AI and Ethics: A Systematic Review of the Ethical Considerations of Large Language Model Use in Surgery Research

Healthcare, Journal Year: 2024, Volume and Issue: 12(8), P. 825 - 825

Published: April 13, 2024

Introduction: As large language models receive greater attention in medical research, the investigation of ethical considerations is warranted. This review aims to explore surgery literature identify concerns surrounding these artificial intelligence and evaluate how autonomy, beneficence, nonmaleficence, justice are represented within discussions provide insights order guide further research practice. Methods: A systematic was conducted accordance with Preferred Reporting Items for Systematic Reviews Meta-Analyses guidelines. Five electronic databases were searched October 2023. Eligible studies included surgery-related articles that focused on contained adequate discussion. Study details, including specialty concerns, collected. Results: The search yielded 1179 articles, 53 meeting inclusion criteria. Plastic surgery, orthopedic neurosurgery most surgical specialties. Autonomy explicitly cited principle. frequently discussed concern accuracy (n = 45, 84.9%), followed by bias, patient confidentiality, responsibility. Conclusion: implications using complex evolving. integration into necessitates continuous discourse ensure responsible use, balancing technological advancement human dignity safety.

Language: Английский

---

Accuracy, readability, and understandability of large language models for prostate cancer information to the public

Prostate Cancer and Prostatic Diseases, Journal Year: 2024, Volume and Issue: unknown

Published: May 14, 2024

Abstract Background Generative Pretrained Model (GPT) chatbots have gained popularity since the public release of ChatGPT. Studies evaluated ability different GPT models to provide information about medical conditions. To date, no study has assessed quality ChatGPT outputs prostate cancer related questions from both physician and perspective while optimizing for patient consumption. Methods Nine cancer-related questions, identified through Google Trends (Global), were categorized into diagnosis, treatment, postoperative follow-up. These processed using 3.5, responses recorded. Subsequently, these re-inputted create simplified summaries understandable at a sixth-grade level. Readability original layperson was validated readability tools. A survey conducted among urology providers (urologists urologists in training) rate accuracy, completeness, clarity 5-point Likert scale. Furthermore, two independent reviewers on correctness trifecta: decision-making sufficiency. Public assessment summaries’ understandability carried out Amazon Mechanical Turk (MTurk). Participants rated demonstrated their understanding multiple-choice question. Results GPT-generated output deemed correct by 71.7% 94.3% raters (36 urologists, 17 residents) across 9 scenarios. this as accurate 8 (88.9%) scenarios sufficient make decision Mean higher than ([original v. ChatGPT, mean (SD), p -value] Flesch Reading Ease: 36.5(9.1) 70.2(11.2), <0.0001; Gunning Fog: 15.8(1.7) 9.5(2.0), < 0.0001; Grade Level: 12.8(1.2) 7.4(1.7), Coleman Liau: 13.7(2.1) 8.6(2.4), 0.0002; Smog index: 11.8(1.2) 6.7(1.8), Automated Index: 13.1(1.4) 7.5(2.1), 0.0001). MTurk workers ( n = 514) (89.5–95.7%) correctly understood content (63.0–87.4%). Conclusion shows promise education contents, but technology is not designed delivering patients information. Prompting model respond with may enhance its utility when used GPT-powered chatbots.

Language: Английский

---

Clinical and Surgical Applications of Large Language Models: A Systematic Review

Journal of Clinical Medicine, Journal Year: 2024, Volume and Issue: 13(11), P. 3041 - 3041

Published: May 22, 2024

Background: Large language models (LLMs) represent a recent advancement in artificial intelligence with medical applications across various healthcare domains. The objective of this review is to highlight how LLMs can be utilized by clinicians and surgeons their everyday practice. Methods: A systematic was conducted following the Preferred Reporting Items for Systematic Reviews Meta-Analyses guidelines. Six databases were searched identify relevant articles. Eligibility criteria emphasized articles focused primarily on clinical surgical LLMs. Results: literature search yielded 333 results, 34 meeting eligibility criteria. All from 2023. There 14 original research articles, four letters, one interview, 15 These covered wide variety specialties, including subspecialties. Conclusions: have potential enhance delivery. In settings, assist diagnosis, treatment guidance, patient triage, physician knowledge augmentation, administrative tasks. documentation, planning, intraoperative guidance. However, addressing limitations concerns, particularly those related accuracy biases, crucial. should viewed as tools complement, not replace, expertise professionals.

Language: Английский

---

Both Patients and Plastic Surgeons Prefer Artificial Intelligence–Generated Microsurgical Information

Journal of Reconstructive Microsurgery, Journal Year: 2024, Volume and Issue: 40(09), P. 657 - 664

Published: Feb. 21, 2024

With the growing relevance of artificial intelligence (AI)-based patient-facing information, microsurgical-specific online information provided by professional organizations was compared with that ChatGPT (Chat Generative Pre-Trained Transformer) and assessed for accuracy, comprehensiveness, clarity, readability.

Language: Английский

---

Evaluating the Reliability of ChatGPT for Health-Related Questions: A Systematic Review

Informatics, Journal Year: 2025, Volume and Issue: 12(1), P. 9 - 9

Published: Jan. 17, 2025

The rapid advancement of large language models like ChatGPT has significantly impacted natural processing, expanding its applications across various fields, including healthcare. However, there remains a significant gap in understanding the consistency and reliability ChatGPT’s performance different medical domains. We conducted this systematic review according to an LLM-assisted PRISMA setup. high-recall search term “ChatGPT” yielded 1101 articles from 2023 onwards. Through dual-phase screening process, initially automated via subsequently manually by human reviewers, 128 studies were included. covered range specialties, focusing on diagnosis, disease management, patient education. assessment metrics varied, but most compared accuracy against evaluations clinicians or reliable references. In several areas, demonstrated high accuracy, underscoring effectiveness. some contexts revealed lower accuracy. mixed outcomes domains emphasize challenges opportunities integrating AI into certain areas suggests that substantial utility, yet inconsistent all indicates need for ongoing evaluation refinement. This highlights potential improve healthcare delivery alongside necessity continued research ensure reliability.

Language: Английский

---

Skin, scalpel and the silicon chip: a systematic review on the accuracy, bias and data governance of artificial intelligence in dermatology, minimally invasive aesthetics, aesthetic, plastic and reconstructive surgery

European Journal of Plastic Surgery, Journal Year: 2025, Volume and Issue: 48(1)

Published: Feb. 5, 2025

Language: Английский

---

Management of Dupuytren’s Disease: A Multi-Centric Comparative Analysis Between Experienced Hand Surgeons Versus Artificial Intelligence

Diagnostics, Journal Year: 2025, Volume and Issue: 15(5), P. 587 - 587

Published: Feb. 28, 2025

Background: Dupuytren's fibroproliferative disease affecting the hand's palmar fascia leads to progressive finger contractures and functional limitations. Management of this condition relies heavily on expertise hand surgeons, who tailor interventions based clinical assessment. With growing interest in artificial intelligence (AI) medical decision-making, study aims evaluate feasibility integrating AI into management by comparing AI-generated recommendations with those expert surgeons. Methods: This multicentric comparative involved three experienced surgeons five systems (ChatGPT, Gemini, Perplexity, DeepSeek, Copilot). Twenty-two standardized prompts representing various scenarios were used assess decision-making. Surgeons provided recommendations, which analyzed for concordance, rationale, predicted outcomes. Key metrics included union accuracy, surgeon agreement, precision, recall, F1 scores. The also evaluated performance unanimous versus non-unanimous cases inter-AI agreements. Results: Gemini ChatGPT demonstrated highest accuracy (86.4% 81.8%, respectively), while Copilot showed lowest (40.9%). Surgeon agreement was (45.5%) (42.4%). performed better (accuracy up 92.0%) than as low 35.0%). Inter-AI agreements ranged from 75.0% (ChatGPT-Gemini) 48.0% (DeepSeek-Copilot). Precision, scores consistently higher other systems. Conclusions: systems, particularly ChatGPT, show promise aligning surgical especially straightforward cases. However, significant variability exists, complex scenarios. should be viewed complementary judgment, requiring further refinement validation integration practice.

Language: Английский

---

AI in Hand Surgery: Assessing Large Language Models in the Classification and Management of Hand Injuries

Journal of Clinical Medicine, Journal Year: 2024, Volume and Issue: 13(10), P. 2832 - 2832

Published: May 11, 2024

Background: OpenAI's ChatGPT (San Francisco, CA, USA) and Google's Gemini (Mountain View, are two large language models that show promise in improving expediting medical decision making hand surgery. Evaluating the applications of these within field surgery is warranted. This study aims to evaluate ChatGPT-4 classifying injuries recommending treatment. Methods: were given 68 fictionalized clinical vignettes twice. The asked use a specific classification system recommend surgical or nonsurgical Classifications scored based on correctness. Results analyzed using descriptive statistics, paired two-tailed t-test, sensitivity testing. Results: Gemini, correctly 70.6% injuries, demonstrated superior ability over (mean score 1.46 vs. 0.87, p-value < 0.001). For management, higher intervention compared (98.0% 88.8%), but lower specificity (68.4% 94.7%). When ChatGPT, greater response replicability. Conclusions: Large like assisting making, particularly surgery, with generally outperforming ChatGPT. These findings emphasize importance considering strengths limitations different when integrating them into practice.

Language: Английский

---

Large Language Models for Intraoperative Decision Support in Plastic Surgery: A Comparison between ChatGPT-4 and Gemini

Medicina, Journal Year: 2024, Volume and Issue: 60(6), P. 957 - 957

Published: June 8, 2024

Background and Objectives: Large language models (LLMs) are emerging as valuable tools in plastic surgery, potentially reducing surgeons’ cognitive loads improving patients’ outcomes. This study aimed to assess compare the current state of two most common readily available LLMs, Open AI’s ChatGPT-4 Google’s Gemini Pro (1.0 Pro), providing intraoperative decision support reconstructive surgery procedures. Materials Methods: We presented each LLM with 32 independent scenarios spanning 5 utilized a 5-point 3-point Likert scale for medical accuracy relevance, respectively. determined readability responses using Flesch–Kincaid Grade Level (FKGL) Flesch Reading Ease (FRE) score. Additionally, we measured models’ response time. compared performance Mann–Whitney U test Student’s t-test. Results: significantly outperformed accurate (3.59 ± 0.84 vs. 3.13 0.83, p-value = 0.022) relevant (2.28 0.77 1.88 0.032) responses. Alternatively, provided more concise readable responses, an average FKGL (12.80 1.56) lower than ChatGPT-4′s (15.00 1.89) (p < 0.0001). However, there was no difference FRE scores 0.174). Moreover, Gemini’s time faster (8.15 1.42 s) ChatGPT’-4′s (13.70 2.87 Conclusions: Although both demonstrated potential tools. Nevertheless, their inconsistency across different procedures underscores need further training optimization ensure reliability decision-support

Language: Английский

---

Artificial Intelligence as a Triage Tool during the Perioperative Period: Pilot Study of Accuracy and Accessibility for Clinical Application

Plastic & Reconstructive Surgery Global Open, Journal Year: 2024, Volume and Issue: 12(2), P. e5580 - e5580

Published: Feb. 1, 2024

Background: Given the dialogistic properties of ChatGPT, we hypothesized that this artificial intelligence (AI) function can be used as a self-service tool where clinical questions directly answered by AI. Our objective was to assess content, accuracy, and accessibility AI-generated content regarding common perioperative for reduction mammaplasty. Methods: ChatGPT (OpenAI, February Version, San Francisco, Calif.) query 20 patient concerns arise in period Searches were performed duplicate both general term specific question. Query outputs analyzed objectively subjectively. Descriptive statistics, t tests, chi-square tests appropriate with predetermined level significance P less than 0.05. Results: From total 40 outputs, mean word length 191.8 words. Readability at thirteenth grade level. Regarding all 97.5% on topic. Medical advice deemed reasonable 100% cases. General queries more frequently reported overarching background information, whereas prescriptive information ( < 0.0001). AI specifically recommended following surgeon provided postoperative instructions 82.5% instances. Conclusions: Currently available tools, their nascent form, provide recommendations With further calibration, interfaces may serve fielding future; however, patients must always retain ability bypass technology able contact surgeon.

Language: Английский

---