How can ChatGPT be used in Healthcare?
There has been great publicity recently on the capabilities of ChatGPT. It is considered the state-of-the-art in conversational AI, given its large training knowledge and its strong natural language understanding. One may ask ChatGPT encyclopedic questions such as “What is the population of the USA?” or domain-specific questions such as “What is the cause of diabetes?” An interesting feature is that it can change the language style of its response, for example by asking “Explain the causes of diabetes to a small child.”
Given the powerful capabilities of ChatGPT, people have been trying to find its potential killer applications in all industries. For example, many people view it as a competitor to Google or to Alexa or Siri.
In this article, we will discuss how ChatGPT can be used in healthcare applications. Let’s first start with a review of healthcare applications for chatbots in general, which we discussed in a previous article. For each application, we will discuss if and how ChatGPT can be leveraged.
1. Self-diagnosis
This is an area that was very hot a few years ago, and has since cooled down a little bit. There are several platforms that allow patients to enter their symptoms, such as Babylon Health, Healthily or WebMD. These platforms use a combination of machine learning and manual expert input. They are programmed to ask the right next question given the patient input so far. This is something that ChatGPT is not good at, that is, ChatGPT is not good at asking clarification questions to solicit the right additional information from a user. Further, it is hard to compete with algorithms trained with extensive human input and calibration in a specialized domain.
To avoid possible liability or confusion, ChatGPT is explicit on not being able to do symptom checking, responding “I am an AI language model and do not have the ability to diagnose medical conditions. It is recommended that you consult a doctor or a medical professional for proper diagnosis and treatment.”
Verdict: Not applicable
2. Healthcare navigation
The goal here is to help patients navigate the complex healthcare landscape; for example, find a doctor in their network, compare prices, or learn about their benefits. These chatbots are typically available only to members of a healthcare system, so they are not freely accessible on the web. The technology of these bots can range from simple multiple-choice diagrams to deep learning models that are trained on existing customer service chats.
ChatGPT is not trained to provide local information, such as look for doctors in an area. Instead, it forwards the patient to other resources: “To find a pediatrician in Irvine, CA, you can use the following resources: Online directories such as ZocDoc, Healthgrades, and WebMD, Your health insurance company's website or customer service line,....”
On the other hand, ChatGPT could be used to answer general healthcare questions such as “What is deductible in healthcare” or “What is a co-pay.”
ChatGPT could be used to augment a specialized healthcare navigation chatbot, by possibly handling general knowledge questions that the specialized chatbot cannot answer.
Verdict: Can complement existing chatbots to make them more responsive and cover more questions.
3. Health coaching and wellness
These chatbots try to engage patients to achieve a more healthy lifestyle. For example, they may remind patients to exercise and eat healthy every day. Examples of such chatbots include Doppel and 1-million-strong-to-prevent-diabetes. They may also assess the progress of a patient using existing clinical scales, such as PHQ9, which is a popular 9-question scale to assess depression.
There is no easy way to program ChatGPT to give wellness advice or to monitor a user’s progress over time. Furthermore, ChatGPT does not support administering questionnaires.
ChatGPT could be used to complement such bots. For example, one could ask ChatGPT if eating pasta is healthy or to suggest ways to burn calories.
Verdict: Can complement existing chatbots to make them more responsive and cover more questions.
4. Healthcare marketing
These bots typically collect information from leads (potential customers/patients), provide them information on a practice’s services or make an appointment. Such bots are more popular in out-of-pocket specialties, such as dentists, chiropractors, plastic surgeons or physical therapy, which are the specialties that invest more in marketing.
A key intuition here is that as patients search for services on Google, if a practice’s website has a chatbot that can engage the patient 24x7, it will make it more likely to attract this patient, especially in the case of millennials, where instant gratification is key.
ChatGPT would not be very useful here, as it is hard to program it to have a goal (e.g. collect a user’s email) and solicit information from users.
Verdict: Not applicable
In summary, we see that ChatGPT, in its present form, is not applicable for some types of healthcare AI applications, and could be used to complement existing technologies in some other applications. As ChatGPT evolves and becomes more customizable, its applicability in healthcare could also increase.
by Vagelis H.