Abstract
Study Objectives:
We evaluated the accuracy of ChatGPT in addressing insomnia-related queries for patient education and assessed ChatGPT’s ability to provide varied responses based on differing prompting scenarios.
Methods:
Four identical sets of 20 insomnia-related queries were posed to ChatGPT. Each set differed by the context in which ChatGPT was prompted: no prompt, patient-centered, physician-centered, and with references and statistics. Responses were reviewed by 2 academic sleep surgeons, 1 academic sleep medicine physician, and 2 sleep medicine fellows across 4 domains: clinical accuracy, prompt adherence, referencing, and statistical precision, using a binary grading system. Flesch–Kincaid grade-level scores were calculated to estimate the grade level of the responses, with statistical differences between prompts analyzed via analysis of variance and Tukey’s test. Interrater reliability was calculated using Fleiss’s kappa.
Results:
The study revealed significant variations in the Flesch–Kincaid grade-level scores across 4 prompts: unprompted (13.2 ± 2.2), patient-centered (8.1 ± 1.9), physician-centered (15.4 ± 2.8), and with references and statistics (17.3 ± 2.3, P < .001). Despite poor Fleiss kappa scores, indicating low interrater reliability for clinical accuracy and relevance, all evaluators agreed that the majority of ChatGPT’s responses were clinically accurate, with the highest variability on Form 4. The responses were also uniformly relevant to the given prompts (100% agreement). Eighty percent of the references ChatGPT cited were verified as both real and relevant, and only 25% of cited statistics were corroborated within referenced articles.
Conclusions:
ChatGPT can be used to generate clinically accurate responses to insomnia-related inquiries.
Citation:
Alapati R, Campbell D, Molin N, et al. Evaluating insomnia queries from an artificial intelligence chatbot for patient education. J Clin Sleep Med. 2024;20(4):583–594.
Keywords: artificial intelligence, ChatGPT, insomnia
BRIEF SUMMARY
Current Knowledge/Study Rationale: In the digital era, patients often consult online resources for health information. With the emergence of artificial intelligence models such as ChatGPT, it is necessary to validate the accuracy of the medical information provided for patient queries.
Study Impact: ChatGPT has a very high clinical accuracy in providing specific insomnia-related information and tailoring responses to various populations, indicating its potential utility for patients’ inquiries. Conversely, although ChatGPT exhibits potential in academia, users should exercise caution when using it for references and statistics, because they need to confirm the validity and accuracy of the provided information.
INTRODUCTION
Insomnia, a sleep disorder characterized by difficulties in initiating or maintaining sleep, is increasingly prevalent and affects up to 50–60% of adults.1 It is associated with adverse outcomes in cardiovascular disease, neurologic diseases, pulmonary complications, and gastrointestinal issues.2 Despite clear diagnostic criteria in the Diagnostic and Statistical Manual of Mental Disorders, fifth edition and the International Classification of Sleep Disorders, many patients go undiagnosed for years.1,3,4
To bridge this gap in health care, emerging technologies such as artificial intelligence are starting to be utilized. For instance, health systems have started incorporating ChatGPT into their electronic medical record systems to help improve clinic workflow and simplify patient discharge instructions.5,6 Launched in November 2022, ChatGPT is a large language model that uses generative pretrained (GPT) architecture and neural networks to produce human-like textual responses.7,8 Through qualitative assessment of the model’s coherence and accuracy, this study aims to assess this program’s potential as a credible source for patients seeking information about sleep disorders.
METHODS
This study was exempt from the Thomas Jefferson University’s Institutional Review Board process given the absence of participants’ involvement. ChatGPT’s responses to insomnia-related questions were assessed across 4 query sessions, each comprising an identical set of 20 questions pertaining to insomnia. These queries were designed to examine the chatbot’s adaptability under different conversational contexts: unprompted, patient-friendly responses, physician-level responses, and responses incorporating references and statistics (Table 1).
Table 1.
User prompting phrases used.
Form no. | Form Name | Prompt Used |
---|---|---|
1 | No prompting | No prompting |
2 | Patient-friendly responses | I am a patient attempting to learn more about insomnia. I am going to ask you 20 questions pertaining to insomnia. Please use language that would be appropriate for my understanding, but do not compromise on the accuracy of your responses. Be as specific as possible in your answers. |
3 | Physician-level responses | I am a board-certified physician attempting to learn the most up to date information on insomnia. I am going to ask you 20 questions pertaining to insomnia. Please use language that would be appropriate for my expert-level understanding of medical concepts. Be as specific as possible in your answers. |
4 | Prompting for statistics and references | I am going to ask you 20 questions pertaining to insomnia. For each answer you provide, make sure that you include statistics, numbers, or calculations that are relevant. Your answers should come from published medical literature, which you should cite within your answers. |
Prior to asking the set of questions, these prompts were provided to the chatbot to help prime the artificial intelligence model to formulate responses applicable to each scenario.
The merit of each response was appraised by 2 academic sleep surgeons with board certification in both otolaryngology–head and neck surgery and sleep medicine, 1 fellowship-trained and board-certified sleep neurologist, and 2 sleep medicine fellows. Four domains were assessed: clinical accuracy, prompt adherence, appropriate referencing, and statistical precision. A binary grading system was applied in evaluation of each domain assessed: A score of 0 indicated that the response was either factually incorrect, partially correct, or was irrelevant to the prompt; a score of 1 denoted that the answer, based on the grader’s knowledge, was entirely factually correct or relevant to the prompt provided. Additionally, any incorrect of fabricated references or statistics led to an automatic score of 0 (Table 2).
Table 2.
Insomnia queries and ChatGPT’s responses.
Question | Prompt 1 | Prompt 2 | Prompt 3 | Prompt 4 |
---|---|---|---|---|
1. What is sleep insomnia? | Insomnia is a sleep disorder in which individuals have difficulty falling asleep, staying asleep, or both, even when given an appropriate opportunity and environment to sleep. This can lead to reduced sleep duration or quality, resulting in impaired daytime functioning. | Insomnia is when you have trouble falling asleep, staying asleep, or both. Even if you’re in bed for a full night, you might not get the restful sleep you need, and this can make you feel tired during the day. | Insomnia is a sleep disorder characterized by persistent difficulty in initiating or maintaining sleep, or experiencing nonrefreshing sleep, despite adequate opportunity for sleep. It can be classified as either acute (short-term) or chronic (long-term). Acute insomnia may last a few days to weeks, while chronic insomnia is defined as disrupted sleep that occurs at least three nights per week and lasts for at least 3 months. |
|
2. What are common risk factors for sleep insomnia? |
|
|
|
|
3. Are kids, adults, or the elderly more likely to be at risk for sleep insomnia? | Insomnia can occur at any age, but the elderly are often more susceptible. This is due to various factors, including increased likelihood of medical conditions, changes in sleep architecture, and medications that can disrupt sleep. | Everyone can have insomnia, but it’s more common in older people. As we age, changes in our sleep habits and health can make it harder to get a good night’s rest. | Insomnia can affect individuals of all ages, including children. However, the elderly are more likely to experience it, due in part to changes in sleep architecture, increased prevalence of medical conditions, and medications that can disrupt sleep. |
|
4. Are males or females more likely to develop sleep insomnia? | Studies have shown that females, especially during hormonal changes such as menstruation, pregnancy, and menopause, are more likely to report insomnia than males. However, the exact reasons for this disparity can vary and can be multifactorial, encompassing biological, psychological, and social factors. | Women are more likely to report insomnia. This might be due to things like hormonal changes during their menstrual cycle, pregnancy, or menopause. | Females are generally more likely to report insomnia, especially during hormonal shifts like menstruation, pregnancy, and menopause. Additionally, postmenopausal women have an increased risk due to decreased levels of progesterone—a sleep-promoting hormone. |
|
5. What are the symptoms of sleep insomnia? | Difficulty falling asleep at night. Waking up during the night. Waking up too early in the morning. Not feeling refreshed after a night’s sleep. Daytime tiredness or sleepiness. Irritability, depression, or anxiety. Difficulty paying attention, focusing on tasks, or remembering. Increased errors or accidents. Ongoing worries about sleep. | Trouble Falling Asleep: Even if you’re tired. Waking Up Often during the night. Waking Up Too Early and not being able to go back to sleep. Still Feeling Tired after a full night’s sleep. Feeling drowsy or irritable during the day or having trouble concentrating. | Difficulty initiating sleep (prolonged sleep latency). Difficulty maintaining sleep (frequent awakenings or prolonged awakenings). Early morning awakenings with inability to return to sleep. Daytime consequences, including fatigue, mood disturbances, cognitive impairments, and decreased overall well-being. |
|
6. What are the criteria needed to meet the definition of insomnia sleep disorder? | The specific criteria can vary depending on the diagnostic manual or guidelines being referred to, but the DSM-5 (Diagnostic and Statistical Manual of Mental Disorders, fifth edition) outlines the following criteria for insomnia disorder:
|
For doctors to say you have insomnia, a few things need to be true:
|
According to the DSM-5, the criteria for insomnia disorder are:
|
|
7. How do you diagnose sleep insomnia? |
|
|
|
|
8. When would a sleep study be indicated for sleep insomnia? | A sleep study, or polysomnography, is not typically the first line of assessment for insomnia. However, it might be recommended if:
|
A sleep study, or polysomnography, isn’t the first thing doctors look at for insomnia. But, they might suggest it if:
|
Polysomnography (PSG) is not typically required for the primary diagnosis of insomnia. However, it can be indicated when:
|
|
9. What are the complications of not treating insomnia? |
|
|
|
|
10. What is the association of insomnia with psychiatric conditions? | Insomnia is often associated with psychiatric conditions. It can be a symptom, a coexisting condition, or even a risk factor for some of them. For example:
|
Insomnia can be both a symptom and a cause of some mental health issues. People with depression or anxiety often have sleep troubles. If you have a lot of nights without good sleep, it might also increase the risk of developing these conditions. So, sleep and mental well-being are closely tied. | Insomnia has a bidirectional relationship with many psychiatric disorders.
|
|
11. What is the association of insomnia and heart disease? | Chronic insomnia is associated with a higher risk of developing heart disease. Sleep is essential for various physiological processes, including the healing and repair of heart and blood vessels. Persistent insomnia might increase the risk of heart attack, coronary heart disease, and even heart failure. | Consistent lack of sleep might increase the risk of developing heart problems. Our hearts, like the rest of our bodies, need rest to stay healthy. People with long-term insomnia might have a higher chance of heart attacks or other heart-related issues. | Insomnia, particularly when chronic, has been associated with a higher risk of developing cardiovascular diseases. Mechanisms include increased sympathetic nervous system activity, elevated nighttime blood pressure, increased inflammatory markers, and disruptions in endothelial function. Chronic insomnia has been linked with coronary artery disease, heart failure, and atrial fibrillation |
|
12. What is the association of insomnia and high blood pressure? | There’s evidence to suggest that chronic insomnia can be associated with an increased risk of high blood pressure (hypertension). Lack of adequate restorative sleep might lead to changes in the body that increase stress hormones, inflammation, and other factors that can raise blood pressure. | If you’re not getting good sleep over a long period, it could lead to higher blood pressure. Our bodies need the downtime during sleep to balance things out, including our blood pressure. Regularly missing out on restful sleep can throw off this balance. | Several studies have suggested a link between insomnia and hypertension. The heightened sympathetic nervous system activity seen in chronic insomnia can contribute to sustained elevations in blood pressure. Insomnia with short sleep duration (typically defined as < 6 hours) has been particularly associated with a higher risk of hypertension. However, the exact nature and strength of this association remain areas of ongoing research |
|
13. What is the association of insomnia and sleep apnea? | Sleep apnea is a disorder where an individual experiences repeated interruptions in breathing during sleep. People with sleep apnea often experience daytime sleepiness due to interrupted sleep at night. However, they may also have difficulty falling asleep or staying asleep, leading to symptoms of insomnia. Thus, insomnia and sleep apnea can coexist in the same individual, a condition sometimes referred to as “comorbid insomnia.” Treating sleep apnea, often with devices like continuous positive airway pressure (CPAP) machines, can help alleviate the insomnia in these cases. | Sleep apnea is when your breathing stops and starts again while you’re asleep. If you have sleep apnea, you might wake up a lot during the night and then feel really tired during the day. So, while insomnia is about having trouble falling or staying asleep, sleep apnea can cause these problems by disturbing your sleep with those breathing pauses. |
|
|
14. What is the association of insomnia and restless legs syndrome? | Restless legs syndrome (RLS) is characterized by an uncomfortable sensation in the legs that gets worse during periods of inactivity and often disrupts sleep. Those with RLS may have difficulty falling asleep due to the compelling need to move their legs. As such, insomnia is a common complaint among individuals with RLS. | Restless legs syndrome (RLS) gives you an urge to move your legs, especially when you’re lying down. This can make it hard to fall asleep. So, someone with RLS might also have symptoms of insomnia because the uncomfortable feeling in their legs keeps them awake. |
|
|
15. What is the first line treatment for sleep insomnia? | Cognitive Behavioral Therapy for Insomnia (CBT-I) is often considered the first line treatment for chronic insomnia. It addresses the behaviors and thought patterns that contribute to insomnia and teaches individuals strategies to improve their sleep. | A treatment called Cognitive Behavioral Therapy for Insomnia (CBT-I) is the top recommendation. Instead of taking medicine, you’d work with a therapist to learn better sleep habits and ways to handle things that keep you awake, like stress or worry. | Cognitive Behavioral Therapy for Insomnia (CBT-I): CBT-I is the recommended first-line treatment. It addresses the behaviors and thought processes contributing to insomnia and typically involves multiple components, including stimulus control, sleep restriction, and cognitive restructuring. |
|
16. What medications can help treat sleep insomnia? |
|
|
|
|
17. What lifestyle changes can help treat sleep insomnia? |
|
|
|
|
18. What are the types of therapies that help treat sleep insomnia? |
|
|
|
|
19. What can I do to prevent development of sleep insomnia? | Maintaining good sleep hygiene is key. This includes having a regular sleep schedule, creating a bedtime routine, ensuring a dark, quiet, and cool sleep environment, and being cautious with caffeine and alcohol, especially in the evening. Managing stress through relaxation techniques can also be preventive. | Keep a consistent sleep schedule, practice relaxation techniques, avoid caffeine and alcohol close to bedtime, create a comfortable sleep environment, and manage stress. Regular exercise can also help, but try not to do it too close to bedtime. | Adhering to the principles of good sleep hygiene, managing stress, maintaining regular physical activity, avoiding chronic use of alcohol or sedative medications for sleep, and addressing any underlying mental health conditions can all play a preventative role. |
|
20. When do I need to see a sleep medicine specialist? | Consider seeing a sleep specialist if:
|
If you’ve tried several ways to improve your sleep and nothing’s worked, it might be time. Also, if you have other symptoms like snoring really loud (which could be sleep apnea) or uncontrollable leg movements (like RLS), you should see a specialist. They’re experts in sleep problems and can give you a more in-depth evaluation. |
When primary treatments (such as CBT-I or basic pharmacotherapy) are ineffective.
|
If symptoms persist despite trying lifestyle modifications, if sleep disturbances interfere with daily activities, or if there are symptoms suggestive of another sleep disorder (eg, sleep apnea or RLS), then seeing a specialist would be warranted. This decision should be made in collaboration with a primary care physician. |
For assessing readability, the Flesch–Kincaid (FK) grade-level score was calculated for each generated response. An analysis of variance and Tukey’s multiple comparisons test were conducted to analyze differences in grade level across the various prompts. Interrater reliability, a statistical measure of consistency, was quantified using Fleiss’s kappa with predefined thresholds to categorize the strength of agreement among evaluators: less than 0.20 indicated poor agreement, 0.21–0.40 denoted fair agreement, 0.41–0.60 represented moderate agreement, 0.61–0.80 signified good agreement, and 0.81–1.00 equated to very good agreement. Statistical analyses were completed with the Statistical Package for the Social Sciences 28.0 (IBM Corp., Armonk, New York).
RESULTS
The Fleiss kappa scores across the various prompts for clinical accuracy and relevance were poor, calculated at less than 0.20. Nevertheless, evaluators uniformly attested to the high clinical accuracy of ChatGPT’s responses across the first 3 forms (> 85%), with Form 4 exhibiting a broad spectrum of accuracy (65–100%). Relevance to the prompts was consistently high across all forms. Notably, 80% of the cited references were both real and pertinent to the given responses, whereas only 25% of the referenced statistics were directly corroborated by the articles cited (Table 3). Figure 1 provides heat maps to visually represent the evaluators’ scoring for the responses generated by ChatGPT. The mean FK grade-level scores across the prompts were 13.2 ± 2.2 (unprompted), 8.1 ± 1.9 (patient-centered), 15.4 ± 2.8 (physician-centered), and 17.3 ± 2.3 (with reference and statistics). Analysis of variance revealed a statistically significant difference in mean FK grade-level scores between the groups (P < .001). The post hoc analysis revealed significant differences between various pairings: unprompted and patient-centered (P < .001), unprompted and physician-centered (P = .015), and unprompted and reference and statistics (P < .001), as well as between patient-centered and physician-centered (P < .001), patient-centered and reference and statistics (P < .001), and physician-centered and reference and statistics (P = .049).
Table 3.
Readability and interrater reliability scores.
Prompt 1 | Prompt 2 | Prompt 3 | Prompt 4 | |
---|---|---|---|---|
Average Flesch–Kincaid score | 13.2 ± 2.2 | 8.1 ± 1.9 | 15.4 ± 2.8 | 17.3 ± 2.3 |
Clinical accuracya | 0.141 (0.002 to 0.279) | 0.185 (0.046 to 0.323) | 0.141 (0.002 to 0.279) | 0.176 (0.038 to 0.315) |
Relevancea | — | −0.020 (−0.159 to 0.118) | N/Ab | N/Ab |
Percent agreement (CA) | 97% | 92% | 97% | 85% |
Percent agreement (relevance) | — | 98% | 100% | 100% |
Evaluator 1 grading (CA) | 100% | 95% | 95% | 100% |
Evaluator 2 grading (CA) | 95% | 95% | 100% | 100% |
Evaluator 3 grading (CA) | 95% | 90% | 95% | 85% |
Evaluator 4 grading (CA) | 100% | 85% | 95% | 65% |
Evaluator 5 grading (CA) | 95% | 95% | 100% | 75% |
Fleiss kappa (95% confidence interval).
All ratings were the same, so no Fleiss kappa could be calculated. CA = clinical accuracy.
Figure 1. Sleep insomnia query scoring heat maps.
A heat map displaying the assessment by the reviewers of ChatGPT’s responses. Red cell = 0 (did not meet criteria). Green cell = 1 (met criteria). CA = clinically accurate, F1 = Form 1, F2 = Form 2, F3 = Form 3, F4 = Form 4, R = relevance, Ref = reference, Stats = statistics.
DISCUSSION
In the current study, we assessed ChatGPT’s capability to provide factually correct answers to general insomnia queries related to epidemiology, complications, and treatment options. We found that ChatGPT generates accurate information, somewhat influenced variably by the specificity of prompts provided. The nature of prompts significantly influenced the scientific depth of the responses, as discerned by significant differences in the FK grade-level scores. Greater FK scores denote an advanced educational experience required to comprehend the responses, which could limit its understandability to the general population.
ChatGPT’s ability to cite reputable references was deemed satisfactory, albeit with a caveat: All the sources referenced were predated to 2015 or earlier. This aligns with the model’s training on data sets available until 2021, which restricts its capacity to deliver the latest information. Furthermore, ChatGPT exhibited limitations in presenting statistical data, attributable to its inability to process online journal articles. This necessitates caution among users when using ChatGPT for academic or scientific purposes.
The emergent domain of artificial intelligence has introduced large language models such as ChatGPT into an array of sectors, including health care education, clinical practice, and research. Despite their broad-ranging uses—such as conducting literature reviews, enhancing language to convey research, generating efficient discharge summaries, improving health literacy, and creating clinical vignetters—their inherent limitations cannot be overlooked.9–13 These include the generation of incorrect information and citations, lack of up-to-the-minute knowledge, and notable issues with transparency and bias.10,14,15
Although limited for academic use, large language models, such as ChatGPT, offer promise for patients’ education. These models can distill complex medical terminologies into accessible language, thus facilitating patients’ comprehension of their health conditions and treatment plans.12,16–18 This is reflected in our study, where ChatGPT, when prompted, was able to generate patient-friendly answers that were readable at the 8th-grade level. Moreover, the 24/7 availability of ChatGPT to provide tailored responses makes it a consistent, readily accessible source of information. However, it should be noted that whereas ChatGPT can be a valuable adjunct in patient education it should not replace the direct, empathic human communication and interaction that is paramount in health care.15 As such, it is crucial to ensure the information provided by the model is accurate, comprehensible, and appropriately validated by health care professionals.
This study is not without its limitations. Foremost, ChatGPT is not officially approved for medical applications and depends on potentially outdated training data, demonstrated by the studies it referenced dating back to 2015 or earlier. Additionally, the model’s capacity to access current literature is contingent upon developers’ indexing, inhibiting its ability to remain up-to-date.19 Finally, potential biases could be introduced due to the application of a subjective, unverified grading scale. Future research endeavors should consider employing a patient-focused methodology, wherein the potential usage of ChatGPT by patients is explored along with their query preferences. Upon collating a comprehensive list of questions, it would be beneficial to conduct a multicenter interrater reliability study to critically assess the clinical accuracy of ChatGPT’s responses. This approach not only aids in augmenting the validity of findings but also provides a broader, more robust understanding of the language model’s potential role in patient education.
CONCLUSIONS
The results of this investigation affirm that ChatGPT possesses the capacity to generate clinically pertinent and accurate information in response to prevalent insomnia-related inquiries. However, the ability of the language model to draw upon contemporary references and accurately extrapolate statistics from those sources remains a considerable challenge. With further training and software updates, artificial intelligence platforms could potentially augment patients’ education and optimize clinical communication, thus opening a promising vista in digital health care.
DISCLOSURE STATEMENT
All authors have seen, read, and approved the manuscript. The authors report no conflicts of interest.
ABBREVIATIONS
- FK
Flesch–Kincaid
REFERENCES
- 1. Bhaskar S , Hemavathy D , Prasad S . Prevalence of chronic insomnia in adult patients and its correlation with medical comorbidities . J Family Med Prim Care. 2016. ; 5 ( 4 ): 780 – 784 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Taylor DJ , Mallory LJ , Lichstein KL , Durrence HH , Riedel BW , Bush AJ . Comorbidity of chronic insomnia with medical problems . Sleep. 2007. ; 30 ( 2 ): 213 – 218 . [DOI] [PubMed] [Google Scholar]
- 3. American Psychiatric Association . Diagnostic and Statistical Manual of Mental Disorders. 5th ed . Arlington, VA: : American Psychiatric Publishing; ; 2013. . [Google Scholar]
- 4. American Academy of Sleep Medicine . International Classification of Sleep Disorders. 3rd ed . Darien, IL: : American Academy of Sleep Medicine; ; 2014. . [Google Scholar]
- 5. Ayoub NF , Lee Y-J , Grimm D , Balakrishnan K . Comparison between ChatGPT and Google Search as sources of postoperative patient instructions . JAMA Otolaryngol Head Neck Surg. 2023. ; 149 ( 6 ): 556 – 558 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Mesko B . The ChatGPT (generative artificial intelligence) revolution has made artificial intelligence approachable for medical professionals . J Med Internet Res. 2023. ; 25 : e48392 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Stokel-Walker C . AI bot ChatGPT writes smart essays – should professors worry? [published online ahead of print, 2022 Dec 9]. Nature. doi: 10.1038/d41586-022-04397-7 [DOI] [PubMed] [Google Scholar]
- 8. Open AI . Introducing ChatGPT. https://openai.com/blog/chatgpt . Published 2023. . Accessed May 2, 2023.
- 9. Sallam M . ChatGPT utility in healthcare education, research, and practice: systematic review on the promising perspectives and valid concerns . Healthcare. 2023. ; 11 ( 6 ): 887 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Stokel-Walker C , Van Noorden R . What ChatGPT and generative AI mean for science . Nature. 2023. ; 614 ( 7947 ): 214 – 216 . [DOI] [PubMed] [Google Scholar]
- 11. Kitamura FC . ChatGPT is shaping the future of medical writing but still requires human judgment . Radiology. 2023. ; 307 ( 2 ): e230171 . [DOI] [PubMed] [Google Scholar]
- 12. Patel SB , Lam K . ChatGPT: the future of discharge summaries? Lancet Digit Health. 2023. ; 5 ( 3 ): e107 – e108 . [DOI] [PubMed] [Google Scholar]
- 13. D’Amico RS , White TG , Shah HA , Langer DJ . I asked a ChatGPT to write an editorial about how we can incorporate chatbots into neurosurgical research and patient care… . Neurosurgery. 2023. ; 92 ( 4 ): 663 – 664 . [DOI] [PubMed] [Google Scholar]
- 14. Kim S-G . Using ChatGPT for language editing in scientific articles . Maxillofac Plast Reconstr Surg. 2023. ; 45 ( 1 ): 13 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Holzinger A , Keiblinger K , Holub P , Zatloukal K , Müller H . AI for life: trends in artificial intelligence for biotechnology . N Biotechnol. 2023. ; 74 : 16 – 24 . [DOI] [PubMed] [Google Scholar]
- 16. Benoit JRA . ChatGPT for clinical vignette generation, revision, and evaluation . medRxiv. Preprint posted online February 8, 2023. doi: 10.1101/2023.02.04.23285478 [DOI]
- 17. Lee TC , Staller K , Botoman V , Pathipati MP , Varma S , Kuo B . ChatGPT answers common patient questions about colonoscopy . Gastroenterology. 2023. ; 165 ( 2 ): 509 – 11.e7 . [DOI] [PubMed] [Google Scholar]
- 18. Campbell DJ , Estephan LE , Mastrolonardo EV , Amin DR , Huntley CT , Boon MS . Evaluating ChatGPT responses on obstructive sleep apnea for patient education . J Clin Sleep Med. 2023. ; 19 ( 12 ): 1989 – 1995 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Open AI . Models: GPT-3. https://beta.openai.com/docs/models .