POTENTIAL CONTRIBUTION OF ChatGPT^® TO LEARNING ABOUT SEPTIC SHOCK IN INTENSIVE CARE

POTENCIAL APORTE DE APLICAR ChatGPT^® EN LA ENSEÑANZA DE SHOCK SÉPTICO EN CUIDADOS INTENSIVOS

ABSTRACT

Objective:

to demonstrate the application of some prompts and to problematize the use of ChatGPT^® to guide the best answers for nursing students and teachers on septic shock in intensive care learning.

Method:

a methodological study where prompt technology was applied in ChatGPT^® to support nursing learning in intensive care with an emphasis on septic shock. The study was organized in 3 stages, covering an understanding of ChatGPT^® and models, as well as testing and exercising prompts.

Results:

applications of prompts were presented, based on a structure of pre-defined stages that made it possible to exemplify the answers given and to organize an output generation diagram as a way of summarizing the process of decision support in intensive care.

Conclusion:

ChatGPT^® is a natural language processing model that uses deep learning approaches to generate human-like answers. However, the generation of prompts for the teaching-learning process in intensive care nursing requires in-depth association with the pillars of evidence-based practice.

DESCRIPTORS:
Artificial Intelligence; Critical care; Intensive care units; Nursing; Information technology; Septic shock

RESUMEN

Objetivo:

demostrar la aplicación de algunos prompts y debatir cómo se utiliza ChatGPT^® para orientar las mejores respuestas a estudiantes y profesores de Enfermería sobre shock séptico en la enseñanza de cuidados intensivos.

Método:

estudio metodológico en el que se aplicó la tecnología de prompts en ChatGPT^® para respaldar la enseñanza de Enfermería en cuidados intensivos con énfasis en shock séptico. El estudio se organizó en 3 etapas, contemplando lo que se sabe sobre ChatGPT^® y diversos modelos, al igual que pruebas y prácticas con prompts.

Resultados:

Se presentaron aplicaciones de prompts a partir de una estructura de etapas predefinidas que permitieron ejemplificar respuestas dadas por el sistema y organizar un diagrama de generación de salidas como una manera de resumir el proceso de apoyo a las decisiones que se toman en cuidados intensivos.

Conclusión:

ChatGPT^® es un modelo de procesamiento del lenguaje natural que utiliza enfoques de aprendizaje profundo para generar respuestas semejantes a las humanas. Sin embargo, la generación de los prompts para el proceso de enseñanza-aprendizaje en Enfermería de cuidados intensivos requiere una profunda asociación con los pilares de la práctica basada en evidencias.

DESCRIPTORES:
Inteligencia Artificial; Cuidados críticos; Unidades de cuidados intensivos; Enfermería; Tecnología de la información; Shock séptico

RESUMO

Objetivo:

Demostrar a aplicação de alguns prompts e problematizar o uso do ChatGPT^® para guiar as melhores respostas aos estudantes e professores de enfermagem sobre choque séptico na aprendizagem em terapia intensiva.

Método:

Estudo metodológico, com aplicação de tecnologia de prompts no ChatGPT^® para apoiar a aprendizagem de enfermagem em terapia intensiva com ênfase no choque séptico. O estudo foi organizado em 3 etapas, contemplando o entendimento sobre o ChatGPT^®, modelos, bem como teste e exercício de prompts.

Resultados:

Foram apresentadas aplicações de prompts a partir de uma estrutura de etapas pré-definidas que permitiram exemplificar respostas dadas e organizar um diagrama de geração de saídas como uma forma de resumir o processo de apoio à tomada de decisão em terapia intensiva.

Conclusão:

O ChatGPT^® é um modelo de processamento de linguagem natural que usa abordagens de aprendizagem profunda para gerar respostas semelhantes às humanas. Contudo, a geração dos prompts para o processo de ensino-aprendizagem em enfermagem de terapia intensiva requer associação profunda com os pilares da prática baseada em evidência.

DESCRITORES:
Inteligência artificial; Cuidados críticos; Unidades de terapia intensiva; Enfermagem; Tecnologia da informação; Choque Séptico

INTRODUCTION

Artificial Intelligence (AI) is a multidisciplinary approach to computer science and linguistics that uses technology to simulate human cognitive and behavioral processes on computers. Computers are designed to show comprehension, reasoning and problem-solving skills similar to those of humans through specific coding. In addition, these systems can operate in areas that normally require intelligence (e.g. working on issues such as language perception, visual object recognition and decision making), using learning, adaptation, reasoning and understanding capabilities, parallel to cognition and complex human attributes such as attention, emotion, creativity, among others1^-2.

AI-based models are progressively being used in health services and these applications will become even more widespread in the future. In this age of transformation in which technology is advancing rapidly, we hope that nurses will recognize AI models and be able to use this technology in nursing practices as well as increase their levels of knowledge and skills. As health services advance technologically, discussing potential advantages and disadvantages of AI-based technology, health care will be made possible by recognizing and participating in the development and use of these technologies.

In this AI universe, there are generative artificial intelligences, classified as generative adversarial neural networks (GANs), which have the capability of learning and generating new data, called generative and discriminative networks. These AIs learn from large databases so that they can acquire the construction pattern of this data. With this acquired understanding, they can generate new data3.

Generative AI thus describes algorithms (such as ChatGPT^®) that can be used to create new content, including audio, code, images, text, simulations and videos. Recent advances in this field show the potential to change the way we approach content creation. It is important to note that ChatGPT^® is not only an easy-to-use computer algorithm that produces short answers, but an intelligent system that can produce entire documents, essays or even computer codes4.

Thus, as an essential part of today's health care, nursing professionals need to adapt to the rapid advances in AI in order to provide efficient, person-centered care, with health accessible to all. An example of this is the Generative Chat Pre-Trained Transformer (ChatGPT^®; launched on 11/30/2022), an open-access platform with possibilities, such as AI-supported chatbots, that will impact education in clinical practice and scientific documents in the health area5^-7.

ChatGPT^® is an AI-based broad language model (LLM-Master of Laws), trained on massive text datasets in multiple languages and capable of generating human-like answers from text input)8. It was developed by OpenAI (OpenAI, L.L.C., San Francisco, CA, USA). The etymology of ChatGPT^® is related to a chatbot (a program capable of understanding and generating answers using a text-based interface) that is based on the architecture of a pre-trained generative transformer (GPT)8^-9.

The GPT architecture uses a neural network to process natural language, thus generating answers based on the text entered into the platform. When compared to its GPT-based predecessors, ChatGPT^®'s superiority can be linked to its ability to answer in multiple languages, generating refined and highly sophisticated answers based on advanced modeling. It has been trained on a large amount of data, allowing it to generate coherent and contextually appropriate answers to a wide range of prompts. With its advanced natural language processing capabilities, ChatGPT^® can perform tasks such as language translation, summarization and answering questions8^-9.

In this sense, AI-powered chatbots can improve training by offering personalized learning opportunities to students and professionals (e.g. possibilities for clinical assessment of patients to make a safer decision) and the experience of future intensive care nurses. However, the integration and adaptation of chatbots into intensive care nursing practice will require a careful and conscious approach to these issues by nurses.. Nurses must use technologies, such as chatbots, correctly and effectively, considering their potential benefits, limitations and risks in order to offer the best care to patients.

Therefore, for ChatGPT^® to enhance intensive care nursing training and practice, taking into account the complex demands of intensive care patients, good questions need to be asked to promote better decision-making for increasingly safe care. Therefore, the aim of this study was to demonstrate the application of some prompts and to problematize the use of ChatGPT^® to guide the best answers to nursing students and teachers about septic shock in intensive care learning.

METHOD

This methodological research study sought to develop a new instrument, method, procedure, product, program, research tool, theory or model. It was also used to validate and check the reliability of the instruments to measure constructs used as variables in the research10.

As the study aimed at new research opportunities based on artificial intelligence, we opted for methodological research with the application of technology involving the Graduate Program in Health Informatics (professional modality). Considering the complexity of the topic of septic shock in the context of nursing training, five professors (with nursing training and experience in intensive care) from the program mentioned above were the participants.

The methodological path was structured in three stages: understanding the structure and operation of ChatGPT^®; writing model prompts and testing the prompts.

The first stage was aimed at understanding the structure and operation of ChatGPT^®. From delving into the literature and understanding prompt engineering and generative AI, prompt was defined as a question, set of instructions or statement used to initiate or guide a task or conversation. When it comes to language processing, a prompt is an input that the model uses to give an answer or output. Thus, prompt engineering is the process of designing and optimizing questions for AI language models (as used in GPT-4).

The quality and effectiveness of the prompts used to train these models can significantly affect their performance and their ability to generate accurate and useful outputs.. In prompt engineering, the goal is to create prompts that effectively convey the model's inputs and outputs, minimizing ambiguity, noise and other factors that can reduce the model's accuracy or effectiveness. This involves selecting appropriate input formats and defining the expected output format, taking into account any restrictions or limitations that may affect the model's performance11^-12.

GPT-3 is unimodal, i.e. it only accepts text entries. It can process and generate various forms of text (with formal and informal language), but it cannot handle images or other types of data. In contrast, GPT-4 is multimodal. It can accept and produce both text and image inputs and outputs, making it more diverse. According to OpenAI, the latest version of Chat GPT is more likely (40.0%) to produce accurate answers and less likely (82.0%) to respond to requests for prohibited content than GPT-3 Chat. Users can feel safer employing GPT-4 Chat because its AI is much less likely to respond to harmful or inappropriate queries11^-12.

Some model prompts were written in the second stage. At their most basic level, both OpenAI's GPT-3 and GPT-4 predict text based on input (prompt). To get the best results, it was then necessary to write clear prompts with a broad context.. At this stage, we chose to address septic shock in Intensive Care Nursing because of the authors' experience in the field and because it is a complex and challenging subject for nursing care. Thus, after a thorough grounding in the literature (and various discussions), it was possible to delimit the scope of the questions and carry out the tests to refine the questions in the ChatGPT^® model. This process led to a consensus that the prompt structures performed best with clear, objective questions in English.

The prompts were tested and practiced in the third stage. As the prompts on orientation, care and septic shock were tested and practiced, it was possible to identify the specificity needed to construct each question to meet the proposed objective, thus broadening the learning process and filling in gaps. In this stage, teaching and professional experience were key to obtaining better output results. After practicing this technology for a few hours, it was possible to organize suggestions for students to write a GPT-3 or GPT-4 prompt.

In this sense, we emphasize that prompt engineering is the science and art of creating effective inputs to guide an AI model and generate the desired output. In this study, the prompts (inputs) were the clinical questions; the better the prompts, the better the results. We would emphasize that the process of generating prompts and verifying answers (outputs) was supported by consolidated scientific evidence, considering the time limitations of the tool itself.

RESULTS

Using ChatGPT^® and similar tools implies risks, as answers can be generated from unreliable sources or provide incomplete sources and the consequences usually fall on the users. Therefore, all information needs to be verified with rigorous evidence, especially in the area of health. Despite the limitations of this technology, it is fast becoming an indispensable ally, which brings us to the heart of the following questions: How to ask the best questions in the intensive care unit? What are the results of the method used?

Prompt engineering is fundamental; its understanding is necessary for the training of nurses and should be added to the curriculum. Mastery of prompt engineering (as with any skill) comes as much from theoretical study of the subject matter as from practical experience. To maximize time and reduce trial and error, some examples are provided below for learning about septic shock in intensive care using ChatGPT^®13.

1. Be as specific as possible: The more specific your prompt, the more precise and focused the answer will be. In this study, prompts in English were used, but they can be written in any language (Figures 1 and 2).

Figure 1 -
Example of less specific prompts: “Tell me about septic shock in the ICU”, 2023.

Figure 2 -
Example of more specific prompts: “What are the main hemodynamic care for a patient facing septic shock?”

In this example, a specific area of intensive care nursing is covered. However, a more general question can be asked (in any area of nursing), narrowing it down to the specific point to be addressed. For example: What are the common challenges faced by nursing students in their training process and what types of support can be provided to overcome these difficulties?

It is important to note that the information (general or specific) provided by ChatGPT^® may vary depending on the individual characteristics of the patient and the protocols of the health institution. We emphasize that close collaboration between members of a multidisciplinary team of health professionals and the support of scientific evidence are crucial to optimizing results in critically ill patients.

2. Describe your objective: Describe exactly what kind of output (answer) you are looking for (Figure 3). Example of a prompt: I would like to have a short list of at least five ideas for a YouTube video about the future of septic shock treatment and prevention.

Figure 3 -
Example of the type of output (answer) obtained after applying the above prompt with emphasis on the objective, 2023.

3. Describe your objective and give the context: Consider your conversation with ChatGPT^® as the one you would have with a professional you have just met who can answer your questions and help you face the challenges of intensive care nursing. The better the description of the purpose and context you would like to use, the more appropriate the result will be (Figure 4). Example of a prompt: I am writing a manuscript about the main nursing care for patients with septic shock in the ICU based on the use of artificial intelligence. You can list some precautions with specific results. Nurses and nursing students, who are outside the area and are still learning, is my audience. Please, use a friendly and approachable tone.

Figure 4 -
Example of the type of output (answer) obtained after applying the prompt with emphasis on configuration and context, 2023.

4. Experiment with different prompt styles: The style of your prompt can significantly affect the answer given. Try out different formats, such as generating a list, providing a summary, main ideas or giving the characteristics of your audience, desired roles, among others. (Figures 5, 6, 7, 8 and 9):

Figure 5 -
Example of prompts applied according to the use of a direct question: What is the nursing care based on the ICNP (International Classification Nursing Practice) for patients in Septic shock?, 2023.

Figure 6 -
Example of prompts applied according to list usage: List all potential alerts for patients with suspected Septic shock, 2023.

Figure 7 -
Example of prompts applied according to summary usage: Summarize the key symptoms and progression of Septic shock, 2023.

Figure 8 -
Example of prompts applied according to the use of Target audience (characteristics): Explain the consequences of a patient's septic shock to his/her family, 2023.

Figure 9 -
Example of prompts applied according to role suggestions: Act like a scientist and explain to the nurses why controlling C-reactive protein in septic shock is important, 2023.

5. Repeat and refine your questions to dig deeper and/or get better answers: It should be noted that the best results are rarely obtained immediately after the first prompt. Therefore, refine your questions, just like when you organize a content test or teach a class in which you look for the best answers indicating that the content has been understood. If you still fail to get the output you want, try guiding the model with continuation prompts. This can be more productive than hoping to get the answer you want with a single prompt. It is also possible to start a conversation like this: ask the AI to consider each step or present the pros and cons before you decide on an answer. The examples shown above can help you better understand this topic of repeating and refining questions.

In addition, if you are searching in unfamiliar territory and looking for information in a field where you are unfamiliar, feedback loops can be used to go into detail. Although ChatGPT^® initially gives generic answers, it is always possible to use the output as input for subsequent prompts, generating a cycle of problems and answers.

6. Use your previous topics: In ChatGPT^® you can return to a specific discussion by clicking on the topic in the left-hand column where the prompts are stored. Thus, you do not need to start again and can simply continue the discussion with ChatGPT^®.

7. Ask open and closed questions: open questions generally produce broader answers, while closed questions produce more specific answers.

8. Request examples: if the answer is unsatisfactory or incomprehensible, say that you didn't understand it; then ask for an example or ask to improve or regenerate the answer.

9. Use your time wisely: If you are asking about a process or schedule, specify this in your prompt (Figures 10 and 11).

Figure 10 -
Example of prompts applied any without time reference: Describe the healing process after a septic shock, 2023.

Figure 11 -
Example of prompts applied with a time reference: What can a patient typically expect during the first six weeks of healing after suffering a septic shock?, 2023.

10. Define realistic expectations: Although GPT-4 is a powerful tool, it does have its limitations. For example: it doesn't allow access to real-time data (although you can adjust this with plugins); there's an end date (2021) which may not be a problem soon)8; references are not made explicit (answers then need to be checked) and it doesn't give personal advice or replace the judgment of a health professional.

Based on the examples above, a diagram of output generation from prompts has been structured. The diagram was designed to summarize the method or process of generating prompts and outputs to support decision-making in the practice of intensive care for the teaching-learning process of septic shock (Figure 12).

Figure 12 -
Diagram corresponding to the process for generating prompts and outputs to support decision-making in intensive care.

The diagram above can serve as a reference to support teachers and professionals in developing new prompts in various areas of practice and learning for nurses. From the perspective of caring for patients with septic shock, we would like to highlight the important contribution of ChatGPT^®, which enables both students and professionals to exercise clinical reasoning based on a structure of logically well-formulated questions and supported by the four pillars of evidence-based practice (research evidence, practice, patient and context).

DISCUSSION

In the scientific and academic community, ChatGPT^® received mixed responses, reflecting the controversies over the benefits versus risks of advanced AI technologies. On the one hand, ChatGPT^® (among other Large Language Models, LLMs) can be beneficial in dialog and writing tasks, helping to increase the efficiency and accuracy of the required output. On the other hand, concerns have been raised that require more supervision and investment in AI exit detectors. This is necessary to deal with possible errors and biases in its output (which can limit its capabilities and result in inaccuracies) based on the data set used to train ChatGPT^®. In addition, security concerns and potential cyber-attacks spreading disinformation via LLMs should also be considered14.

In the literature, ChatGPT^® has been approached as an evolved form of search engine, in the sense that it can suggest themes, questions and research objectives by relating them to the initial question1. The experience gained during the preparation of the manuscript showed that ChatGPT^® can be a component of the teaching and learning system as a tutor as well as in self-learning in nursing.

By practicing the prompts for learning about septic shock, teachers and students manage their time better by obtaining quick and accurate information. By incorporating the information obtained and seeking confirmation and further study in the scientific literature, there was a stimulus to research, study and clinical reasoning. This process must be incorporated into teaching so that artificial intelligence can contribute more and more to teaching and learning15.

ChatGPT^® can therefore be used as a tool to support decision-making in the learning process. It has the potential to help research and accelerate the technological transformation of clinical and translational nursing, as well as discovering new care for different drugs in development, predicting diseases and complications, diagnosing and assessing before health problems set in16.

Generative AI such as ChatGPT^® can stimulate the development of digital literacy, a core competence in nursing informatics, encouraging critical thinking about the integration of AI into health care. In addition, it can offer valuable support to students and teachers, improving the quality of writing, assisting in research and teaching tasks (organizing, summarizing and simplifying ideas) and helping to interpret data17.

It is fundamental to ensure the privacy and security of confidential patient information, as is the case with any digital health technology when applied directly to care. Sending confidential information to ChatGPT^® can have serious consequences. Health organizations that intend to implement models such as ChatGPT^® must have comprehensive guidelines for using such tools to handle patient data and implement measures to protect data privacy, such as anonymizing identifiable data, encryption, compliance with national health regulations, among others18.

Users of some recently developed AI-based software claim to take advantage of ChatGPT^®'s cutting-edge features to meet data security and privacy requirements, although the effectiveness of these tools in a clinical environment requires a rigorous validation and quality control process1^,16. Interdisciplinary collaboration is needed between AI developers, health professionals, policymakers and data security experts. This reflection should also be observed in the professional training process, according to the possibilities presented here.

Integrating ChatGPT^® with effective prompt generation techniques can simplify health documentation, but must be approached with care to manage ethical challenges and prevent harms. The adoption of this technology can generate benefits in health documentation, the teaching process and decision-making by health professionals, improving their productivity and patient care. Guidelines for the use of AI tools in the documentation of patient data will be an important step. However, research efforts will be needed to investigate the effectiveness of these tools compared to existing methods19.

Among other aspects, we highlight the importance of learning for a robust, dynamic and progressive nursing experience based on evidence. This requires curricula that are flexible and attentive to the global AI movement in teaching and care processes, ensuring nurses' place in care environments and improving patient outcomes20.

Empathetic communication is the basis of the nurse-patient relationship and over-reliance on chatbots can lead to a reduction in empathy among nurses. For example, providing pre-prepared answers to nurse-patient dialogues can make these interactions more impersonal and less therapeutic. Additionally, chatbot answers may not be reliable or evidence-based20. The OpenAI website itself admits that ChatGPT^® can generate false or misleading information and produce offensive or biased content, and recommends caution in its use. However, technologically competent nurses can assess the risk of using this tool considering its limitations. Nurses will continue to be responsible for their clinical decisions, including those made on the basis of the chatbot's answers.

In relation to the incorporation of confidential or personally identifiable data, the ChatGPT^® platform warns users and organizations that the information processed in it is stored provisionally on OpenAI servers and does not guarantee its security. Therefore, nursing professionals must take precautions to protect sensitive or confidential data of patients and people in general, as well as those of the health organization. Security measures can include the use of encryption or the non-disclosure of information online.

Any app programmed to implement ChatGPT^® must also adhere to data protection in accordance with national legislation1. The study limitations include the fact that it is still in a theoretical stage and needs to be applied to students, teachers and nurses in the context of intensive care.

CONCLUSION

The ChatGPT^® discussions and examples were initiated in intensive care applied to septic shock learning, as it is important to understand how this type of artificial intelligence technology works in a specific nursing focus.

In a logical and increasingly complex sequence, ChatGPT^® make learning friendly and dynamic in the care of patients with septic shock and other problems, although it is a recent topic in nursing.

Health decisions are ethical actions and the use of this tool in teaching and care requires constant consultation of safe and up-to-date evidence without forgetting its limitations.

REFERENCES

• 1. Berşe S, Akça K, Dirgar E, Serin EK. The role and potential contributions of the artificial intelligence language model ChatGPT^® Ann Biomed Eng [Internet]. 2023 [cited 2023 Jul 12];52:130-3. Available from: https://doi.org/10.1007/s10439-023-03296-w
  » https://doi.org/10.1007/s10439-023-03296-w
• 2. Sallam M. ChatGPT^® Utility in healthcare education, research, and practice: Systematic review on the promising perspectives and valid concerns. Healthcare [Internet]. 2023 [cited 2023 Jul 12];11(6):887. Available from: https://doi.org/10.3390/healthcare11060887
  » https://doi.org/10.3390/healthcare11060887
• 3. Arora A, Arora A. Generative adversarial networks and synthetic patient data: Current challenges and future perspectives. Future Healthc J [Internet]. 2022 [cited 2023 Jul 12];9(2):190-3. Available from: https://doi.org/10.7861/fhj.2022-0013
  » https://doi.org/10.7861/fhj.2022-0013
• 4. Bajwa J, Munir U, Nori A, Williams B. Artificial intelligence in healthcare: Transforming the practice of medicine. Future Healthc J [Internet]. 2021 [cited 2023 Jul 12];8(2):e188-94. Available from: https://doi.org/10.7861/fhj.2021-0095
  » https://doi.org/10.7861/fhj.2021-0095
• 5. Choi EPH, Lee JJ, Ho MH, Kwok JYY, Lok KYW. Chatting or cheating? The impacts of ChatGPT^® and other artificial intelligence language models on nurse education. Nurse Educ Today [Internet]. 2023 [cited 2023 Jul 12];125:105796. Available from: https://doi.org/10.1016/j.nedt.2023.105796
  » https://doi.org/10.1016/j.nedt.2023.105796
• 6. Haleem A, Javaid M, Singh RP. An era of ChatGPT^® as a significant futuristic support tool: A study on features, abilities, and challenges. TBench [Internet]. 2022 [cited 2023 Jul 10];2(4):100089. Available from: https://doi.org/10.1016/j.tbench.2023.100089
  » https://doi.org/10.1016/j.tbench.2023.100089
• 7. Scerri A, Morin KH. Using chatbots like ChatGPT^® to support nursing practice. J Clin Nurs [Internet]. 2023 [cited 2023 Jul 5];32(15-16):4211-3. Available from: https://doi.org/10.1111/jocn.16677
  » https://doi.org/10.1111/jocn.16677
• 8. OpenAI. OpenAI: Models GPT-3 [Internet]. 2023 [cited 14 Jul 2023]. Available from: https://beta.openai.com/docs/models
  » https://beta.openai.com/docs/models
• 9. Brown T, Mann B, Ryder N, Subbiah M, Kaplan JD, Dhariwal P, et al. Language models are few-shot learners. Adv Neural Inf Process Syst [Internet]. 2020 [cited 2023 Jul 5];33:1877-901. Available from: https://papers.nips.cc/paper_files/paper/2020/file/1457c0d6bfcb4967418bfb8ac142f64a-Paper.pdf
  » https://papers.nips.cc/paper_files/paper/2020/file/1457c0d6bfcb4967418bfb8ac142f64a-Paper.pdf
• 10. Abedellah FG, Levine E. Better patient care through nursing research. New York, NY(US): MacMillan; 1965.
• 11. Adelson M. The ChatGPT^® Goldrush: Profiting from the ai revolution: Prompt Engineering Mastery with ChatGPT^®-4. Prompt Library, List of 200 AI apps; 2023.
• 12. John I. The Art of Asking ChatGPT^® for high-quality answers: A complete guide to prompt engineering techniques. Nzunda Technologies Limited, Library of Congress Control/USA; 2023.
• 13. Ekin S. Prompt Engineering For ChatGPT: A Quick Guide To Techniques, Tips, And Best Practices. TechRxiv [Internet]. Preprint. 2023 [cited 2023 Jul 5]. Available from: https://doi.org/10.36227/techrxiv.22683919.v2
  » https://doi.org/10.36227/techrxiv.22683919.v2
• 14. The Lancet Digital Health. ChatGPT^®: Friend or foe? Lancet Digit Health [Internet]. 2023 [cited 2023 Jul 5];5(3):e102. Available from: https://doi.org/10.1016/S2589-7500(23)00023-7
  » https://doi.org/10.1016/S2589-7500(23)00023-7
• 15. Miao H, Ahn H. Impact of ChatGPT^® on Interdisciplinary Nursing Education and Research.Asian Pac Isl Nurs J [Internet]. 2023 [cited 2023 Sep 22];7:e48136. Available from: https://doi.org/10.2196/48136
  » https://doi.org/10.2196/48136
• 16. Xue VW, Lei P, Cho WC. The potential impact of ChatGPT^® in clinical and translational medicine. Clin Transl Med [Internet]. 2023 [cited 2023 Jul 5];13(3):e1216. Available from: https://doi.org/10.1002/ctm2.1216
  » https://doi.org/10.1002/ctm2.1216
• 17. Alexandre Castonguay A, Farthing P, Davies S, Vogelsang L, Kleib M, Risling T, et al. Revolutionizing nursing education through Ai integration: A reflection on the disruptive impact of ChatGPT. Nurse Educ Today [Internet]. 2023 [cited 2023 Sep 22];129:105916. Available from: https://doi.org/10.1016/j.nedt.2023.105916
  » https://doi.org/10.1016/j.nedt.2023.105916
• 18. Wang C, Liu S, Yang H, Guo J, Wu Y, Liu J. Ethical considerations of using ChatGPT^® in health care. J Med Internet Res [Internet]. 2023 [cited 2023 Sep 22];25:e48009. Available from: https://doi.org/10.2196/48009
  » https://doi.org/10.2196/48009
• 19. Fabrizzio GC, Oliveira LM de, Costa DG, Erdmann AL, Santos JLG. Virtual assistant: A tool for health co-production in coping with covid-19. Texto Contexto Enferm [Internet]. 2023 [cited 2023 Jul 5];32:e20220136. Available from: https://doi.org/10.1590/1980-265X-TCE-2022-0136en
  » https://doi.org/10.1590/1980-265X-TCE-2022-0136en
• 20. Nguyen J, Pepping CA. The application of ChatGPT^® in healthcare progress notes: A commentary from a clinical and research perspective. Clin Transl Med [Internet]. 2023 [cited 2023 Jul 5];13(7):e1324. Available from: https://doi.org/10.1002/ctm2.1324
  » https://doi.org/10.1002/ctm2.1324

NOTES

• TRANSLATED BY

  Leonardo Parachú

Edited by

EDITORS

Associated Editors: Manuela Beatriz Velho, Maria Lígia Bellaguarda. Editor-in-chief: Elisiane Lorenzini.

Publication Dates

• Publication in this collection
  17 May 2024
• Date of issue
  2024

History

• Received
  26 July 2023
• Accepted
  29 Sept 2023