A Heuristic for Notifying Patients About AI: From Institutional Declarations to Informed Consent
Open Peer Commentaries
Matthew Elmore, Nicoleta Economou-Zavlanos, Michael Pencina
The American Journal of Bioethics, 24 February 2025
Excerpt
    The principle of respect for autonomy, often expressed as the right to understand and make decisions about one’s care, has recently gained attention in AI-related bioethics. Hurley et al. (2025) have made an important contribution by examining the Whitehouse’s Blueprint for an AI Bill of Rights, asking how the right to notice and explanation might apply in healthcare contexts. They propose three possible functions for this right in patient care: (1) to provide a simple “FYI” to patients about the use of AI; (2) to foster education and trust; and (3) to serve as part of a patient’s right to informed consent.
This commentary offers a heuristic for determining how best to plot these three aims (Table 1). Simplifying the recent work of Rose and Shapiro (2024), our heuristic lays out the functions described by Hurley et al. on a four-quadrant grid, scaling notification practices along two axes: the degree of AI autonomy and the degree of clinical risk. The need for robust consent increases when clinical risk is higher and when AI has greater autonomy in decision-making. This heuristic is adaptable to institution-specific measures of clinical risk, and it also provides flexibility for institutions to address their unique workflows…

Clarifying When Consent Might Be Illusory in Notice and Explanation Rights
Bryan Pilkington, Charles E. Binkley
The American Journal of Bioethics, 24 February 2025
Excerpt
In “Patient Consent and The Right to Notice and Explanation of AI Systems Used in Health Care,” Hurley et al. (2025) helpfully summarize key features of the AI Bill of Rights, focusing on the right to notice and explanation (RNandE) and arguing that greater clarity, including a specified goal, is needed. Though we concur with their overall assessment and appreciate the authors’ reliance on our work (Binkley and Pilkington 2023a) on consent, further clarification is needed, as some of the nuance of our position appears to have been lost in their goal categorizations. In exploring the normative function of RNandE, the authors ask what RNandE “is meant to achieve and how (and why) it is morally important at the individual patient level in healthcare?” We hold that for the three categories of use cases that the authors reference (chatbot, diagnostic and prognostic), the benefit to patients in claiming a RNandE of AI systems in their healthcare is not that AI models per se will be used. Rather the benefit to patients is the RNandE about how the output of the AI systems (chats, diagnoses, prognoses) will be used in their care, how patients will be informed, and with whom, besides patients and the clinicians involved in their care, the information will be shared.

Potential role of ChatGPT in simplifying and improving informed consent forms for vaccination: a pilot study conducted in Italy
Original Research
Claudia Cosma, Alessio Radi, Rachele Cattano, Patrizio Zanobini, Guglielmo Bonaccorsi, Chiara Lorini, Marco Del Riccio
BMJ Health & Care Informatics, 22 February 2025
Open Access
Abstract
Objectives
Informed consent forms are important for assisting patients in making informed choices regarding medical procedures. Because of their lengthy nature, complexity and specialised terminology, consent forms usually prove challenging for the general public to comprehend. This pilot study aims to use Chat Generative Pretrained Transformer (ChatGPT), a large language model (LLM), to improve the readability and understandability of a consent form for vaccination.
Methods
The study was conducted in Italy, within the Central Tuscany Local Health Unit. Three different consent forms were selected and approved: the standard consent form currently in use (A), a new form totally generated by ChatGPT (B) and a modified version of the standard form created by ChatGPT (C). Healthcare professionals in the vaccination unit were asked to evaluate the consent forms regarding adequacy, comprehensibility and completeness and to give an overall judgement. The Kruskal–Wallis test and Dunn’s test were used to evaluate the median scores of the consent forms across these variables.
Results
Consent forms A and C achieved the top scores in every category; consent form B obtained the lowest score. The median scores were 4.0 for adequacy on consent forms A and C and 3.0 on consent form B. Consent forms A and C received high overall judgement ratings with median scores of 4.0, whereas consent form B received a median score of 3.0.
Conclusions
The findings indicate that LLM tools such as ChatGPT could enhance healthcare communication by improving the clarity and accessibility of consent forms, but the best results are seen when these tools are combined with human knowledge and supervision.

Analyzing patient perspectives with large language models: a cross-sectional study of sentiment and thematic classification on exception from informed consent
Scientific Reports
Aaron E. Kornblith, Chandan Singh, Johanna C. Innes, Todd P. Chang, Kathleen M. Adelgais, Maija Holsti, Joy Kim, Bradford McClain, Daniel K. Nishijima, Steffanie Rodgers, Manish I. Shah, Harold K. Simon, John M. VanBuren, Caleb E. Ward, Catherine R. Counts
Nature, 20 February 2025
Open Access
Abstract
Large language models (LLMs) can improve text analysis efficiency in healthcare. This study explores the application of LLMs to analyze patient perspectives within the exception from informed consent (EFIC) process, which waives consent in emergency research. Our objective is to assess whether LLMs can analyze patient perspectives in EFIC interviews with performance comparable to human reviewers. We analyzed 102 EFIC community interviews from 9 sites, each with 46 questions, as part of the Pediatric Dose Optimization for Seizures in Emergency Medical Services study. We evaluated 5 LLMs, including GPT-4, to assess sentiment polarity on a 5-point scale and classify responses into predefined thematic classes. Three human reviewers conducted parallel analyses, with agreement measured by Cohen’s Kappa and classification accuracy. Polarity scores between LLM and human reviewers showed substantial agreement (Cohen’s kappa: 0.69, 95% CI 0.61–0.76), with major discrepancies in only 4.7% of responses. LLM achieved high thematic classification accuracy (0.868, 95% CI 0.853–0.881), comparable to inter-rater agreement among human reviewers (0.867, 95% CI 0.836–0.901). LLMs enabled large-scale visual analysis, comparing response statistics across sites, questions, and classes. LLMs efficiently analyzed patient perspectives in EFIC interviews, showing substantial sentiment assessment and thematic classification performance. However, occasional underperformance suggests LLMs should complement, not replace, human judgment. Future work should evaluate LLM integration in EFIC to enhance efficiency, reduce subjectivity, and support accurate patient perspective analysis.

Transforming Informed Consent Generation Using Large Language Models: Mixed Methods Study
Qiming Shi, Katherine Luzuriaga, Jeroan J Allison, Asil Oztekin, Jamie M Faro, Joy L Lee, Nathaniel Hafer, Margaret McManus, Adrian H Zai
JMIR Medical Informatics, 13 February 2025
Abstract
Background
Informed consent forms (ICFs) for clinical trials have become increasingly complex, often hindering participant comprehension and engagement due to legal jargon and lengthy content. The recent advances in large language models (LLMs) present an opportunity to streamline the ICF creation process while improving readability, understandability, and actionability.
Objectives
This study aims to evaluate the performance of the Mistral 8x22B LLM in generating ICFs with improved readability, understandability, and actionability. Specifically, we evaluate the model’s effectiveness in generating ICFs that are readable, understandable, and actionable while maintaining the accuracy and completeness.
Methods
We processed 4 clinical trial protocols from the institutional review board of UMass Chan Medical School using the Mistral 8x22B model to generate key information sections of ICFs. A multidisciplinary team of 8 evaluators, including clinical researchers and health informaticians, assessed the generated ICFs against human-generated counterparts for completeness, accuracy, readability, understandability, and actionability. Readability, Understandability, and Actionability of Key Information indicators, which include 18 binary-scored items, were used to evaluate these aspects, with higher scores indicating greater accessibility, comprehensibility, and actionability of the information. Statistical analysis, including Wilcoxon rank sum tests and intraclass correlation coefficient calculations, was used to compare outputs.
Results
LLM-generated ICFs demonstrated comparable performance to human-generated versions across key sections, with no significant differences in accuracy and completeness (P>.10). The LLM outperformed human-generated ICFs in readability (Readability, Understandability, and Actionability of Key Information score of 76.39% vs 66.67%; Flesch-Kincaid grade level of 7.95 vs 8.38) and understandability (90.63% vs 67.19%; P=.02). The LLM-generated content achieved a perfect score in actionability compared with the human-generated version (100% vs 0%; P<.001). Intraclass correlation coefficient for evaluator consistency was high at 0.83 (95% CI 0.64-1.03), indicating good reliability across assessments.
Conclusions
The Mistral 8x22B LLM showed promising capabilities in enhancing the readability, understandability, and actionability of ICFs without sacrificing accuracy or completeness. LLMs present a scalable, efficient solution for ICF generation, potentially enhancing participant comprehension and consent in clinical trials.

ChatGPT’s role in alleviating anxiety in total knee arthroplasty consent process: a randomized controlled trial pilot study
Randomised Controlled Trial
Wenyi Gan, Jianfeng Ouyang, Guorong She, Zhaowen Xue, Lingxuan Zhu, Anqi Lin, Weiming Mou, Aimin Jiang, Chang Qi, Quan Cheng, Peng Luo, Hua Li, Xiaofei Zheng
International Journal of Surgery, 4 February 2025
Open Access
Abstract
Background
Recent advancements in artificial intelligence (AI) like ChatGPT have expanded possibilities for patient education, yet its impact on perioperative anxiety in total knee arthroplasty (TKA) patients remains unexplored.
Methods
In this single-blind, randomized controlled pilot study from April to July 2023, 60 patients were randomly allocated using sealed envelopes to either ChatGPT-assisted or traditional surgeon-led informed consent groups. In the ChatGPT group, physicians used ChatGPT 4.0 to provide standardized, comprehensive responses to patient queries during the consent process, while maintaining their role in interpreting and contextualizing the information. Outcomes were measured using Hospital Anxiety and Depression Scales (HADS), Perioperative Apprehension Scale-7 (PAS-7), Visual Analogue Scales for Anxiety and Pain (VAS-A, VAS-P), Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC), and satisfaction questionnaires.
Results
Of 55 patients completing the study, the ChatGPT group showed significantly lower anxiety scores after informed consent (HADS-A: 10.48 ± 3.84 vs 12.75 ± 4.12, P = .04, Power = .67; PAS-7: 12.44 ± 3.70 vs 14.64 ± 2.11, P = .01, Power = .85; VAS-A: 5.40 ± 1.89 vs 6.71 ± 2.27, P = .02, Power = .75) and on the fifth postoperative day (HADS-A: 8.33 ± 3.20 vs 10.71 ± 3.83, P = .01, Power = .79; VAS-A: 3.41 ± 1.58 vs 4.64 ± 1.70, P = .008, Power = .85). The ChatGPT group also reported higher satisfaction with preoperative education (4.22 ± 0.51 vs 3.43 ± 0.84, P<.001, Power = .99) and overall hospitalization experience (4.11 ± 0.65 vs 3.46 ± 0.69, P = .001, Power = .97). No significant differences were found in depression scores, knee function, or pain levels.
Conclusions
ChatGPT-assisted informed consent effectively reduced perioperative anxiety and improved patient satisfaction in TKA patients. While these preliminary findings are promising, larger studies are needed to validate these results and explore broader applications of AI in preoperative patient education.

Artificial Intelligence and Informed Consent: an information science perspective on privacy policies and terms of use in major AI platforms
Conference Presentation
Jonas Ferrigolo Melo, Moises Rockembach
Artificial Intelligence in Library and Information Science: Exploring the Intersection, January 2025; Istanbul, Turkey
Abstract
This study examines how AI platforms incorporate the principles of informed consent (IC) concerning transparency, privacy, and user autonomy. The research aims to identify areas of convergence, divergence, and potential integration between the conceptual framework of IC in Information Science and the Privacy Policies of ChatGPT, Gemini, and Co-pilot. By analyzing these policies, the study explores how these platforms communicate data practices, ensure user control over personal information, and align with ethical standards of informed decision-making. The findings contribute to a broader understanding of AI governance and user rights in digital environments.

ChatGPT in Dental Education: Enhancing Student Proficiency in Informed Consent
Short Communication
Les Kalman, Arman Danesh
Medical Science Educator, 22 January 2025
Abstract
There is a scarcity of research regarding how ChatGPT can be integrated in dental curricula to develop the understanding of students regarding informed consent. Our study implemented an educational exercise where D3 dental students compared ChatGPT-generated consent forms with professionally validated consent forms, qualitatively assessing the educational value that the chatbot provides in this setting. The findings of the present study encourage dental educators to acknowledge the importance of supervised learning in helping the next generations of dentists prepare for the safe use of AI technologies, such as ChatGPT, in their future practice.

Use of ChatGPT to Generate Informed Consent for Surgery in Urogynecology
Emily S. Johnson, Eva K. Welch, Jacqueline Kikuchi, Heather Barbier, Christine M. Vaccaro, Felicia Balzano, Katherine L. Dengler
Urogynecology, 17 January 2025
Abstract
Importance
Use of the publicly available Large Language Model, Chat Generative Pre-trained Transformer (ChatGPT 3.5; OpenAI, 2022), is growing in health care despite varying accuracies.
Objective
The aim of this study was to assess the accuracy and readability of ChatGPT’s responses to questions encompassing surgical informed consent in urogynecology.
Study Design
Five fellowship-trained urogynecology attending physicians and 1 reconstructive female urologist evaluated ChatGPT’s responses to questions about 4 surgical procedures: (1) retropubic midurethral sling, (2) total vaginal hysterectomy, (3) uterosacral ligament suspension, and (4) sacrocolpopexy. Questions involved procedure descriptions, risks/benefits/alternatives, and additional resources. Responses were rated using the DISCERN tool, a 4-point accuracy scale, and the Flesch-Kinkaid Grade Level score.
Results
The median DISCERN tool overall rating was 3 (interquartile range [IQR], 3–4), indicating a moderate rating (“potentially important but not serious shortcomings”). Retropubic midurethral sling received the highest overall score (median, 4; IQR, 3–4), and uterosacral ligament suspension received the lowest (median, 3; IQR, 3–3). Using the 4-point accuracy scale, 44.0% of responses received a score of 4 (“correct and adequate”), 22.6% received a score of 3 (“correct but insufficient”), 29.8% received a score of 2 (“accurate and misleading information together”), and 3.6% received a score of 1 (“wrong or irrelevant answer”). ChatGPT performance was poor for discussion of benefits and alternatives for all surgical procedures, with some responses being inaccurate. The mean Flesch-Kinkaid Grade Level score for all responses was 17.5 (SD, 2.1), corresponding to a postgraduate reading level.
Conclusions
Overall, ChatGPT generated accurate responses to questions about surgical informed consent. However, it produced clearly false portions of responses, highlighting the need for a careful review of responses by qualified health care professionals.

The ethical implications of using children’s photographs in artificial intelligence: challenges and recommendations
Original Research
Wael Badawy
AI and Ethics, 15 January 2025
Abstract
The rapid advancement of artificial intelligence (AI) technologies has brought to the forefront the ethical implications of using children’s photographs within these systems. This paper aims to explore the multifaceted ethical considerations surrounding the utilization of children’s images in AI, highlighting the need for a delicate balance between technological progress and the protection of children’s rights. We examine the challenges of obtaining informed consent, the risks associated with data security and misuse, the potential for bias and discrimination, and the psychological impacts on children whose images are used in AI applications. Through a critical analysis of existing legal frameworks and ethical guidelines, this study identifies gaps in the protection of children’s digital identities and proposes a set of ethical recommendations and best practices for the responsible use of children’s photographs in AI. By advocating for enhanced safeguards, transparency, and accountability, this paper contributes to the ongoing dialogue on ethical AI practices, underscoring the importance of prioritizing children’s privacy and well-being in the digital age.