BMC Medical Ethics最新文献

Background: Artificial intelligence (AI) offers significant potential to drive advancements in healthcare; however, the development and implementation of AI models present complex ethical, legal, social, and technical challenges, as data practices often undermine regulatory frameworks in various regions worldwide. This study explores stakeholder perspectives on the development and deployment of AI algorithms for diabetic retinopathy (DR) screening, with a focus on ethical risks, data practices, governance, and emerging shortcomings in the Global South AI discourse.

Methods: Fifteen semi-structured interviews were conducted with ophthalmologists, program officers, AI developers, bioethics experts, and legal professionals. Thematic analysis was guided by OECD principles for responsible AI stewardship. Interviews were analyzed using MAXQDA software to identify themes related to AI trustworthiness and ethical governance.

Results: Six key themes emerged regarding the perceived trustworthiness of AI: algorithmic effectiveness, responsible data collection, ethical approval processes, explainability, implementation challenges, and accountability. Participants reported critical shortcomings in AI companies' data collection practices, including a lack of transparency, inadequate consent processes, and limited patient awareness about data ownership. These findings highlight how unchecked data collection and curation practices may reinforce data colonialism in low and middle-income healthcare systems.

Conclusion: Ensuring trustworthy AI requires transparent and accountable data practices, robust patient consent mechanisms, and regulatory frameworks aligned with ethical and privacy standards. Addressing these issues is vital to safeguarding patient rights, preventing data misuse, and fostering responsible AI ecosystems in the Global South.

Background: Workplace dignity is a fundamental ethical concern in nursing, directly related to professional identity, human rights, and the moral foundations of care. Despite increasing global attention, few qualitative studies have explored how nurses themselves perceive and articulate workplace dignity, particularly within sociocultural contexts marked by rapid healthcare transformation.

Methods: This qualitative study was conducted in January 2025 at a tertiary hospital in Hangzhou, China. Sixteen registered nurses were recruited through purposive sampling and interviewed using semi-structured face-to-face interviews. Data were analyzed using Colaizzi's method.

Results: Nurses described workplace dignity as a subjective experience shaped by respect, equality, and recognition of self-worth. Influencing factors included sociodemographic characteristics, patient-related factors, organizational factors, and social factors.

Conclusion: Workplace dignity is dynamic and socially constructed. Enhancing dignity requires supportive management, fair compensation, positive workplace culture, and improved societal recognition of the nursing profession.

Background: The patient-doctor relationship significantly influences treatment effectiveness. This relationship unfolds amid clinical complexity and emotional vulnerability. By framing the initial consultation as the beginning of Prof. Józef Tischner's dramatic arc, this study suggests that this encounter, where knowledge meets hope, embodies the classical Greek ideal of Kalos kagathos, understood here as the ethical and aesthetic unity of goodness and beauty, also referred to as moral beauty. Apathy emerges as the primary barrier to realizing these ethical ideals in clinical practice, diminishing the profession's value by negating the intrinsic beauty and moral significance of aiding those in distress. This ethical vision is herein interpreted through a Christian lens, where faith and spirituality, defined as the personal search for transcendent meaning and moral purpose, provide a deeper moral grounding for both empathy and vocation.

Methods: To examine both the philosophical and practical aspects of the patient-doctor relationship, we conducted a survey of 173 participants among patients, medical students, and doctors. The survey assessed perceptions of the therapeutic alliance, the role of empathy, and the impact of apathy. Statistical analysis identified key patterns and associations.

Results: The findings indicate that while doctors play a crucial role in shaping the therapeutic alliance, patients' engagement and attitudes are equally important. Apathy was identified as a major challenge, weakening the ethical foundation of medical practice and diminishing the depth of doctor-patient interactions. Participants emphasized the need for mutual effort in fostering a meaningful dialogue throughout the treatment process. Notably, participants who valued religion and spirituality, defined here as the personal orientation toward meaning, moral depth, and potentially transcendent connection, expressed a greater emphasis on empathy, shared moral responsibility, and the integration of spiritual care.

Conclusions: The formation of a strong therapeutic alliance requires shared responsibility between doctor and patient. Addressing apathy and enhancing engagement on both sides may strengthen this bond, improving the quality of care and patient outcomes. By drawing on Christian values and the spiritual dimension of care, this study highlights how faith can enrich the humanistic core of medicine.

This study investigates how different forms of literacy shape trust in medical AI and its transfer in healthcare contexts. Based on a survey of 1,250 participants, three findings emerge. First, digital literacy and AI literacy exert opposite influences on medical AI trust: while digital literacy enhances trust, higher AI literacy unexpectedly reduces it. This paradox highlights a theoretical puzzle in technology acceptance, suggesting that deeper knowledge can generate informed skepticism rather than blind confidence. Second, trust in medical AI transfers hierarchically, flowing to hospitals only through physician trust as a critical intermediary, underscoring the role of interpersonal trust in institutional trust building. Third, scientific literacy moderates this process, with higher literacy dampening trust transfer, reflecting the impact of cognitive processing differences. These results extend theories of trust and technology acceptance by integrating multiple literacies and uncovering divergent cognitive pathways. Practically, they call for communication strategies and policy designs that calibrate trust-strengthening physicians' role as trust brokers, balancing education about AI's capacities and risks, and leveraging explainable AI tools to sustain appropriate confidence in medical AI.

Background: The advancement of digital technologies has brought transformative changes across the healthcare sector, and nursing is no exception. However, existing research has largely overlooked the ethical challenges nursing students face in real-world clinical settings, including dilemmas arising from the integration of artificial intelligence, patient privacy, and digital communication. This study explored nursing students' perceptions and challenges regarding nursing ethics in the digital age, focusing on the ethical dilemmas they face and their strategies for resolution.

Methods: We conducted two focus group interviews with 13 nursing students from two universities in Seoul, South Korea. Discussions centered on their experiences with ethical conflicts in digital healthcare environments.

Results: Nursing students frequently experienced ethical tensions between utilizing digital information and protecting patient privacy, including concerns about photographing EMR screens and using patient data without explicit consent. Many reported confusion in ethical decision-making due to unclear institutional guidelines and insufficient practical guidance. A notable gap existed between theoretical ethics education and its application in digital contexts. Introducing AI in clinical settings further complicates ethical reasoning, raising questions about trust, accountability, and verification of AI-driven decisions. Students also highlighted ethical risks in digital communication, such as improper use of hospital messaging systems and social media, which threaten professionalism and confidentiality. These findings emphasize students' strong demand for enhanced digital ethics education through case-based learning, simulations, and experiential training.

Conclusions: Nursing students stressed the need for practical ethics education that bridges the gap between theory and real-world application. This study underscores the importance of mandatory, practice-focused training linked to clinical environments, reinforced patient privacy protections, and ongoing education addressing ethical issues posed by emerging technologies like AI. To develop ethically competent nurses, digital ethics education must adopt a comprehensive, practice-oriented framework grounded in the realities of modern healthcare.

Background: The professionalism perceptions of newly enrolled residents determine how to implement targeted professionalism training during residency, but few studies have paid attention to this point. Moreover, it is unclear whether residents' perceptions reflect their attitudes and behaviors regarding professionalism when examined quantitatively. This study aimed to investigate residents' professionalism perceptions (their understanding of professionalism and approach to teaching professionalism), and to examine the relationships between perceptions, attitudes, and behaviors regarding professionalism.

Methods: Using cluster sampling, we conducted a cross-sectional survey of newly enrolled residents from 23 training bases of China Medical University in Liaoning Province, China. A self-designed qualitative questionnaire was used to understand the residents' professionalism perceptions, and the revised Penn State College of Medicine Professionalism Questionnaire and the China Medical Professionalism Inventory were used to assess their professionalism attitudes and behaviors respectively. Directed and conventional content analysis was used to analyze qualitative data. Descriptive statistics, Spearman rank correlation, and univariate/multivariable linear regressions were used to analyze the quantitative data.

Results: A total of 838 (84.3%) residents effectively completed the survey. Through analysis of residents' professionalism perceptions, we identified 23 professionalism attributes across five domains (Compassion, respect, communication, and collaboration; Integrity and duty; Pursuit of excellence; Equity in health care, humanitarianism, adherence to ethical guidelines; and Other), 25 undergraduate courses involving professionalism in four disciplines, and 12 approaches to improving professionalism at five levels. We also found positive associations between professionalism perceptions and attitudes (β = 0.54, SE = 0.26), and between attitudes and behaviors (β = 1.18, SE = 0.04), but no association between perceptions and behaviors.

Conclusions: Residents collectively demonstrated a comprehensive understanding of professionalism, but there are inadequacies at the individual level, suggesting the need to strengthen the professionalism education during residency. No association was found between perceptions and behaviors, indicating the gap between them, and efforts are still needed to realize the translation from knowing into doing.

Background: End-of-life (EOL) care practices, particularly the withholding and withdrawing (W/W) of life-sustaining treatments (LSTs), remain underexplored in North Africa. This study examined factors that influence EOL practices in Morocco.

Methods: A nationwide online survey was conducted over one month (from July to August 2023) among Moroccan intensivists, assessing their perspectives on W/W LSTs, decision-making processes, and influencing factors, including ethical, cultural, and religious considerations. Univariable analyses were performed to screen for potential associations, followed by multivariable logistic regression to identify factors independently associated with W/W decisions.

Results: Of 351 invited intensivists, 151 completed the survey, yielding a 41% response rate. The mean age of respondents was 47 ± 9 years, and 84% were male. Most interpreted EOL care as providing palliative care (74%) or ensuring a dignified death (59%), while only 23% explicitly associated it with the cessation of life-sustaining treatment. Nearly 40% reported never having made withholding or withdrawing (W/W) decisions, and 88% made fewer than one such decision per week. While 59% considered withholding LSTs ethically acceptable, only 5% supported both withholding and withdrawing. The most frequently cited barriers were the absence of a legal framework (75%), sociocultural constraints (44%), and discomfort discussing EOL issues with families (58%). Patient wishes were considered in fewer than half of cases, whereas family preferences predominated in 66%. Documentation of W/W decisions was uncommon (27%), and formal institutional protocols were largely absent (94.5%). In multivariable analysis, practicing in a public hospital (odds ratio [OR] = 3.16, p = 0.005) and believing that Islam permits W/W decisions (OR = 3.49, p = 0.006) were independently associated with a higher likelihood of making such decisions.

Conclusion: Moroccan intensivists face major ethical and practical challenges in EOL care, including legal ambiguity, lack of protocols, limited patient involvement, and difficulty communicating with families. The findings highlight the urgent need for legislative reforms, standardized protocols, and improved education to support ethical, culturally sensitive, and patient-centred EOL practices.

Introduction: The ability of artificial intelligence (AI) to analyze data in real-time and improve patients' diagnosis has led to a rapid growth of AI- health research in Uganda. Yet, there are no national guidelines on how to conduct AI-research in an ethical manner. Recent studies have reported that ethics committees lack resources, expertise and training to appropriately address the risks that may arise from AI health research. This study aimed to develop a guide for ethical review of the development of AI systems in health research in Uganda.

Methods: This study employed an exploratory qualitative approach between March - October 2024, involving 35 stakeholders in two public universities in Uganda. In-depth interviews were conducted with twelve members of ethics committees who had ever reviewed AI- protocols, six bioethicists, eight health researchers and nine members of AI-development teams. A thematic approach was used to interpret the results.

Results: Six themes emerged from this data including promoting social value and equity; ensuring participants and end-user autonomy and safety; addressing data acquisition, access and sharing gaps; ensuring responsible data use and data minimization; promoting responsible AI and fostering collaborative partnerships. Respondents opined that AI holds promise for improving health research. However, its successful implementation demands ethical considerations to minimize harm to participants and end-users.

Conclusion: Overall, respondents felt that developing a guide for ethics review of AI-research may minimize potential harms that could arise from using AI tools in research. We recommend training of ethics committees on key ethical considerations for development of responsible AI tools.

Background: Large language models (LLMs) such as GPT-4 are increasingly used to simplify radiology reports and improve patient comprehension. However, excessive simplification may undermine informed consent and autonomy by compromising clinical accuracy. This study investigates the ethical implications of readability thresholds in AI-generated radiology reports, identifying the minimum reading level at which clinical accuracy is preserved.

Methods: We retrospectively analyzed 500 computed tomography and magnetic resonance imaging reports from a tertiary hospital. Each report was transformed into 17 versions (reading grade levels 1-17) using GPT-4 Turbo. Readability metrics and word counts were calculated for each version. Clinical accuracy was evaluated using radiologist assessments and PubMed-BERTScore. We identified the first grade level at which a statistically significant decline in accuracy occurred, determining the lowest level that preserved both accuracy and readability. We further assessed potential clinical consequences in reports simplified to the 7th-grade level.

Results: Readability scores showed strong correlation with prompted reading levels (r = 0.80-0.84). Accuracy remained stable across grades 13-11 but declined significantly below grade 11. At the 7th-grade level, 20% of reports contained inaccuracies with potential to alter patient management, primarily due to omission, incorrect conversion, or inappropriate generalization. The 11th-grade level emerged as the current lower bound for preserving accuracy in LLM-generated radiology reports.

Conclusions: Our findings highlight an ethical tension between improving readability and maintaining clinical accuracy. While 7th-grade readability remains an ethical ideal, current AI tools cannot reliably produce accurate reports below the 11th-grade level. Ethical implementation of AI-generated reporting should include layered communication strategies and model transparency to safeguard patient autonomy and comprehension.