Other than scheduled check-ups, when we have new ailments, we all begin with self-triage in one form or another. Symptom checkers offer a convenient, digital way to sift through symptoms, then receive healthcare advice and connections to appropriate level of care. There are two formats used to do this sifting and both have strengths and weaknesses.
For severe symptoms, both quickly instruct the user to call 911 or go to the emergency room. But, most people aren’t taking the time to use symptom checkers when symptoms are severe and clear emergencies. After determining level of severity, both types of digital symptoms checkers can link to locations and care options such as online scheduling, telehealth, and nearby urgent or emergency care locations.
Let’s start with the chatbot:
Using an automated chatbot, patients are asked a series of questions pertaining to their constellation of symptoms in a friendly, conversational format. It simulates a texting conversation. At the end, there is a final segment whereby the bot can suggest the next steps for appropriate level of healthcare. “One benefit to a good chat configuration is its ability to guide the user through their own assessment,” says Sue Riffel, CEO of Self Care Decisions. “When done well, the user is asked questions which walk them through grouped concerns relative to their symptom: appearance, behavior, duration, related symptoms, etc. The primary downside is the chatbot may not reveal which of the user’s answers drove the final outcome. The chatbot may reach the right result, but as far as knowledge transfer, it’s a ‘black box’ (no transparency) as to how the result was reached.”
Some bots go beyond triage and match answers to the most likely diagnoses. These tools claim to improve their accuracy (through AI) the more they are used. If true, they might be better at rare diagnoses that present with more unusual constellation of symptoms. The ability to generate a group of likely diagnoses make these AI chatbots particularly well-suited for medical teaching. However, for more common ailments, questions tend to go on for quite a while, especially if more than one symptom is entered. It can be difficult for the user to know when the process is reaching the end, thus leading to a higher rate of abandonment.
Now let’s examine the checklist format:
The checklist format starts with matching the user’s chief complaint with the best care guide. At this point, the user is presented with a list of medical scenarios starting from most serious, moving down to the least. Starting from the top, the user reads each short phrase and selects the best match to their situation. Their selection then presents a recommended action (level of care needed and in what time frame) and links to the appropriate options for care. It might be to get care now, online telehealth, scheduling an in-person visit or to call during office hours.
As users read the symptom list in order of severity, they learn which scenarios signify a more serious condition. They also learn what signs they need to keep an eye on. In many cases, they are reassured when they reach the bottom and learn they can focus on care at home, for now. At any phase of this process the patient can skip directly to the self-care advice to learn more. This knowledge gain in two areas of the process is not something that the chatbot format can easily do.
Pediatric settings (practices, Children’s Hospitals) gravitate to checklist symptom checkers. Their audiences, parents – especially new parents, are educated by each symptom check and the care advice has been shown to build self-care skills. So, the transparency of how the decision is made is especially helpful in developing parents’ assessment skills. The checklist offers new knowledge which they will build upon with each use.
Why not offer both?
As we proceed within this digital health age, some are kicking and screaming but most seem quite pleased. It is clear more patients are looking for digital solutions and the pandemic has escalated that pursuit. There may be an advantage to offer both chatbot and checklist formats so that a patient could choose what they prefer. Though chatbots are popular, it is worth understanding where they shine best and how that aligns with the needs of their intended audience. The solution ‘fit’ may come down to the needs and preferences of the primary audience. Self Care Decisions will be supporting both formats in 2022.
Bonnie Offit, MD is a general pediatrician trained at Children’s Hospital of Philadelphia (CHOP). For over 30 years she has provided care for children in both primary and urgent care settings. She recently served as clinical director for telehealth in the Office of Digital Health at CHOP prior to her current role of Chief of Health Innovation at Self Care Decisions. Her interests focus on using technology to improve the patient experience and ultimately improve outcomes by getting the right care at the right time.
General references:
- Palanica1 A, Flaschner2 P, Thommandram1 A, et al. Physicians’ perceptions of Chatbots in health care: Cross-sectional web-based survey. Journal of Medical Internet Research.https://www.jmir.org/2019/4/e12887/. Accessed October 26, 2021.
- Philipson B. Chatbots in Healthcare: A more humanized, patient-oriented approach. Healthcare IT News. https://hitconsultant.net/2021/09/10/chatbots-healthcare-humanized-patient-oriented-approach/#.YXhbFNnMKbg. Published September 10, 2021. AccessedOctober 26, 2021.
- Lerman R. The robot will see you now: Health-care chatbots boom but still can’t replace doctors. The Washington Post. https://www.washingtonpost.com/technology/2021/07/26/healthcare-chatbots-pandemic-rise/. Published August 13, 2021. Accessed October 26, 2021.