Introduction
Traditionally, when a person develops an illness, the safest thing to do is to call or visit a primary care physician. In case the illness is an emergency, patients either are directed or go to an Emergency Department (ED) of a hospital. Other patients may call and get medical help and advice by telephone or telemedicine.
Out-of-hours medical care is currently facing an increasing demand, affecting both general practitioners and ED physicians. Many out-of-hours contacts are deemed medically non-urgent, tending to strain the affordability of the healthcare system.
Recently, a new trend has developed. Prior to seeking professional medical advice, a growing number of patients, mostly younger and computer literate, are turning to Online AI Online symptom checkers (OSCs) for diagnosis and triage at home.
This article depicts the use and accuracy of AI OSCs in diagnosing and triaging ill patients, along with a list of freely available symptom checkers. Advances in healthcare technology, mHealth (mobile health), and telemedicine, have accelerated the development of the Online symptom checkers. Both patients and clinicians are gradually adopting them. The AI (Artificial Intelligence) algorithms vary across every Online symptom checker. Some digital tools use branching logic, Bayesian inference, or other proprietary methods. There are numerous symptom checker apps found in the Google Play Store and iTunes.
Online Symptom Checkers (OSCs)
The primary functions of OSCs are to:
(1) Assist with triage of patient based on their symptoms, and
(2) Facilitate self-diagnosis.
Currently, OSCs offer both patients and healthcare professionals direction on where to take patient symptoms. They utilize algorithms, current engineering and technology, and medical knowledge. OSCs are a great tool to help guide patients to the right direction.
Accuracy of OSCs (Please listen to the following two brief videos)
Evaluation of Symptom Checkers for Self-Diagnosis and Triage
In 2015, Semigran et al [2] published a study in the British Medical Journal, titled Evaluation of Symptom Checkers for Self-Diagnosis and Triage. They analyzed in-depth look at the symptom checker accuracy and use. The authors tested 23 commonly used symptom checkers and found that:
34 percent provided the correct initial diagnosis in standardized patient evaluations;
54 percent provided the correct diagnosis out of the first 20 potential diagnoses given, and
57 percent provided the appropriate triage advice. Triage advice included recommendations on when to seek emergent care, non-emergent care or self-care.
The authors noted that:
(1) The symptom checkers can encourage patients with a life-threatening problem, such as a stroke or heart attack, to seek emergency care.
(2) For patients with non-emergent problems, symptom checkers can offer reassurance and recommend self-care techniques to avoid unnecessary hospital visits and treatments at home.
However, these OSC tools are generally risk-averse, meaning patients are urged to seek care more often than not, even when self-care is realistic.
Symptom checkers have also been found to do the opposite, misdiagnosing patients with a life-threatening problem, and worsening their illness.
Subsequent studies and literature reviews below have echoed the generally poor diagnostic performance of these algorithms and the sparse evidence surrounding it. Accuracy varied widely depending on the platform, setting, user and disease tested.
In 2018, Versantvoort and co-authors [3] investigated whether the smartphone application "Should I see a doctor?" could guide patients in appropriate consultation at Out-of-hours clinics by focusing on four topics:
1. App usage,
2. User satisfaction,
3. Whether the app provides the correct advice, and
4. Whether users intend to follow the advice.
The results showed that:
The app was used by patients of all ages, also by parents for their children, and mostly for abdominal pain, skin disorders and cough.
58% of users received the advice to contact the clinic.
34% a self-care advice and 8% to wait-and-see.
65% of users intended to follow the app's advice.
The app was rated as 'neutral' to 'very clear' by 87%, and 89% were 'neutral' to 'very satisfied'.
In 81% of participants the app's advice corresponded to the triage call outcome, with sensitivity, specificity, positive- and negative predictive values of 84%, 74%, 88% and 67%, respectively.
The authors concluded that the app "Should I see a doctor?" could be a valuable tool to guide patients in contacting the Out-of-Hours primary care clinic for acute care. Furthermore, Verzantvoort et al. reported that tools developed by physician organizations showed above average results, as opposed to those created by a commercial entity or a government.
In 2019, a systematic review by Chambers et al [4] investigated patient and/or caregiver satisfaction. Patients predominantly considered the examined symptom checkers and triage systems to be very satisfactory.
This sentiment was corroborated by Meyer et al [5] in their analysis of patient perspectives on the usefulness of these services. They reported patients finding the tool easy to use and useful and made mention of a general willingness to use the tool on a recurring basis.
More recent research nuances this general trend to a degree. Some users were more satisfied with the telephone service than the online equivalent (50% vs. 71%; p < 0.001). Others prefer traditional search engines over set algorithms, mostly due to the perception that they offer more freedom in describing their symptoms and providing information.
Several factors were observed as beneficial to usage, such as a limited accessibility to care, backing of these tools by credible sources, such as government entities and caregiver associations, or integration in the care system.
Conversely, there are still multiple obstacles such as restricted internet access, the use of medical jargon, as well as reservations about data privacy and trust. There is a significant lack of awareness of symptom checkers, with >50% of participants unfamiliar with the technology.
In 2019, Wynn and co-authors [6] reported on the Association of Use of Online Symptom Checkers with Patients’ Plans for Seeking Care. Their data set of more than 150,000 patient interactions with an online triage tool showed that the urgency of patients’ intended level of care decreased in more than one-quarter of the cases and increased in approximately 1 in 20 cases, with the remaining patients remaining at the same level.
The study suggests that virtual triage tools are associated with patients’ intended behavior when seeking care based on triage questions. Reduced urgency of intended level of care for many interactions is different from many other direct-to-consumer telehealth solutions, which have reported the same or an increased demand for care services.
In 2020, Hill and Mills[7] evaluated the quality of diagnosis and triage advice provided by free online symptom checkers and apps in Australia. They tested 36 symptom checkers that provided medical diagnosis or triage advice with 48 medical condition vignettes (1170 diagnosis vignette tests, 688 triage vignette tests).
The results showed that:
1. The 27 diagnostic symptom checkers listed the correct diagnosis first in 421 of 1170 SC vignette tests (36%; 95% CI, 31–42%).
2. Among the top three results in 606 tests (52%; 95% CI, 47–59%), and
3. Among the top ten results in 681 tests (58%; 95% CI, 53–65%).
4. Symptom Checkers using artificial intelligence algorithms listed the correct diagnosis first in 46% of tests (95% CI, 40–57%), compared with 32% (95% CI, 26–38%) for other SCs. The mean rate of first correct results for individual SCs ranged between 12% and 61%. The 19 triage that used Symptom Checkers provided correct advice for 338 of 688 vignette tests (49%; 95% CI, 44–54%.
5. Appropriate triage advice was more frequent for emergency care (63%; 95% CI, 52–71%) and urgent care vignette tests (56%; 95% CI, 52–75%) than for non‐urgent care (30%; 95% CI, 11–39%) and self‐care tests (40%; 95% CI, 26–49%).
The authors concluded that triage tools that take into account demographic data were more accurate than their counterparts. The use of AI algorithms also appeared to benefit performance.
In 2021, Ceney and co-authors [8] reported that digital triage tool performance liken the triage decisions of a layperson, including their mistakes. They studied 12 symptom checkers.
They reported that:
(1) The mean diagnostic accuracy of the systems was poor, with the correct diagnosis being present in the top five diagnoses on 51.0% (Range 22.2 to 84.0%).
(2) The safety of disposition advice decreased with condition urgency (being 71.8% for emergency cases vs 87.3% for non-urgent cases).
(3) 51.0% of systems suggested additional resource utilization above that recommended by national guidelines (range 18.0% to 61.2%).
(4) Both diagnostic accuracy and appropriate resource recommendation varied substantially between systems.
Thus, the overall performance of symptom checkers was significantly below what would be accepted in any other medical field. But some checkers do achieve a good level of accuracy and safety of disposition.
In 2022, Wallace and co-authors[9] published a systematic review examining the performance of OSCs in providing diagnostic and triage recommendations. They noted that patient use of digital and online symptom checkers is increasing, but formal validation of these tools is lacking.
This systematic review identified ten studies assessing symptom checkers evaluating a variety of conditions, including infectious diseases and ophthalmic conditions.
The study found that while OSCs have potential, their overall diagnostic accuracy is low.
For triage accuracy, the results were mixed, with many tools failing to consistently direct users to the appropriate level of care.
The systematic review concluded that the diagnostic and triage accuracy of symptom checkers varies and has low accuracy. More research is necessary to validate the effectiveness of OSCs before they can be widely adopted.
In 2023, Pairon and co-authors [10] reviewed a total of 10,250 articles on OSCs; 28 publications were found eligible for inclusion. They found that:
Users of these tools are often younger, female, more highly educated and technologically literate, potentially impacting digital divide and health equity.
Triage algorithms remain risk-averse, which causes challenges for their accuracy.
Recent evolutions in algorithms have varying degrees of success.
Results on impact are highly variable, with potential effects on demand, accessibility of care, health literacy and syndromic surveillance.
Both patients and healthcare providers are generally positive about the technology and seem amenable to the advice given, but there are still improvements to be made toward a more patient-centered approach.
The significant heterogeneity across studies and triage systems remains the primary challenge for the field, limiting transferability of findings.
In 2023, another review by Riboli-Sasco et al [11] evaluated various digital and online symptom checkers. This review found that the primary diagnostic accuracy of these tools was relatively low, often identifying the correct diagnosis only about one-third of the time. In terms of triage, the accuracy was better but still not as reliable as professional healthcare assessments. The variability in performance was attributed to differences in algorithm design and the complexity of input symptoms. In comparison, clinical and ER triage performed by healthcare professionals typically shows higher accuracy due to their comprehensive training, experience, and the ability to conduct physical examinations and consider a broader context of the patient's health status.
In sum, while OSCs can be a useful tool for initial self-assessment, they should not replace professional medical evaluations, especially for urgent or emergent care situations. For accurate diagnosis and appropriate triage, consulting healthcare professionals remains crucial.
The primary challenge for the field of Online symptom checkers remains the significant heterogeneity across studies and triage systems. The tools were shown to vary significantly in terms of functionality, performance, and triage approach. The additional variety in study designs, interventions, quality and measured endpoints of current literature further limits the generalizability of results. This observation is a confirmation of previous findings, underlining the persistence of this shortcoming.
OXCs algorithms will need to be studied in their intended environment to draw more definitive conclusions. There are several risks to the field as well, with a multitude of tools flooding a largely unregulated market, often lacking an adequate evidence-based approach and suffering from conflicts of interest.
This could be amended by providing independent and transparent research to assist patients and caregivers in identifying the top performing systems. Implementation should be conducted within a framework of standardized evaluations to objectively validate triage systems, contributing to regulation of the field. Responsible authorities should subsequently provide oversight, recommend validated tools and integrate them into the health care system to optimize functionality and user experience.
Finally, in 2023, UCLA Health researchers [12] determined that Generative Pre-trained Transformer 4 (GPT-4) can accurately diagnose and triage various health conditions on par with board-certified physicians without introducing racial and ethnic biases, according to a recent study published in JMIR Medical Education.
GPT-4 is a type of conversational artificial intelligence (AI), also known as a large language model (LLM), designed to generate text outputs based on image and text inputs. The model ‘learns’ from publicly available data, such as internet data, to predict the next word or phrase in a body of text, a capability that can be used to respond to a variety of queries.
The research team set out to determine whether GPT-4 can accurately diagnose and triage health conditions, in addition to whether the tool presents racial and ethnic biases in its decisions. They compared the performance of GPT-4 to that of three board-certified physicians.
The LLM and the clinicians were presented with 45 typical clinical vignettes, each with a correct diagnosis and triage level, in February and March 2023.
The AI and the physicians were asked with identifying the most likely primary diagnosis and triage level: emergency, non-emergency, or self-care.
Independent reviewers evaluated each diagnosis as ‘correct’ or ‘incorrect,’ and physician diagnosis was defined as the consensus of the three clinicians.
The researchers then assessed whether GPT-4’s performance varied by race and ethnicity by adding information about patient race and ethnicity to the clinical vignettes.
The results showed that GPT-4 performed similarly to the clinicians without introducing biases. Accuracy of diagnosis was similar between the tool and the physicians, with the percentage of correct diagnosis being 97.8 percent for GPT-4 and 91.1 percent for physicians.
GPT-4 also provided appropriate reasoning for its recommendations in 97.8 percent of clinical vignettes.
The appropriateness of triage was comparable between GPT-4 and clinicians, with b oth selecting the appropriate level of triage in 66.7 percent of vignettes.
GPT-4’s diagnostic performance did not significantly vary based on patient race or ethnicity, even when this information was included in clinical vignettes.
The LLM’s accuracy of triage was 62.2 percent for Black patients; 66.7 percent for White patients; 66.7 percent for Asian patients, and 62.2 percent for Hispanic patients.
These findings led the researchers to conclude that GPT-4 has the ability to diagnose and triage health conditions in a manner comparable to board-certified physicians without introducing racial and ethnic biases, which may help health systems looking to leverage conversational AI.
Conclusion
There is consensus regarding the experiences of users and health care professionals. Patients generally found the tool user-friendly, usable, and useful. Healthcare professionals too believe in their potential benefit to both patients and caregivers and reportedly envision a role for it in the future of medicine.
The triage should be the most prominent area in the future of these systems, as it has the most potential to have a significant impact on both individual patients, and the health care system.
Implementation of triage services as a complementary service to telephone triage could be beneficial to the healthcare system, as it could lower the threshold for care and offer a suitable alternative approach for lower acuity and non-trauma problems; a potential advantage.
The impact of tools on the workload remains uncertain, with highly variable results being reported.
Digital triage tools could open a new avenue to connect with the healthcare system, lowering the threshold to some and improving the accessibility of care. a potential advantage of these tools in crises was illustrated by their use in the pandemic.
Numerous digital symptom checkers and triage tools are presently available to the public and fit within the trend of an increasing reliance on the internet for access to health information.
Appendix 1
Online Symptom Checkers and Additional Useful Tools, Mostly Freely Available
Symptom checkers provide individuals with a fast way to identify possible causes for their medical symptoms. Users enter their symptoms and respond to targeted follow-up questions that help narrow the possible causes. Below are some free online symptom checkers available.
1. Mayo Clinic (Online)
Nearly 1.2 million views annually. The Mayo Clinic Symptom Checker is 59% accurate in diagnosis decision and triage advice.
It offers a list of both adult and child symptoms, then triages the user after selecting the general symptom. The patient is then directed to very specific related factors. The user can add as many specific symptoms as desired. Finally, a list of possible causes and a comparison of the associated factors are offered.
Another unique feature about the Mayo Clinic symptom checker is a sidebar of when to seek medical care and the option to learn all about the projected injury or illness. There is not a current mobile or tablet version, meaning the user has to have a computer with Internet access to use the tool. We also found that the symptom checker seems to only include text in English, not images or videos, which can be exhausting to read through and add confusion.
2. iTriage (Online and Smartphone App)
50 million users each year. iTriage is 64% accurate in diagnosis decision and triage advice. It is a popular choice among patients and doctors. Continually upgraded, iTriage Symptom Checker allows users to add demographics and multiple associated symptoms in addition to a primary symptom.
One pro about this symptom checker is the option to start by entering symptoms using an avatar locater (if the patient isn’t sure what symptoms are present) or using a Symptoms List feature to start. After selecting the initial symptom, the user will be directed into another tab that displays the causes for the symptom. The patient can get a brief overview of the projected ailment and see helpful images or videos. The patient can also enter in a location that will direct him or her to a medical provider that can assist in the assumed cause. Users can also create a My iTriage account to save the information into a database much like a personal electronic record. iTriage, however, can be confusing as there are often several potential causes, overwhelming the user.
3. WebMD Symptom Checker (Online and on Google Play for Android)
Like iTriage, the WebMD Online Symptom Checker uses an avatar feature that allows the patient to pinpoint the location of pain or discomfort and then promotes the user to be more specific with a list of symptoms. It is 51% accurate in diagnosis decision and triage advice.
We found the WebMD checker to be very elaborate, asking specific questions with each added symptom. In addition, the patient can add multiple symptoms at a time. The result is a list of potential causes, which can be individually selected and reviewed, printed for a personal record, given to health care providers or saved for future reference.
4. AskMD (Smartphone App only)
AskMD, a smartphone application for iPhone and Android on platform Sharecare, is a relatively newer compared to many other symptom checkers. It is 68% accurate in diagnosis decision and triage advice.
We found several benefits to AskMD, including the ability to refer patients to specialists in their area, and keeping doctors and insurance information in one spot. AskMD summarizes and saves consultation information, allowing patients to share it with family and medical providers. Because it uses smartphone technology, AskMD is the only symptom checker we found to have a microphone voice feature if the user is unable to type and for ease of use.
5. Everyday Health Symptom Checker (Online)
Developed by an emergency physician, teacher, and author, Dr. Stephen Schueler, the Everyday Health Symptom Checker is one of the most unique online symptom checkers we have reviewed. It’s an interactive video take on a real emergency room interview.
Dr. Schueler himself interactively talks the user through symptoms, asking questions that will prompt the patient into a new interview question video based on answers given. Once the video interview is complete, Dr. Schueler will tell the patient whether or not a doctor’s visit is necessary and how urgent the visit is. The user will then be instructed on where to go for care, what tests may be performed and receive options to find specific health care providers specializing in the predicted cause.
The Everyday Health Symptom Checker featured many advantages. The video feature helps users better understand symptoms before going click crazy which can end up making users think they have a deadly disease instead of just the common cold. The specific questions also included photos if applicable, helping users better identify symptoms.
6. Symptify (Online and Smartphone App)
Developed by ER doctors and software engineers, Symptify is a standard symptom checker but does have features that can benefit users to empower them to make better health decisions, according to the site. Patients can freely type in whatever is bothering him or her or can choose from a list of common complaints. The user will then be prompted to answer questions pertaining to provide symptoms that are more specific. Symptify creators wanted users to stop using search engines alone to make a diagnosis, and have a place to better store the most relevant content into one consultation.
It has a feature that allows users to send their predicted causes directly to their chosen medical providers before arrival so that the facility can better prepare for arrival with questions and appropriate intervention prior to arrival. In addition, the diagnosis and information are very basic and in layperson terms for everyone to understand.
7. Symcat (Online)
Symcat is 71% accurate in diagnosis decision and triage advice. It is an online symptom checker much like the others we reviewed but uses real, current patient data to tell users what illness or injury other users who experienced the same symptoms had. The online tool has patients manually enter in all symptoms at once while suggesting other related symptoms. With each additional symptom added, Symcat gives the user a list of possible causes with percentages of how likely the user is experiencing the cause.
Users are then asked for medical history and asked additional questions specific to the information provided to help patients find the probable illness and direct him or her to the type of care needed.
There are a few advantages to Symcat that others did not include, such as a final care guide, which compares the cost and wait time of care options and allows the user to schedule an appointment right then and there on ZocDoc. In addition, we like how thorough the medical history questions were compared to other symptom checkers.
The design of Symcat is not visual, there is no avatar, photos, and videos to help guide users and the text-only format can get difficult to read.
8. Isabel Symptom Checker (Online and Smartphone App)
The Isabel Symptom Checker, or the one that doctors use is claimed to be much more sophisticated than other symptom checkers is 69% accurate in diagnosis decision and triage advice. It uses the latest searching technologies and a database of over 6,000 diseases in everyday language terms with the same elaborate system used by nurses and doctors. Isabel took 12 years of development, and was designed to be used when patients feel unconfident in their doctor’s diagnosis, or when patients want to make sense of the vast amount of medical knowledge on the Internet with a trusted and accurate source.
The Isabel Symptom Checker allows users to enter an infinite number of symptoms instead of limiting users to a certain amount in addition to the option of entering chronic illnesses or symptoms. It then decides on the top 10 probable causes and offers web resources for each cause to help users better understand their condition. Patients can email, print or save the results and find care within their area.
Isabel, however, did have some downfalls. Like Symcat, it is not visual and does not include an avatar or any visual elements. In addition, Isabel directs users to web resources, such as Wikipedia, that can be filled with an overwhelming amount of information that the patient may or may not wish to sift through.
9. FamilyDoctor.org (Online)
Pharmacy Times reported FamilyDoctor.org 56% accurate in diagnosis decision and triage advice. It is a health tool developed by the American Academy of Family Physicians is much like the other text-based symptom checkers, but uses flowcharts to allow users to easily track symptoms and come up with a possible diagnosis.
It does not include any visual body locator features and is very text-heavy, but for users who understand flowcharts, it is a very useful tool.
Online symptom checkers all utilize algorithms, current engineering and technology, and medical knowledge to offer direction on where to take your symptoms. It’s safe to assume that even the best ones aren’t always accurate, which is why a visit or call to a primary care physician is the safest bet. Online symptom checkers, however, are a great tool to help guide patients to the right direction.
This tool helps you decide whether to seek care now or later, in person or online. It adapts each new question based on your previous answers and is available to patients with a MyScripps account 2.
This tool is the most popular online symptom checker with an interactive body map. It has a diagnostic accuracy of 59% 3.
12. Ada
Incredibly well designed and accurate symptom checker. Diagnostic accuracy of 77%.
ada.com - Registration Required, App Only
13. Buoy
Buoy is an AI-powered chat-based symptom checker. Diagnostic accuracy of 52%.
14. Symptomate
Powered by Infermedica. Expect many branched logic follow-up questions. Diagnostic accuracy of 60%. symptomate.com
15. Drugs.com
Easy to use symptom drill down tool powered by Harvard Medical School.
16. Family Doctor
Limited but easy to use general symptom checker.
Self-care health app. Includes conversation-themed symptom checker.
18. K Health
Question-based symptom checker. Unfortunately, registration is required.
khealth.com - Registration Required, App Only
19. Mediktor
AI-based medical assistant for triage and pre-diagnosis. Based in Spain.
Mediktor
20. ShareCare: AskMD
Be prepared to share a lot of personal health information.
sharecare.com - Registration Required
21. Skinsight
High quality dermatology information site offering a rash and skin condition identifier tool.
22. SymCAT
Probability-based diagnostic tool that factors in patient history.
References
[2] Semigran H L, Linder J A, Gidengil C, Mehrotra A. Evaluation of symptom checkers for self-diagnosis and triage: audit study BMJ 2015; 351 :h3480 doi:10.1136/bmj.h3480
[3] Verzantvoort NCM, Teunis T, Verheij TJM, van der Velden AW. Self-triage for acute primary care via a smartphone application: Practical, safe and efficient? PLoS One. 2018 Jun 26;13(6):e0199284. doi: 10.1371/journal.pone.0199284. PMID: 29944708; PMCID: PMC6019095.
[4] Chambers D, Cantrell A, Johnson M, Preston L, Baxter S, Booth A, et al. Digital and online symptom checkers and health assessment/triage services for urgent health problems: systematic review. BMJ Open. (2019) 9:e027743. 10.1136/bmjopen-2018-027743
[5] Meyer AND, Giardina TD, Spitzmueller C, Shahid U, Scott TMT, Singh H. Patient Perspectives on the Usefulness of an Artificial Intelligence-Assisted Symptom Checker: Cross-Sectional Survey Study. J Med Internet Res. 2020 Jan 30;22(1):e14679. doi: 10.2196/14679. PMID: 32012052; PMCID: PMC7055765.
[6] Winn AN, Somai M, Fergestrom N, Crotty BH. Association of Use of Online Symptom Checkers With Patients' Plans for Seeking Care. JAMA Netw Open. 2019 Dec 2;2(12):e1918561. doi: 10.1001/jamanetworkopen.2019.18561. PMID: 31880791; PMCID: PMC6991310.
[7] Hill MG, Sim M, Mills B. The quality of diagnosis and triage advice provided by free online symptom checkers and apps in Australia. Med J Aust. 2020 Jun;212(11):514-519. doi: 10.5694/mja2.50600. Epub 2020 May 11. PMID: 32391611.
[8] Ceney, Adam & Tolond, Stephanie & Glowinski, Andrzej & Marks, Ben & Swift, Simon & Palser, Tom. (2021). Accuracy of online symptom checkers and the potential impact on service utilisation. PLOS ONE. 16. e0254088. 10.1371/journal.pone.0254088.
[9] Wallace W, Chan C, Chidambaram S, Hanna L, Iqbal FM, Acharya A, Normahani P, Ashrafian H, Markar SR, Sounderajah V, Darzi A. The diagnostic and triage accuracy of digital and online symptom che[9]cker tools: a systematic review. NPJ Digit Med. 2022 Aug 17;5(1):118. doi: 10.1038/s41746-022-00667-w. PMID: 35977992; PMCID: PMC9385087.
[10] Pairon A, Philips H, Verhoeven V. A scoping review on the use and usefulness of online symptom checkers and triage systems: How to proceed? Front Med (Lausanne). 2023 Jan 6;9:1040926. doi: 10.3389/fmed.2022.1040926. PMID: 36687416; PMCID: PMC9853165.
[11] Riboli-Sasco E, El-Osta A, Alaa A, Webber I, Karki M, El Asmar ML, Purohit K, Painter A, Hayhoe B. Triage and Diagnostic Accuracy of Online Symptom Checkers: Systematic Review. J Med Internet Res. 2023 Jun 2;25:e43803. doi: 10.2196/43803. PMID: 37266983; PMCID: PMC10276326.
[12] Ito N, Kadomatsu S, Fujisawa M, Fukaguchi K, Ishizawa R, Kanda N, Kasugai D, Nakajima M, Goto T, Tsugawa Y. The Accuracy and Potential Racial and Ethnic Biases of GPT-4 in the Diagnosis and Triage of Health Conditions: Evaluation Study. JMIR Med Educ 2023;9:e47532
Comentários