How reliable are ChatGPT and Google’s answers to frequently asked questions about unicondylar knee arthroplasty from a scientific perspective?

Abstract

Introduction

Unicondylar knee arthroplasty (UKA) is a minimally invasive surgical technique that replaces a specific compartment of the knee joint. Patients increasingly rely on digital tools such as Google and ChatGPT for healthcare information. This study aims to compare the accuracy, reliability, and applicability of the information provided by these two platforms regarding unicondylar knee arthroplasty.

Materials and Methods

This study was conducted using a descriptive and comparative content analysis approach. 12 frequently asked questions regarding unicondylar knee arthroplasty were identified through Google’s “People Also Ask” section and then directed to ChatGPT-4. The responses were compared based on scientific accuracy, level of detail, source reliability, applicability, and consistency. Readability analysis was conducted using DISCERN, FKGL, SMOG, and FRES scores.

Results

A total of 83.3% of ChatGPT’s responses were found to be consistent with academic sources, whereas this rate was 58.3% for Google. ChatGPT’s answers of 142.8 words, compared to Google’s 85.6-word average. Regarding source reliability, 66.7% of ChatGPT’s responses were based on academic guidelines, whereas Google’s percentage was 41.7%. The DISCERN score for ChatGPT was 64.4, whereas Google’s was 48.7. Google had a higher FRES score.

Conclusion

ChatGPT provides more scientifically accurate information than Google, while Google offers simpler and more comprehensible content. However, the academic language used by ChatGPT may be challenging for some patient groups, whereas Google’s superficial information is a significant limitation. In the future, the development of Artificial Intelligence-based medical information tools could be beneficial in improving patient safety and the quality of information dissemination.

Keywords

unicondylar knee arthroplasty artificial intelligence chatgpt google

Introduction

Osteoarthrosis is the leading cause of knee pain in the elderly, with a lifetime risk of 50% for symptomatic knee osteoarthritis.¹ Knee arthroplasty is a well-established, effective surgical treatment routinely applied to appropriate patients for the management of gonarthrosis.² For the treatment of unicompartmental gonarthrosis, unicondylar knee prosthesis, first introduced in 1954, can be utilized.³ When compared to total knee arthroplasty, unicondylar knee arthroplasty is considered a less invasive approach due to its smaller skin incision, reduced blood loss, and superior preservation of the knee’s native anatomy.^4,5

The rapid advancements in artificial intelligence have introduced novel analytical techniques and algorithms, significantly enhancing their application within medical practices and demonstrating the potential to reshape medical disciplines, including orthopedics and traumatology.^6,7 In 2018, OpenAI introduced the ChatGPT model, a language model with advanced linguistic capabilities, made publicly available as a computer program. ChatGPT is increasingly being adopted in the medical field,^8,9 serving as a communication tool for patients to acquire information, share knowledge, and seek solutions to their health-related concerns.¹⁰ Since 1998, Google, the most widely used search engine globally, has facilitated the easy access to health information.¹¹ However, patients frequently rely on search engines or artificial intelligence applications for obtaining health information without consulting a healthcare provider, which can sometimes lead to inaccurate or misleading results.¹²

This study aims to compare the information provided by ChatGPT and Google regarding unicondylar knee prosthesis, focusing on accuracy, reliability, detail level, and applicability. The findings will assess the reliability of artificial intelligence patient education and evaluate the advantages and limitations of artificial intelligence in comparison to traditional search engine-based information sources.

Materials and Methods

This study aims to compare the responses provided by Google’s “People Also Ask” section and ChatGPT-4 regarding patient education on unicondylar knee prosthesis in terms of accuracy, scope, level of detail, and applicability. The study was conducted using a descriptive and comparative content analysis methodology. It is a preliminary study conducted between January 1st, 2025, and February 28th, 2025. Given the nature of the study, blinding was not applicable, as the data consisted of publicly accessible information retrieved from internetbased platforms. The responses from ChatGPT-4 were based on its training data, with a knowledge cut-off of April 2023.

All academic sources used in this study were carefully verified. Peer-reviewed articles were selected primarily from PubMed, Scopus, and Web of Science, and assessed for relevance, scientific quality, and publication credibility. When possible, DOIs and PMIDs were checked to confirm source authenticity.

Additionally, when appropriate, the content of responses was evaluated in light of established clinical guidelines, including those published by NICE and AAOS. The study adhered to ethical standards, including the principles outlined in the NICE guidelines and the Declaration of Helsinki.

The research was conducted in two phases. In the first phase, the 12 most frequently asked patient questions concerning unicondylar knee prosthesis were identified through the “People Also Ask” section on Google. This methodological approach was chosen due to its reflection of patient information-seeking behavior and its ability to provide analysis grounded in real-world data. These questions were selected to highlight the most commonly sought topics by patients and to facilitate a comparative analysis. Google searches were performed using keywords such as “Unicondylar Knee Prosthesis Treatment,” “Unicondylar Knee Prosthesis Recovery,” and “Unicondylar Knee Prosthesis Surgery,” and the questions suggested by the search algorithm were recorded. In the second phase, the same 12 questions were directed to ChatGPT-4, and the generated responses were documented. In cases where multiple responses were produced by ChatGPT, only the first response was considered. To ensure objectivity, all responses were independently evaluated by two medically trained reviewers, who resolved any differences by mutual agreement.

The responses obtained from Google and ChatGPT were compared according to five primary criteria: scientific accuracy, level of detail, source reliability, applicability, and response consistency. Scientific accuracy was assessed by evaluating the alignment of the responses with evidence-based medical guidelines, academic literature, and current research. This criterion is essential to ensure that accurate and reliable information is provided during the patient education process. High scientific accuracy reduces the risk of patients encountering misleading or incomplete information, thereby supporting clinical decision-making. The level of detail was evaluated based on the comprehensiveness of the information presented. This evaluation considered the amount of medical information included, the level of technical terminology used, and the depth of the explanations. Additionally, metrics such as the DISCERN score and FKGL readability scores, which have been used in previous patient education content, were employed to assess the level of detail. Source reliability was analyzed by examining the academic credibility of the information and whether the responses were corroborated by recognized medical authorities. Applicability was evaluated based on the clarity and practical relevance of the responses for patients. Response consistency was measured by assessing the coherence between multiple answers generated on the same topic (Table 1).

Table 1.

Rothwell classification of questions related to unicondylar knee prosthesis.

Rothwell’s Classification	Description
Fact	Provides information regarding the accuracy of the question, based on objective data
Policy	Questions whether a specific treatment approach should be followed
Value	Requests an evaluation of a thought, object, or event

Class	Subcategory	Description
Question classification and explanations
Fact	Activity restrictions	Questions related to post-operative activity limitations, such as the ability to engage in sports and the return to activity time frame after unicondylar knee prosthesis
	Recovery process	Specific questions about the post-surgical recovery process and timelines
	Technical details	Questions regarding the type of prosthesis used, surgical methods, and implants
	Cost	Inquiries about the costs associated with surgery and the prosthesis
	Anatomy / Function	Structural information about the knee joint and the relationship with the prosthesis
	Diagnosis	Questions regarding the diagnostic criteria for unicondylar knee prosthesis
Policy	Indications and management	Inquiries regarding when unicondylar knee prosthesis is indicated and how the decision for surgery is made
Policy	Risks and complications	Questions concerning post-surgical risks, such as infection, loosening, and other potential complications
Value	Pain management	Questions about post-surgical pain management and the use of medications
	Longevity	Questions regarding the durability, lifespan, and long-term outcomes of the prosthesis
	Surgical success evaluation	Evaluation questions concerning the effectiveness of unicondylar knee prosthesis surgeries and patient satisfaction
	Comparison with other treatments	Comparative questions regarding unicondylar knee prosthesis versus total knee prosthesis or conservative treatment options

Category	Description
Website categories (Source evaluation)
Academic	Websites of universities, academic medical centers, and scientific publications
Commercial	Websites of medical companies, prosthesis manufacturers, and pharmaceutical companies offering health information
Government Sources	Official health organizations, ministries, and national health guidelines
Medical practice	Websites of hospitals and orthopedic clinics
Single-surgeon personal	Personal websites, doctor blogs, and individual surgeon pages
Social media	Blogs, patient forums, and user-generated content on the internet.

Cohen’s Kappa statistic was used to calculate inter-rater agreement in the statistical analysis. Differences between Google and ChatGPT responses in terms of scientific accuracy and level of detail were statistically analyzed using Fisher’s exact test or chi-square test. The lengths of the responses and the technical terms used were compared using the Mann-Whitney U test. These analyses aimed to determine the reliability of both platforms for patient education and highlight the advantages and limitations of both systems. Responses from both ChatGPT and Google were compared using the DISCERN score, FKGL, SMOG, and FRES. Parametric data were analyzed using independent samples t test, and non-parametric data were analyzed using the MannWhitney U test.

Discern score

DISCERN is a 16-item tool designed to assess the quality of written materials provided for patient education. These items are grouped into three sections: reliability of the publication (Items 1–8), presentation of treatment options (Items 9–15), and overall assessment (Item 16). Each item is scored from 1 (very poor) to 5 (very good). The total score is calculated out of 80: 63–80: Excellent; 51–62: Good; 39–50: Fair; <39: Poor.

Readability metrics

The comprehensibility of the responses was automatically calculated using the following metrics:Flesch-Kincaid Grade Level (FKGL): Indicates the average education level required to understand the text. SMOG (Simple Measure of Gobbledygook): Assesses the academic complexity of the text. FRES (Flesch Reading Ease Score): The higher the score (closer to 100), the easier the text is to read.

Results

In this study, the responses provided to 12 patient queries were compared between ChatGPT and Google in terms of scientific accuracy, level of detail, source reliability, applicability, and consistency of responses (Table 2).

Table 2.

Most frequently asked questions regarding unicondylar knee arthroplasty according to google and ChatGPT.

	Google	ChatGPT
1	Is unicondylar knee arthroplasty a major surgery? (Evaluation of surgery)	How long does it take to fully recover from unicondylar knee arthroplasty? (Timeline of recovery)
2	What is the recovery time after unicondylar knee arthroplasty? (Timeline of recovery)	What are the potential risks and complications of unicondylar knee arthroplasty? (Risks/complications)
3	What is the difference between unicondylar knee arthroplasty and total knee arthroplasty? (Technical details)	When can I return to sports or physical activities after unicondylar knee arthroplasty? (Specific activities)
4	How long does pain last after unicondylar knee arthroplasty? (Pain	What type of anesthesia is used during unicondylar knee arthroplasty? (Technical details)
5	Will my knee return to its original state after surgery? (Evaluation of surgery)	When should I start physical therapy after unicondylar knee arthroplasty? (Timeline of recovery)
6	Is unicondylar knee arthroplasty worth having? (Evaluation of surgery)	Will I need to use a cane or assistive devices after unicondylar knee arthroplasty? (Indications/management)
7	Will I lose weight after unicondylar knee arthroplasty? (Risks/complications)	How long should I wear a knee brace after surgery? (Timeline of recovery)
8	What will my knee look like after unicondylar knee arthroplasty? (Indications/management)	Will I need another surgery after unicondylar knee arthroplasty? (Risks/complications
9	Why is the recovery process so painful after unicondylar knee arthroplasty? (Pain)	How soon after unicondylar knee arthroplasty can I drive? (Specific activities)
10	What is the lifespan of a unicondylar knee prosthesis? (Technical details)	What restrictions should I follow after unicondylar knee arthroplasty? (Restrictions)

The responses generated by ChatGPT were based on literature, medical terminology, and evidencebased information, with 83.3% of ChatGPT’s answers aligning with scholarly sources. In contrast, only 58.3% of Google’s responses reflected evidence-based content. Furthermore, Google’s responses often relied more on patient experiences and commercial content. The average word length of ChatGPT’s responses was 142.8 ± 21.3 words, whereas Google’s responses averaged 85.6 ± 18.5 words, indicating that ChatGPT provided more detailed, technical information while Google’s answers tended to be more superficial (Table 3).

Table 3.

Comparison of google and ChatGPT responses to 12 questions regarding unicondylar knee arthroplasty.

Question	Google Response	ChatGPT Response
1. What is unicondylar knee arthroplasty and who is it suitable for?	It is suitable for patients with damage to only part of the knee joint and is a minimally invasive surgery	It is applied to only the medial or lateral compartment of the knee, offering a 50% faster recovery time
2. Should unicondylar or total knee arthroplasty be preferred?	Unicondylar knee arthroplasty provides a smaller incision and faster recovery, while total knee arthroplasty is suitable for advanced osteoarthritis cases	Total knee arthroplasty involves the replacement of the entire joint, whereas unicondylar arthroplasty only covers part of the joint. Patient selection is crucial
3. What is the recovery time after unicondylar knee arthroplasty?	Walking begins within the first week, full range of motion is achieved in 6-8 weeks, and return to sports occurs within 3-6 months	Daily activities can generally be resumed within 4-6 weeks, with full functional recovery taking 3-6 months
4. What activities can be performed after unicondylar knee arthroplasty?	Low-impact exercises like swimming and walking can be done, but high-impact sports are not recommended	Light sports activities can resume within 6-8 weeks, but high-impact sports should only be performed with the recommendation of a physician
5. How long does pain last after unicondylar knee arthroplasty?	Pain may be intense during the first week but will largely subside within 6 weeks with proper pain management	Pain is more intense during the first 2 weeks after surgery, but it should subside completely within 6-8 weeks
6. Is physical therapy required after unicondylar knee arthroplasty?	Physical therapy is generally not recommended but may benefit some patients	Physical therapy is recommended for early mobilization and should begin within the first 2-4 weeks
7. Is there a risk of infection after unicondylar knee arthroplasty?	The risk of infection is low (<1%), but wound infection signs should be monitored	Infection rates are less than 1%, and proper care and antibiotic use minimize the risk
8. Will another surgery be necessary after unicondylar knee arthroplasty?	The prosthesis lasts between 15 and 20 years; revision may be necessary if there is wear or loosening	The lifespan of the prosthesis depends on the patient’s activity level and typically lasts 15-20 years, though revision surgery may be needed
9. What is the lifespan of a unicondylar knee prosthesis?	The average lifespan is 15-20 years, with wear depending on the patient	Similarly, the prosthesis typically lasts 15-20 years, depending on the patient’s weight and activity level
10. How many days will I stay in the hospital after unicondylar knee arthroplasty?	Most patients stay for 1-2 days, though some may be discharged on the same day	The hospital stay is usually 1-2 days, with early mobilization being encouraged
11. Will the prosthesis loosen after unicondylar knee arthroplasty?	Over time, the prosthesis may loosen, requiring revision surgery	The risk is less than 5% if the prosthesis is correctly implanted, though the patient’s activity level may influence this risk
12. When can I drive after unicondylar knee arthroplasty?	Generally, driving can begin 4-6 weeks after surgery, though it may take longer if the right leg is operated on	Driving may be possible within 3-6 weeks, depending on the individual recovery process

The responses from ChatGPT were based on academic guidelines, orthopedic expert articles, and scientific sources, with 66.7% of the responses relying directly on scholarly literature. In contrast, 41.7% of Google’s responses included academic and medical content, with a greater prevalence of content derived from commercial clinical websites. While 75% of Google’s answers offered practical advice for daily life, ChatGPT’s responses included 58.3% of such content (Figure 1). Despite the higher level of technical detail in ChatGPT’s responses, some patient queries were more clearly answered by Google, offering more accessible and practical information (Table 4).

Figure 1.

Distribution of content in ChatGPT and google responses.

Table 4.

Comparison of the first 5 frequently asked questions on unicondylar knee arthroplasty between google and ChatGPT responses.

Question	Google Response	ChatGPT Response
1. Is unicondylar knee arthroplasty a major surgery?	It is a minimally invasive surgery with smaller incisions and faster recovery compared to total knee arthroplasty, but the surgical outcomes and patient suitability are critical	Unicondylar knee arthroplasty is performed on one part of the knee, requiring smaller incisions and offering faster recovery. However, it should be applied to more selected patient groups compared to total knee arthroplasty
2. How long does the recovery take after unicondylar knee arthroplasty?	Recovery time varies based on the patient’s age, weight, and physical activity. Generally, walking begins in the first week, and full recovery may take 3-6 months	Recovery time can vary, but most patients can return to daily activities within 4-6 weeks, with full functional recovery and return to sports taking 3-6 months
3. What is the difference between unicondylar knee arthroplasty and total knee arthroplasty?	Unicondylar knee arthroplasty is suitable for patients with damage to only part of the knee and is performed with smaller incisions, while total knee arthroplasty replaces the entire knee joint and caters to a broader group of patients	Unicondylar knee arthroplasty is for patients with wear in only one compartment of the knee and preserves some natural joint tissue, whereas total knee arthroplasty replaces the entire joint surface and is typically performed in advanced osteoarthritis cases
4. How long does pain last after unicondylar knee arthroplasty?	Post-surgery pain is typically intense during the first week but gradually decreases, with most patients experiencing significant relief within 6 weeks	Pain is most intense in the first days post-surgery, but proper pain management should reduce it significantly within the first 2 weeks, with complete relief usually within 6-8 weeks
5. When can I resume sports after unicondylar knee arthroplasty?	Low-impact sports (e.g., swimming, walking, cycling) can be resumed 6-8 weeks after surgery, while running and high-impact activities are not recommended	Light activities and low-impact exercises can typically begin after 6-8 weeks, but high-impact sports should only be resumed with a physician’s approval

The consistency of ChatGPT’s responses was calculated to be 8.3%, while Google’s responses showed variability of 25% across different search results, with some queries yielding different answers.

ChatGPT’s responses had higher DISCERN and SMOG scores, indicating a more academic and comprehensive content structure. However, the FKGL and FRES results suggested that the content might be more challenging to read for patients. Google responses, while more superficial, were presented in simpler language and were easier to read according to the FRES scores (Table 5).

Table 5.

Measurement metrics.

Metric	ChatGPT (Average)	Google (Average)
DISCERN Score	64.4	48.7
FKGL	15.92	10.88
SMOG	16.52	12.63
FRES	19.29	50.52

Discussion

Artificial intelligence systems, such as ChatGPT, have inherent limitations related to their training data, as well as drawbacks such as the inaccessibility of certain referenced links.^13,14 Conversely, Google offers advantages, including the continuous repetition of links, greater accessibility, and ease of readability; however, the propagation of inaccurate information, which may be mistakenly perceived as valid, can have adverse consequences for patients.^13,14 Artificial intelligence methodologies should be employed interactively to compensate for each other’s shortcomings, with specialized system training required for data analysis.

In the study by Dubin et al. on arthroplasty, a comparison of Google and ChatGPT data revealed that the majority of the questions were ‘fact-based,’ with 25% of them being similar.¹⁵ In our research, Google web searches, classified according to the Rotwell taxonomy, included questions in all three categories—‘fact,’ ‘policy,’ and ‘value,’ with 50% of questions categorized as ‘value.’ On the other hand, ChatGPT predominantly contained questions in the ‘fact’ and ‘policy’ categories, with ‘fact’-based questions accounting for over 50%. Additionally, ChatGPT did not include any questions in the ‘value’ category. There were very few overlapping questions between Google and ChatGPT within this category.

Recent literature suggests that ChatGPT’s responses are more comprehensive, with greater reliance on academic references compared to Google.^16,17 Eryılmaz et al. demonstrated that ChatGPT provided more objective and comprehensive answers than Google in the context of arthroscopic meniscus surgery.¹⁸ In the present study, we found that ChatGPT’s responses were more aligned with the scientific literature, incorporating medical terminology and evidence-based information, whereas Google’s responses were more influenced by patient experience. Moreover, when analyzing content length, Google responses were more succinct.

Mika et al. suggested that ChatGPT, when addressing frequently asked patient questions regarding total hip arthroplasty, predominantly provided evidence-based responses and could be a valuable tool for patient education.¹⁹ In a separate study, Michael et al. noted that for numerical response queries, both Google and ChatGPT provided highly similar answers.²⁰

In 2016, Lawson et al. identified that the highest DISCERN score on academic websites was 51, while the general DISCERN average across all websites was 44.²¹ Hurley et al. (2023) reported an average DISCERN score of 60 in their study examining AI’s knowledge on shoulder instability.²² In studies with high-quality medical content, readability and informational quality were often suboptimal, with Lawson et al. reporting an FRES score of 50.17, corresponding to a readability level equivalent to 10.98.²¹ Daraz et al., in their 2018 systematic review, found that FKGL values among orthopedic surgeons were slightly above 10.²³ In our study, the DISCERN scores exceeding 60 suggest that the content is generally appropriate for patient education. However, the elevated FKGL and SMOG scores imply that the content may be challenging for the majority of patient populations to comprehend. These findings are further supported by the FRES scores. On the other hand, Google-based responses, while having lower DISCERN scores, were more successful in terms of readability, suggesting that Google’s content is more superficial but easier for patients to understand. However, due to the lack of scientific grounding, it is not advisable for patients to base their treatment decisions solely on such information. Ethical concerns arise when patients rely solely on AI-generated information for medical decision-making. Although ChatGPT can improve access to health information, it is not a substitute for professional medical advice. Ensuring physician oversight and promoting digital health literacy are crucial for safe and effective use.²⁴

Conclusion

In our study, it was observed that ChatGPT responses were more likely to cite academic articles and scientific sources, whereas Google responses were predominantly focused on practical, daily-life recommendations. Concerning unicondylar knee prosthesis information, ChatGPT offered more detailed medical content, while Google provided more accessible and actionable advice relevant to daily life.

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iDs

Ali Aydilek

Ömer Levent Karadamar

References

Murphy

Schwartz

Helmick

, et al. Lifetime risk of symptomatic knee osteoarthritis. Arthritis Rheum 2008; 59(9): 1207–1213.

Deng

Zhang

, et al. Unicondylar knee replacement versus total knee replacement for the treatment of medial knee osteoarthritis: a systematic review and meta-analysis. Arch Orthop Trauma Surg 2021; 141: 1361–1372.

Johal

Nakano

Baxter

, et al. Unicompartmental knee arthroplasty: the past, current controversies, and future perspectives. J Knee Surg 2018; 31(10): 992–998.

Freigang

Rupp

Pfeifer

, et al. Patient-reported outcome after patient-specific unicondylar knee arthroplasty for unicompartmental knee osteoarthritis. BMC Muscoskelet Disord 2020; 21: 773.

Koskinen

Paavolainen

Eskelinen

, et al. Unicondylar knee replacement for primary osteoarthritis: a prospective follow-up study of 1,819 patients from the Finnish Arthroplasty register. Acta Orthop 2007; 78(1): 128–135.

Cheng

Guo

, et al. Emergency surgery in the era of artificial intelligence: ChatGPT could be the doctor’s right-hand man. Int J Surg 2023; 109: 1816–1818.

Temsah

Khan

Chaiah

, et al. Overview of early ChatGPT’s presence in medical literature: insights from a hybrid literature review by ChatGPT and human experts. Cureus 2023; 15: e37281.

Tharakan

Klein

Bartlett

, et al.

Do ChatGPT and Google differ in answers to commonly asked patient questions regarding total shoulder and total elbow arthroplasty?

J Shoulder Elb Surg 2024; 33: e429–437. DOI: 10.1016/j.jse.2023.11.014.

Jamal

Solaiman

Alhasan

, et al. Integrating ChatGPT in medical education: adapting curricula to cultivate competent physicians for the AI era. Cureus 2023; 15: e43036.

10.

Sun

Zhang

Gwizdka

, et al. Consumer evaluation of the quality of online health information: systematic literature review of relevant criteria and indicators. J Med Internet Res 2019; 21: e12522.

11.

Kothari

Moolani

. Reliability of “Google” for obtaining medical information. Indian J Ophthalmol 2015; 63: 267–269. DOI: 10.4103/0301-4738.156934.

12.

Stokel-Walker

. ChatGPT listed as author on research papers: many scientists disapprove. Nature 2023; 613: 620–621.

13.

Lee

Bubeck

Petro

. Benefits, limits, and risks of GPT-4 as an AI chatbot for medicine. N Engl J Med 2023; 388: 1233–1239.

14.

Giorgino

Alessandri-Bonetti

Del Re

, et al. Google bard and ChatGPT in orthopedics: which is the better doctor in sports medicine and pediatric orthopedics? The role of AI in patient education. Diagnostics 2024; 14(12): 1253.

15.

Dubin

Bains

Chen

, et al. Using a Google web search analysis to assess the utility of ChatGPT in total joint arthroplasty. J Arthroplast 2023; 38: 1195–1202.

16.

Tharakan

Klein

Bartlett

, et al.

Do ChatGPT and Google differ in answers to commonly asked patient questions regarding total shoulder and total elbow arthroplasty?

J Shoulder Elb Surg 2024; 33: e429–437.

17.

Shen

Pratap

, et al. How does ChatGPT use source information compared with Google? A text network analysis of online health information. Clin Orthop Relat Res 2024; 482: 578–588.

18.

Eryilmaz

Aydin

Turemis

, et al.

ChatGPT-4.0 vs. Google: which provides more academic answers to patients' questions on arthroscopic meniscus repair?

Cureus 2024; 16: e76380.

19.

Mika

Martin

Engstrom

, et al. Assessing ChatGPT responses to common patient questions regarding total hip arthroplasty. J Bone Joint Surg Am 2023; 105: 1519–1526.

20.

Gaudiani

Castle

Abbas

, et al. ChatGPT-4 generates more accurate and complete responses to common patient questions about anterior cruciate ligament reconstruction than Google’s search engine. Arthrosc Sports Med Rehabil 2024; 6(3): 100939.

21.

Lawson

Codella

Ciccotti

, et al. Evaluation of internet information about rotator cuff repair. Am J Orthop (Belle Mead NJ) 2016; 45: 136–142.

22.

Hurley

Crook

Lorentz

, et al. Evaluation highquality of Information from ChatGPT (artificial intelligencelarge language model) artificial intelligence on shoulder stabilization surgery. Arthroscopy 2023; 40: 726–731.e6.

23.

Daraz

Morrow

Ponce

, et al. Readability of online health information: a meta-narrative systematic review. Am J Med Qual 2018; 33: 487–492. DOI: 10.1177/1062860617751639.

24.

Mennella

Maniscalco

De Pietro

, et al. Ethical and regulatory challenges of AI technologies in healthcare: A narrative review. Heliyon. 2024; 10(4). DOI: 10.1016/j.heliyon.2024.e26297.