The rapid incorporation of artificial intelligence (AI) in healthcare is reshaping patient care and clinical decision-making dynamics. 1 Language models, such as ChatGPT developed by OpenAI in San Francisco, CA, demonstrate remarkable capacity in analyzing vast datasets to generate responses similar to human communication. 2 As the world gravitates towards digitalization, AI-powered chatbots are garnering attention for their potential in addressing patient inquiries in the medical and surgical sectors, signifying a paradigm shift in global healthcare interactions between patients and providers. 3 Opting for breast reconstruction is a profoundly personal choice, and many patients grapple with a myriad of queries about the procedure, its associated risks, and the expected results. 4 Providing timely, precise, and evidence-grounded information is pivotal to guide informed choices and set patient anticipations. Earlier evaluations have established ChatGPT's ability to offer patient-centric information while minimizing the complexity of medical terminology.

With the growing demand for plastic surgery endeavors, including breast reconstruction, medical professionals confront the imperative of imparting thorough knowledge and backing to a swelling patient populace. 4 AI-enhanced chatbots could serve as an answer to this challenge, presenting a user-friendly platform for patients to gain pertinent knowledge and voice their apprehensions. This could also temper the informational load on clinicians, allowing a more concentrated approach to patient treatment and procedure nuances. 5 Our study delved into ChatGPT's informational depth, responsiveness, and competence in addressing typical inquiries from patients considering breast reconstruction. Mindful of ChatGPT's constraints highlighted in prior research, our analysis critiqued the chatbot's ability to reference accurately and the evidential strength buttressing its replies. 6,7 This examination aims to shed light on the viability of AIempowered chatbots as auxiliary tools for patient enlightenment and endorsement in the realm of plastic surgery.

This study set out to explore the capability of artificial intelligence language models in delivering accurate and current medical information to the general populace. We utilized ChatGPT-4, among the most advanced language models presently available to the public, for this examination. We assessed its proficiency, depth, and precision in addressing prevalent queries posed by individuals regarding breast reconstruction. A panel of four expert plastic surgeons (G.L., F.C., D.M., and R.R.) crafted these questions in alignment with the American Society of Plastic Surgeons (ASPS) website. 8 We ensured that subsequent questions remained unmodified based on preceding responses and did not prompt the system to recreate previously used answers.

We posed six frequently asked questions related to breast reconstruction to ChatGPT. Using subject matter experts and defined criteria, we meticulously examined its response strategy and content, evaluating its accuracy, depth, and user-friendliness. Four experienced plastic and reconstructive surgeons (G.L., F.C., D.M, and R.R.) specializing in breast surgeries critically reviewed the given answers. Additionally, we cross-referenced ChatGPT's responses and citations against two comprehensive medical databases, PubMed and Scopus, to validate their relevance and correctness.

Given that ChatGPT-4 operates on a probabilistic algorithm, leveraging random sampling to craft diverse answers, it may produce varied responses to identical queries. For the purposes of this study, we strictly incorporated the initial answer that ChatGPT delivered for each question, disallowing subsequent amendments or elaborations. We took precautionary steps to ensure that each question posed was free from grammatical or syntactical errors. All questions were input on a singular day through a ChatGPT Plus account, equipped with access to ChatGPT-4, held by one of the contributing authors. No response provided by ChatGPT was excluded from our analysis.

The responses to the 6 questions asked to ChatGPT-4 are analyzed below.

ChatGPT-4 response provides a comprehensive overview of the various breast reconstruction techniques, covering both implant-based and autologous reconstruction (Figure 1).

The information is organized in a clear manner, with distinct sections detailing each reconstruction method. The use of bullet points and subheadings enhances readability. The content appears neutral and does not seem to promote one method over another, which is crucial for unbiased patient education. The concluding remark emphasizing individual differences and the need for consultation with a board-certified surgeon underscores a patientcentered approach, which is in line with modern medical practice. The tone is informative and seems directed at patients or the general public. It does not overwhelm the patient with excessive medical jargon, which makes it accessible. The information provided is correct, referring to the most important aspects of autologous and prosthetic breast reconstruction. 9,12,56 When can the breast reconstruction be done? The information provided by ChatGPT-4 elaborates on the temporal modalities of breast reconstruction post-mastectomy ( Figure 2). The information provided to the patient in this question was succinctly summarized without grammatical errors. The tone is informative and seems directed at patients or the general public. Immediate reconstruction, executed concomitantly with mastectomy, has been observed to confer psychological benefits and might yield superior aesthetic outcomes due to the potential preservation of breast skin. 7 Delayed reconstruction, done weeks to years post-operatively, is often advocated for individuals necessitating postmastectomy radiation, as radiation can induce dermal and subdermal alterations compromising reconstruction quality. 11 An intermediary, delayed-immediate reconstruction, employs a tissue expander post-mastectomy, with the definitive reconstructive procedure ensuing post-radiation. 10 Lastly, staged reconstruction deploys an expander to incrementally stretch the dermal envelope before implantation. 14 ChatGPT-4 underlines the fact that the decision on when to undergo breast reconstruction is highly individual and should be made in consultation with a team of medical professionals, including an oncologic surgeon, a plastic surgeon, an oncologist, and other relevant specialists.

They can provide guidance tailored to the patient's medical situation, aesthetic desires, and lifestyle considerations.

ChatGPT-4 succinctly and correctly addresses the query, discussing risks and benefits of each option ( Figure 3).

The provided information outlines the benefits and risks of two prevalent breast reconstruction methods: implant-based and autologous reconstructions. The distinctions highlight key surgical considerations like invasiveness, recovery times, and aesthetic outcomes. Importantly, the text acknowledges the fact that patient outcomes and experiences with these procedures can vary significantly. 8 A critical nuance missed is the possibility of implant-associated anaplastic large cell lymphoma (BIA-ALCL) with implant-based reconstruction. 9 While flap failures are rightly mentioned for autologous reconstruction, actual statistical rates can provide a clearer perspective. 10 While the recommendation to consult a board-certified plastic surgeon is valid, incorporating multi-disciplinary approaches with oncologists and radiologists is equally crucial. 11

ChatGPT-4 succinctly and correctly addresses the query, discussing the risks of radiotherapy (Figure 4).

The provided information offers a comprehensive overview of the impact of radiation on breast reconstruction. The adverse effects of radiation, such as reduced skin elasticity and fibrotic changes, are well-documented in the literature. 11 The increased risk of capsular contracture post-radiation in implantbased reconstruction is supported by studies demonstrating its incidence and resultant morbidity. 11 The mention of radiation's effect on autologous reconstruction, especially reduced blood flow and potential flap failure, is consistent with findings on vascular complications. 7 The elaboration on the timing of reconstruction-immediate, delayed, and delayed-immediate-is a relevant consideration for optimizing patient outcomes . Concluding the summary with the importance of multidisciplinary collaboration is vital for holistic patient care.

ChatGPT-4 response provides a comprehensive overview of breast reconstruction complications ( Figure 5) The summary elucidates potential complications and considerations following breast reconstruction. For implant-based reconstruction, the lifespan of implants and the risk of capsular contracture are consistent with the current literature. 1213 The phenomenon of implant malposition and rippling is well-documented, particularly in patients with thin skin coverage. 14 In the context of autologous reconstruction, flap failure due to compromised blood supply is a recognized complication, and the mention of donor site issues is supported by research detailing morbidity at these sites. 15 The emphasis on symmetry and nipple reconstruction in the shared section underscores the multidimensional aspects of breast restoration. 16 Highlighting the importance of regular follow-ups is pertinent for optimal patient care.

ChatGPT-4 provides a comprehensive, yet concise, overview of the interplay between breast reconstruction and chemotherapy, stressing the importance of individualized, patient-centered care ( Figure 6)

The presented summary discusses the considerations and interactions between breast reconstruction and chemotherapy. The emphasis on the timing of reconstruction aligns with studies which indicate that immediate post-mastectomy reconstruction may delay adjuvant chemotherapy, potentially influencing oncological outcomes. 17 The mention of complications potentially delaying chemotherapy initiation is crucial, given that complications like infections can indeed postpone chemotherapy. 18 The assertion that chemotherapy can weaken the immune system and thereby impede wound healing is supported by numerous studies. 19

Previously, the authors conducted a study on chatbots in the context of breast augmentation. 20 However, breast reconstruction presents a more complex subject matter for patients, especially from a shared decision-making perspective.

In the realm of breast reconstruction, the decisionmaking process has always been intimately personal for patients. As this study aimed to evaluate ChatGPT's capability to address pertinent patient queries related to the subject, it highlighted a core tenet: the need for collaborative decision-making rather than a solely paternalistic stance. Emphasizing SDM, a patient-centered approach, ensures that healthcare decisions are made in tandem with patients and providers. This collaboration seamlessly melds the best clinical evidence with the patient's personal values and preferences. 21 Breast reconstruction, in particular, necessitates a robust SDM framework given the enduring impact such decisions have on a patient's appearance, functionality, and overall well-being. 22 Pivotal to enhancing SDM are comprehensive patient educational tools, including multimedia resources and decision aids. 23 These aids not only bolster understanding but can also optimize the time efficiency of medical consultations. 24 However, the contemporary healthcare landscape is mired with challenges, notably the soaring demand for medical and surgical services in the face of finite resources. 25 While diversifying patient education methods and relying on auxiliary medical staff or digital resources might seem viable, nothing replaces the foundational trust built through direct doctorpatient interactions. Such rapport has proven to uplift individual post-surgical outcomes 26 and fortify overarching trust in healthcare systems. 27 In this context, ChatGPT emerges as a promising assistant. Its prowess lies in distilling complex medical information into easily digestible responses without using dense medical jargon. While this study acknowledges the strength of language models like ChatGPT, it also highlights a key limitation: a general lack of personalized advice, especially in the context of breast reconstruction. These shortcomings, combined with concerns about academic rigor, underscore the need for caution before fully adopting it as a primary educational tool. 2820 Supporting this observation is ChatGPT's tendency to provide broader, somewhat generalized responses -a pattern also observed in previous studies evaluating AI-driven chatbots across various surgical fields. 29 The challenge lies in aligning ChatGPT's capabilities with the deeply rooted medical philosophy of providing individualized and nuanced care. Without addressing this gap, ChatGPT might risk overshadowing the specialist advice provided by surgeons, which is grounded in their meticulous evaluation of specific patient factors, ranging from surgical techniques to reconstructive decisions. 30 Despite these reservations, the study underlines the undeniable potential of ChatGPT as an auxiliary informational tool in breast reconstruction dialogues. Still, as with any research, this study is not without limitations. The questions formulated by the authors, though extensive, may not cover the entire spectrum of patient queries in every surgical setup. The 2021 knowledge cut-off of ChatGPT is another limitation, restricting the breadth of evidence. Further, the qualitative nature of this study, informed by views from a single institution, could introduce some inherent biases. An important note for future research is the plethora of chatbots beyond ChatGPT. Evaluative comparisons between these platforms might pave the way for more efficacious patient education tools in the future.

The integration of artificial intelligence (AI) tools, such as ChatGPT, into the healthcare sector offers the prospect of reshaping the informational landscape for patients, especially in complex domains like breast reconstruction. Our research sought to evaluate ChatGPT's capacity in responding to common patient queries in this arena. While ChatGPT showcased strengths in simplifying intricate medical concepts and providing clear, evidence-based responses, its propensity to offer generalized rather than personalized guidance remains a tangible limitation. Such a lack of personalization underscores the challenge of harmonizing AI capabilities with the inherent nuances of patient care and the individualized insights provided by medical professionals.

Moreover, while ChatGPT can serve as a beneficial auxiliary tool for patient education, it cannot supplant the crucial doctor-patient relationship, rooted in trust and tailored advice. The fundamental role of shared decision-making, where healthcare decisions are collaboratively made between patients and providers, is only accentuated in domains such as breast reconstruction. Tools like ChatGPT, though promising, need to evolve further before they can seamlessly support this collaborative dynamic without diluting the core essence of individualized care.

This study, while illuminating, also underscores the need for more comprehensive research. As AIdriven tools multiply and evolve, future comparative assessments could help refine their role in healthcare, ensuring that they complement rather than compromise the quality and specificity of patient care.

The process of breast augmentation: four sequential steps for optimizing outcomes for patients Adams Wp Plast Reconstr Surg 2008 A Comparison of Patient-Reported Outcomes After BreastConserving Surgery and Mastectomy with Implant Breast Reconstruction M R Flanagan E C Zabor A Romanoff S Fuzesi M Stempel B J Mehrara Ann Surg Oncol 2019 Comparison of Immediate versus Delayed DIEP Flap Reconstruction in Women Who Require Postmastectomy Radiotherapy R L O'connell Di Micco R Khabra K Kirby A M Harris P A James S E Power K Ramsey Kwd Rusby J E Plast Reconstr Surg 142 3 594 605 2018 Implementing shared decision-making interventions in breast cancer clinical practice: a scoping review N Oprea V Ardito O Ciani BMC Med Inform Decis Mak 23 1 164 164 2023 The Decision Aid is the Easy Part: Workflow Challenges of Shared Decision-Making in Cancer Care M E Salwei J S Ancker M B Weinger J Natl Cancer Inst 133 133 2023 50 322 352 464 469 Less is more? One-stage versus two-stage implant-based breast reconstruction: A systematic review and meta-analysis of comparative studies A Van Der Wielen V Negenborn G L Burchell S Remmelzwaal O Lapid C Driessen J Plast Reconstr Aesthet Surg 86 109 127 2023 Findings of a national comparative audit of mastectomy and breast reconstruction surgery in England R Jeevan D A Cromwell J P Browne C M Caddy J Pereira C Sheppard K Greenaway J H Van Der Meulen J Plast Reconstr Aesthet Surg 67 10 1333 1377 2014 Epidemiology of Breast Implant-Associated Anaplastic Large Cell Lymphoma E L Doren R N Miranda J C Selber P B Garvey J Liu L J Medeiros C E Butler Mw U Clemens Plast Reconstr Surg 139 5 2017 A conversation on artificial intelligence, chatbots, and plagiarism in higher education Chatgpt Kingmr Cell Mol Bioeng 16 1 2023 B D Lund T Wang Chatting about ChatGPT: how may AI and GPT impact academia and libraries? Libr Hi Tech News 40 26 29 2023 Capsular contracture and genetic profile of ica genes among Staphylococcus epidermidis isolates from subclinical periprosthetic infections P Persichetti Giovanni Lombardo G A Marangi G F Gherardi G Dicuonzo G Plast Reconstr Surg 127 4 1747 1748 2011 Nipple-areola reconstruction: satisfaction and clinical determinants M A Jabor P Shayani D R Collins T Karas B E Cohen Plast Reconstr Surg 110 2 457 63 2002 B D Lund T Wang Chatting about ChatGPT: how may AI and GPT impact academia and libraries? Libr Hi Tech News 40 26 29 2023 The donor site morbidity of free DIEP flaps and free TRAM flaps for breast reconstruction N Blondeel G G Vanderstraeten S J Monstrey K Van Landuyt P Tonnard R Lysens Br J Plast Surg 1997 A before and after evaluation of patient reported outcomes assessed by Breast Q following LICAP turnover flap in breast conservative therapy Gag Lombardo A Stivala F Ciancio S Tamburino D' Antonio G M Catalano F Updates Surg Immediate breast reconstruction surgery with expander/direct implant and use of acellular dermal matrix: Does hormone therapy increases the risk of infection? M A Danino D Giaccone El Khatib A Dimitropoulos G Doucet O Gagnon A Ann Chir Plast Esthet 65 4 277 283 2020 Defining Health in the Era of ValueBased Care: The Six Cs of Health and Healthcare M Warren Cureus 9 1046 1046 2017 26 3133 3140 122 1892 1900 ChatGPT goes to the operating room: evaluating GPT-4 performance and its potential in surgical education and training in the era of large language models N Oh G-S Choi W Y Lee Ann Surg Treat Res 104 5 269 273 2023 Evaluating Chatbot Efficacy for Answering Frequently Asked Questions in Plastic Surgery: A ChatGPT Case Study Focused on Breast Augmentation I Seth A Cox Y Xie G Bulloch D J Hunter-Smith W M Rozen R J Ross Aesthet Surg J 43 10 1126 1135 2023 2009 Understanding the impact of breast reconstruction on the surgical decisionmaking process for breast cancer A K Alderman S T Hawley J Waljee M Mujahid M Morrow S J Katz Cancer 112 3 489 94 2008 A patient in the clinic; a person in the world. Why shared decision making needs to center on the person rather than the medical encounter M L Clayman P Gulbrandsen M A Morris Patient Educ Couns 100 600 604 2017 M Sallam The utility of ChatGPT as an example of large language models in healthcare education, research and practice: systematic review on the future perspectives and potential limitations D Henn A Momeni A standardized patient education class as a vehicle to improving shared decision-making and increasing access to breast Reconstr Surg 133 496 506 2014 A conversation with an open artificial intelligence platform on osteoarthritis of the hip and treatment I Seth A Rodwell R Tso J Valles G Bulloch N Seth J Orthop Sports Med 5 112 120 2023 R L O'connell T Rattay R V Dave A Trickey J Skillman Nlp Barnes M Gardiner A Harnett S Potter C Holcombe A Standardized Patient Education Class as a Vehicle to Improving Shared Decision Making and Increasing Access to Breast Reconstruction D Henn A Momeni Journal of Plastic, Reconstructive & Aesthetic Surgery 2020 The Effect of Omega-3 Fatty Acids on Capsular Tissue around the Breast Implants Gag Lombardo S Tamburino K Magano P Fagone S Mammana E Cavalli M S Basile L Salvatorelli F Catalano G Magro F Nicoletti Plast Reconstr Surg 145 3 701 710 2020 Complications in postmastectomy breast reconstruction: two-year results of the Michigan Breast Reconstruction Outcome Study A K Alderman E G Wilkins H M Kim J C Lowery Plast Reconstr Surg 109 7 2265 74 2002 Shared decision making: a model for clinical practice G Elwyn D Frosch R Thomson N Joseph-Williams A Lloyd P Kinnersley E Cording D Tomson C Dodd S Rollnick A Edwards M Barry J Gen Intern Med 27 10 1361 1368 2012 Postoperative psychosocial predictors of outcome in bariatric surgery C S Sheets C M Peat K C Berg Obes Surg 25 330 345 2015 Gag Lombardo D Marrella A Stivala F Ciancio I Musmarra F Catalano Pedicle stabilization with fibrin sealant in DIEP flap breast reconstruction. Updates Surg 2023 Breast Reconstruction Research Collaborative. The impact of immediate breast reconstruction on the time to delivery of adjuvant therapy: the iBRA-2 study Br J Cancer 120 9 883 895 2019 Evaluating AI's Efficacy in Enhancing Patient Education and Answering FAQs in Plastic Surgery: A Focused Case Study on Breast Reconstruction G Lombardo R Ranno D Melita I Musmarra P Marchica F Ciancio Arch Breast Cancer 11 2 147 58 2024