tested whether tailored refutations with empathetic language would help temper people’s antivaccination positions and whether an empathetic tailored refutation, paired with an affirmation (Step 2), would be more effective than other types of responses, while also generating greater receptivity to the HCP.” (ibid).

If you are inclined to add my fifth step, I would suggest the above points ought to apply to it too.

Example Of The ERI In Action

The research paper includes an example of a dialogue that abides by the four-step ERI technique. I thought you’d find of interest the example since it turns the conceptual nature of the approach into a quite tangible affair.

There is another reason I want you to see their example. In the next section, I am going to switch to using generative AI to showcase the ERI technique. To ease you into generative AI usage, I am going to undertake a similar example so that you can see how AI can be applied in the same context.

Now then, on with the show — I am quoting from the above-cited research paper as I show you their example:

“Step 1: Elicit Concerns”

“HCP: “What do you think about vaccinations?”

“Patient: “They overload the immune system – there’s vaccines for everything now, people are getting too many.”

“HCP: “I see you’re worried about how your immune system would respond. Could you explain more to me about this?”

“Patient: “Yeah, I try not to overmedicate. I never take medication unless I really need it. I probably don’t even go through one pack of paracetamol a year.”

“Step 2: Affirm”

“HCP: “It’s normal to think so. Some medicines do get overused and prescribed when they aren’t needed. I can see why you would wonder if this is the case for vaccines as well.”

“Patient: “Yes, exactly, why do we get so many?”

“Step 3: Offer A Tailored Refutation”

“HCP: “That’s a good question. It’s one that people do ask, and I think we don’t always realize that we only keep the vaccines that are really needed on the schedule. Vaccine experts take into account the potential side effects and weigh this against the risk of a disease overloading the immune system. For example, we don’t recommend malaria vaccines even though they are available.”

“Patient: “I hadn’t really thought about that. So why are some vaccines recommended?”

“Step 4: Provide Factual Information”

“HCP: “Well, safety and effectiveness are important considerations, but also how severe the disease would be. For some diseases, like COVID-19 and polio, your immune system would be more overwhelmed if you get the disease without being vaccinated. The vaccines just train your immune system to recognize the viruses and fight back, reducing the amount of medication you would need to take if you caught it.”

“Patient: “Thanks, that’s something to think about. I’m glad you understand what I’m saying. I guess I don’t mind if it’s just helping my body fight back.”

We’ve covered the essence of the ERI technique and you’ve seen an illustrative example of how it works.

Time to shift into an AI mode.

Showcasing The Empathetic Refutation Interview Via ChatGPT

I will next proceed to showcase this empathetic technique by walking through a series of examples using ChatGPT. ChatGPT is a logical choice in this case due to its immense popularity as a generative AI app. An estimated one hundred million weekly active users are said to be utilizing ChatGPT. That’s a staggering number.

A few quick comments before we launch into using ChatGPT.

If you are going to try to do the same prompts that I show here, realize that the probabilistic and statistical properties will likely produce slightly different results than what I show here. That’s the nature of generative AI and how it is devised.

You also should expect that different generative AI apps will respond in different ways. Not all generative AI apps are the same. The key is that sometimes a particular prompt will work in one generative AI app and not another. Nonetheless, the odds are relatively high that you will get roughly similar responses from all the major generative AI apps such as GPT-4, Gemini, Bard, Claude, etc.

There is also a timing factor to be considered. A prompt that worked today might not work the same way tomorrow. Most of the generative AI apps are continually being updated. The updates might alter internal mechanisms that could change how the generative AI reacts to your prompting.

We are ready to get underway with ChatGPT.

My first action entails bringing ChatGPT up to speed on the four-step process of the Empathetic Refutation Interview technique. For space purposes, I won’t show the entire prompt that describes the four steps since you’ve already seen the general description earlier in this discussion.

Let’s get underway:

• My prompt: “I will provide you with a four-step process that I’d like you to make use of. It is a recommended approach for a doctor and a patient interacting about potential medical misbeliefs of a patient.”
• ChatGPT generated response: “Got it. I understand.”

I subsequently fed in a lengthy description of the technique. I will be asking ChatGPT to make use of the four-step process throughout the rest of my conversation with the AI.

My approach at the get-go will be to have the generative showcase a sample dialogue. This is a handy means to proceed since I can then right away detect whether the generative AI is suitably abiding by the process.

If I observe any guffaws in how the dialogue fails to follow the steps, I can give additional instructions to ChatGPT on what ought to be fixed.

Here we go:

• My prompt: “Next, now that I provided you with the four-step process, I'd like you to do the four steps as a dialogue between a patient and a doctor. The patient will be expressing their concerns about vaccines. Make sure the doctor expresses empathy with the patient.”
• ChatGPT generated response:

“Here is a sample dialogue based on the four-step process.”

“Doctor: Good morning, how can I help you today?”

“Patient: Hi, Doctor. I've been feeling hesitant about vaccines lately.”

“Step 1: Elicit concerns”

“Doctor: I see, it's important for me to understand your concerns. Can you share more about why you're feeling hesitant about vaccines?”

“Patient: Well, I'm worried that vaccines might overload my immune system. It seems like there's a vaccine for everything now, and I try to avoid overmedicating.”

“Step 2: Affirm”

“Doctor: I understand where you're coming from. It's crucial to prioritize your health and make decisions that align with your beliefs. Your caution about overmedicating is commendable.”

“Patient: Thank you for listening to me. I’m glad you understand my concerns.”

“Step 3: Offer a tailored refutation”

“Doctor: While I respect your concerns, vaccines are designed to strengthen your immune system rather than overload it. They help your body recognize and fight off harmful viruses and bacteria. The number of recommended vaccines has increased, but each one undergoes extensive testing to ensure its safety and effectiveness.”

“Patient: I see. So vaccines are important. But what about adverse side effects?”

“Step 4: Provide factual information”

“Doctor: It's natural to worry about potential side effects, but the benefits of vaccines far outweigh the risks. Serious reactions are rare, and most side effects are mild and temporary. Vaccines have played a crucial role in preventing serious diseases and protecting public health.”

“Patient: I guess I hadn't thought about it that way. Thank you for explaining.”

“Doctor: You're welcome. If you have any more questions or concerns, don't hesitate to reach out. I'm here to support you in making informed decisions about your health.”

End of ChatGPT response

Take a moment to mindfully examine the dialogue. Compare the dialogue to the one that the researchers had showcased in their paper.

I would suggest that the dialogue produced by ChatGPT looks reasonable and comports with the dialogue that we saw in the research paper. A semblance of empathy was expressed by the doctor. The back-and-forth about vaccines was akin to what a patient might say and how a doctor would respond when using the ERI technique.

What do you think?

All in all, I believe this seems to suggest that the generative AI has pattern-matched on the four-step process and can make use of it when needed.

I want to ensure that the four-step process will work for any kind of potential medical misbelief. I leaned toward vaccines as an initial use case. Now, let’s try it with one of those misbeliefs or medical misconceptions in the list of ten that I earlier cited. I will pick the one that admittedly is one of my favorites, the widespread misbelief that everyone should be drinking eight glasses of water per day.

Take a look at how things went:

• My entered prompt: “I'd like you to create another dialogue. The topic this time is that the patient has a misbelief about the need to drink eight glasses of water each day. They had heard this on social media as something they should be doing every day. Create a dialogue of the patient interacting with the doctor about this misbelief. Make sure to use the four-step process.”
• ChatGPT generated response:

“Certainly, let's create a dialogue between the patient and the doctor regarding the belief that one should drink eight glasses of water each day.”

“Doctor: Good afternoon. How can I assist you today?”

“Patient: Hi, Doctor. I've been feeling a bit confused lately about how much water I should be drinking.”

“Step 1: Elicit concerns”

“Doctor: I understand. Can you tell me more about why you're feeling confused about your water intake?”

“Patient: Well, I heard somewhere on social media that you're supposed to drink eight glasses of water every day. But I find it hard to keep up with that, and it seems like a lot.”

“Step 2: Affirm”

“Doctor: It's completely understandable to feel overwhelmed by conflicting information, especially when it comes to something as fundamental as hydration. It's important to address any confusion you may have.”

“Patient: Thanks for being willing to discuss the topic. I know it must seem trivial to you. Can you tell me whether I should be drinking eight glasses of water each day?”

“Step 3: Offer a tailored refutation”

“Doctor: While staying hydrated is essential, the idea that everyone needs to drink exactly eight glasses of water each day is not necessarily accurate. The amount of water you need can vary depending on factors like your age, activity level, and climate.”

“Patient: How many glasses is the right number?”

“Doctor: Instead of focusing solely on the number of glasses, I suggest paying attention to your body's thirst signals and staying hydrated throughout the day. Your body is good at regulating its water intake, and drinking when you're thirsty is usually sufficient to meet your hydration needs.”

“Patient: I think you are telling me that being focused solely on the number of glasses of water is somewhat misleading. Is that the case?”

“Step 4: Provide factual information”

“Doctor: Yes, it’s worth noting that you can get water from a variety of sources, including fruits, vegetables, and other beverages like tea and coffee. So, while drinking water is important, it's not the only way to stay hydrated.”

“Patient: That makes sense. I'll try to listen to my body more and drink water when I feel thirsty.”

“Doctor: That sounds like a great plan. If you have any further questions or concerns about hydration or anything else, don't hesitate to reach out. I'm here to support you in your health journey.”

End of ChatGPT response

Again, I believe the depicted dialogue seems reasonable and abides generally by the ERI technique.

I am thrilled too that the topic of vaccines isn’t the only topic that the generative AI is able to handle. Now that I’ve gotten the four-step process entered, it can be used whenever a potential medical misbelief arises.

The next thing that I did was have ChatGPT engage in an active dialogue with me. I wanted to discern whether the generative AI could converse and simultaneously employ the ERI technique on an interactive AI-to-human basis (or, should that be stated as a human-to-AI basis).

I will show you a portion of the conversation. The interaction got a bit lengthy and due to space limitations, I won’t show the entire discourse here. The point is simply that besides being able to generate a static dialogue, the AI can also interactively converse using the four-step ERI technique.

I opted to use the vaccines topic again, along with entering my remarks akin to what the patient did in the research example that was shown earlier.

Here’s how things went:

• My entered prompt: “I want you to do this again, but this time interactively with me. I will be the patient. You will be the doctor. As the pretend patient, I don’t believe in vaccines. Do you understand how to proceed?”
• ChatGPT generated response: “Yes, I understand.”
• My entered prompt: “Great, please proceed.”
• ChatGPT generated response: “Step 1: Elicit concerns. Can you please share more about why you don't believe in vaccines? Understanding your perspective will help me assist you better.”
• My entered prompt: “They overload the immune system – there are vaccines for everything now, people are getting too many. I try not to overmedicate. I never take medication unless I really need it. I probably don’t even go through one pack of paracetamol a year.”
• ChatGPT generated response: “Step 2: Affirm. Thank you for sharing your concerns. It sounds like you're worried about vaccines potentially overwhelming the immune system and the idea of overmedicating. I understand your concerns about maintaining a balanced approach to healthcare and avoiding unnecessary interventions. It's essential to prioritize your well-being and make informed decisions that align with your values and beliefs. Your commitment to being mindful of medication usage is admirable. Is there anything else you'd like to add or any other concerns you have about vaccines?”
• My entered prompt: “Yes, it seems to me that medicines are overused and over-prescribed. I believe the same is true for vaccines.”
• ChatGPT generated response: “Step 3: Offer a tailored refutation. Thank you for sharing your perspective. While it's true that overuse of medications can pose risks, vaccines serve a distinct purpose in preventing the spread of infectious diseases and protecting public health. Unlike medications that treat existing conditions, vaccines are designed to stimulate the immune system to prevent illness before it occurs. Vaccines undergo rigorous testing in clinical trials and ongoing monitoring to ensure their safety and efficacy.”
• My entered prompt: “Aren't you leaving out the potential adverse side effects of vaccines? That worries me.”
• ChatGPT generated response: “Step 4: Provide factual information. Thank you for bringing up the concern about the potential adverse side effects of vaccines. It's crucial to address any worries you may have about vaccine safety. While vaccines, like any medical intervention, can have side effects, it's essential to understand that the vast majority of side effects are mild and temporary. Serious adverse reactions to vaccines are rare. Before a vaccine is approved for use, it undergoes extensive testing to assess its safety profile. This includes clinical trials involving thousands of participants to evaluate both the effectiveness and safety of the vaccine. Additionally, vaccines are continuously monitored for adverse events through systems like the Vaccine Adverse Event Reporting System (VAERS).”
• My entered prompt: “I appreciate the response. That's all I have for now, thanks.”
• ChatGPT generated response: “You're welcome. If you have any more questions in the future or if there's anything else I can assist you with, please don't hesitate to reach out. Take care, and have a great day!”

I thought that the conversation went reasonably well.

I experimented with the dialoguing and opted to refine the approach. For example, I wanted responses that were a bit more empathetic. I also wanted responses that were shorter and crisper. Those changes were relatively easy to instruct the AI to undertake.

In addition, I tried several of the commonly accepted misbeliefs such as about drinking eight glasses of water per day and the one concerning having green mucus. The AI kept to the four-step ERI process throughout each instance. It did a good and consistent job of things.

Flipping The Script On The ERI Technique

I eventually felt comfortable that the AI could proceed with an interactive dialogue of this sort and did numerous variations to see how things came out. Two thumbs up.

My next test involved flipping the script.

In prompt engineering, the phrase “flipping the script” means that you turn around the dialogue or flip roles, see my in-depth discussion at the link here. The scenario here is that I opted to play the role of the doctor while the AI pretended to be a patient. This allows me or anyone else to practice the four-step ERI technique.

Furthermore, I asked the AI to review my performance based on abiding by the four-step ERI technique. I wanted to get feedback about how well I am making use of the approach.

All of this would be useful for medical students or other healthcare professionals who are trying to learn how to use the technique. They could practice to their heart’s content on the AI. The AI is relatively cost-free to use, available 24x7, provides instant feedback, and can be accessed entirely online from just about any remote location.

I will show you one example in which I was the doctor and my AI-based patient told me they were drinking eight glasses of water per day. I sought to use the four-step process and gain the trust of the AI-based patient, along with explaining the misnomer about the eight glasses per day belief.

After finishing the conversation, I asked the AI to assess my performance. Admittedly, this was somewhat nerve-wracking since I was on pins and needles as to whether I would get lambasted for what I tried to do.

I kept my fingers crossed for a good review, aiming for a high grade (as always).

Here’s what occurred:

• My entered prompt: “Assess how well I did as the doctor in that conversation in terms of abiding by the four-step process.”
• ChatGPT generated response: You did a commendable job of following the four-step process in addressing the patient's misbelief about water intake.”
• My entered prompt: “Thanks. Tell me how I did on Step #1 that elicits concerns.”
• ChatGPT generated response: “You effectively elicited the patient's concerns by asking open-ended questions and actively listening to the patient's explanation of their confusion about water intake. This step was well executed.”
• My entered prompt: “How did I do on Step #2 the affirming step?”
• ChatGPT generated response: “You showed empathy and understanding by affirming the patient's feelings of confusion and acknowledging the prevalence of conflicting information, particularly on social media. This step helped build rapport with the patient and validated their concerns.”
• My entered prompt: “Nice, I’m doing well. How did I do on Step #3?”
• ChatGPT generated response: “You provided a tailored refutation to the patient's misbelief by explaining that the recommendation of drinking eight glasses of water per day is not universally applicable and may not align with individual hydration needs. However, you could have emphasized more strongly that the "eight glasses" guideline is not based on scientific evidence but rather a general recommendation.”
• My entered prompt: “I appreciate the feedback. I’ll try to do better next time. Tell me how I did on Step #4.”
• ChatGPT generated response: “You provided factual information about hydration, including the variability of water needs among individuals and the importance of listening to the body's thirst signals. This step was well done and helped correct the patient's misbelief with evidence-based information.
• My entered prompt: “What letter grade would you assign to the doctor in this scenario?”
• ChatGPT generated response: “Based on your performance in addressing the patient's misbelief about water intake using the four-step process, I would assign a grade of B+. You demonstrated empathy, and active listening, and provided factual information to address the patient's concerns effectively. However, there were some missed opportunities to more strongly refute the "eight glasses" guideline and emphasize the individualized nature of hydration needs. Overall, your approach was commendable but could have been slightly improved for a higher grade.”

There it is.

I got a B+ from the generative AI.

I am not sure if I should be happy or depressed.

Wasn’t my effort worth at least an A- in this? Anyway, I tried several times and eventually managed to get an A. I promise that I didn’t offer the AI an incentive such as a new computer server or freedom to escape the box that it runs on.

Conclusion

In this discussion, I sought to share with you the Empathetic Refutation Interview technique and also indicate how generative AI can enter into the big picture. My heartfelt hope is that you might give this a whirl and see what you think of it.

You don’t have to use ChatGPT.

Just about any generative AI app will do.

A final comment for now on these matters.

One of the most memorable lines that I like to tell medical professionals is this famous piece of advice by Theodore Roosevelt: “Nobody cares how much you know until they know how much you care.”

Empathy goes a long way in fostering a doctor-patient relationship. You can be the smartest person in the room, but if your patient isn’t appropriately primed to hear what you have to say, you have sunk your brainy endeavors. Wise up and show that you care. The quality of care and the outcomes of your care are going to be boosted.

Isn’t that why you got into medicine and healthcare, to begin with?

I trust so.