The ethics of ChatGPT in medicine and healthcare: a systematic review on Large Language Models (LLMs).

With the introduction of ChatGPT, Large Language Models (LLMs) have received enormous attention in healthcare. Despite potential benefits, researchers have underscored various ethical implications. While individual instances have garnered attention, a systematic and comprehensive overview of practical applications currently researched and ethical issues connected to them is lacking. Against this background, this work maps the ethical landscape surrounding the current deployment of LLMs in medicine and healthcare through a systematic review. Electronic databases and preprint servers were queried using a comprehensive search strategy which generated 796 records. Studies were screened and extracted following a modified rapid review approach. Methodological quality was assessed using a hybrid approach. For 53 records, a meta-aggregative synthesis was performed. Four general fields of applications emerged showcasing a dynamic exploration phase. Advantages of using LLMs are attributed to their capacity in data analysis, information provisioning, support in decision-making or mitigating information loss and enhancing information accessibility. However, our study also identifies recurrent ethical concerns connected to fairness, bias, non-maleficence, transparency, and privacy. A distinctive concern is the tendency to produce harmful or convincing but inaccurate content. Calls for ethical guidance and human oversight are recurrent. We suggest that the ethical guidance debate should be reframed to focus on defining what constitutes acceptable human oversight across the spectrum of applications. This involves considering the diversity of settings, varying potentials for harm, and different acceptable thresholds for performance and certainty in healthcare. Additionally, critical inquiry is needed to evaluate the necessity and justification of LLMs' current experimental use.

The Role of Transparency in Digital Contact Tracing During COVID-19: Insights from an Expert Survey.

Health technologies such as apps for digital contract tracing [DCT] played a crucial role in containing and combating infections during the COVID-19 pandemic. Their primary function was to prevent the spread of SARS-CoV-2 by consistently generating and disseminating information related to various events such as encounters, vaccinations or infections. While the functionality of DCT has been well researched, the necessity of transparency in the use of DCT and the consent to share sensitive information such as users' health, vaccination and location status remains unclear. On one hand, DCT enabled the continuous monitoring of various risk factors, including data-based calculations of infection probabilities. On the other hand, digital monitoring of health risks was closely associated with various uncertainties, such as the ambiguous storage of personal data and its potential future misuse, e.g., by tech companies or health authorities. Our contribution aims to retrospectively analyze the COVID-19 pandemic from a post-pandemic perspective and utilize it as a case study for the implementation of new technological measures. We argue that under the condition of voluntary use of DCT, transparency plays a key role in convincing individuals to install health technologies on their mobile devices, keep them activated and consent to the sharing of sensitive data. We support our argument with qualitative data from an expert survey conducted between 2020 and 2021 and analyzed according to the principles of Grounded Theory.

Increasing efficiency and well-being? a systematic review of the empirical claims of the double-benefit argument in socially assistive devices.

BACKGROUND: Socially assistive devices (care robots, companions, smart screen assistants) have been advocated as a promising tool in elderly care in Western healthcare systems. Ethical debates indicate various challenges. One of the most prevalent arguments in the debate is the double-benefit argument claiming that socially assistive devices may not only provide benefits for autonomy and well-being of their users but might also be more efficient than other caring practices and might help to mitigate scarce resources in healthcare. Against this background, we used a subset of comparative empirical studies from a comprehensive systematic review on effects and perceptions of human-machine interaction with socially assistive devices to gather and appraise all available evidence supporting this argument from the empirical side. METHODS: Electronic databases and additional sources were queried using a comprehensive search strategy which generated 9851 records. Studies were screened independently by two authors. Methodological quality of studies was assessed. For 39 reports using a comparative study design, a narrative synthesis was performed. RESULTS: The data shows positive evidential support to claim that some socially assistive devices (Paro) might be able to contribute to the well-being and autonomy of their users. However, results also indicate that these positive findings may be heavily dependent on the context of use and the population. In addition, we found evidence that socially assistive devices can have negative effects on certain populations. Evidence regarding the claim of efficiency is scarce. Existing results indicate that socially assistive devices can be more effective than standard of care but are far less effective than plush toys or placebo devices. DISCUSSION: We suggest using the double-benefit argument with great caution as it is not supported by the currently available evidence. The occurrence of potentially negative effects of socially assistive devices requires more research and indicates a more complex ethical calculus than suggested by the double-benefit argument.

Smartphone Apps for Containing the COVID-19 Pandemic in Germany: Qualitative Interview Study With Experts Based on Grounded Theory.

BACKGROUND: Smartphone apps, including those for digital contact tracing (DCT), played a crucial role in containing infections during the COVID-19 pandemic. Their primary function is to generate and disseminate information to disrupt transmissions based on various events, such as encounters, vaccinations, locations, or infections. Although the functionality of these apps has been extensively studied, there is still a lack of qualitative research addressing critical issues. OBJECTIVE: We will demonstrate that the use of DCT presents a challenge due to the tension between continuous health monitoring and uncertainties related to transparency and user sovereignty. On one hand, DCT enables the monitoring of various risk factors, including data-based calculations of infection probabilities. On the other hand, continuous risk management is intertwined with several uncertainties, including the unclear storage of personal data, who has access to it, and how it will be used in the future. METHODS: We focus on the German "Corona-Warn-App" and support our argument with empirical data from 19 expert interviews conducted between 2020 and 2021. The interviews were conducted using a semistructured questionnaire and analyzed according to the principles of grounded theory. RESULTS: Our data underscores 3 dimensions: transparency, data sovereignty, and the east-west divide. While transparency is considered an essential foundation for establishing trust in the use of DCT by providing a sense of security, data sovereignty is seen as a high value during the pandemic, protecting users from an undesired loss of control. The aspect of the east-west divide highlights the idea of incorporating sociocultural values and standards into technology, emphasizing that algorithms and data-driven elements, such as distance indicators, encounters, and isolations, are also influenced by sociocultural factors. CONCLUSIONS: The effective use of DCT for pandemic containment relies on achieving a balance between individual control and technological prevention. Maximizing the technological benefits of these tools is crucial. However, users must also be mindful of the information they share and maintain control over their shared data.

Cultural Implications Regarding Privacy in Digital Contact Tracing Algorithms: Method Development and Empirical Ethics Analysis of a German and a Japanese Approach to Contact Tracing.

BACKGROUND: Digital contact tracing algorithms (DCTAs) have emerged as a means of supporting pandemic containment strategies and protecting populations from the adverse effects of COVID-19. However, the impact of DCTAs on users' privacy and autonomy has been heavily debated. Although privacy is often viewed as the ability to control access to information, recent approaches consider it as a norm that structures social life. In this regard, cultural factors are crucial in evaluating the appropriateness of information flows in DCTAs. Hence, an important part of ethical evaluations of DCTAs is to develop an understanding of their information flow and their contextual situatedness to be able to adequately evaluate questions about privacy. However, only limited studies and conceptual approaches are currently available in this regard. OBJECTIVE: This study aimed to develop a case study methodology to include contextual cultural factors in ethical analysis and present exemplary results of a subsequent analysis of 2 different DCTAs following this approach. METHODS: We conducted a comparative qualitative case study of the algorithm of the Google Apple Exposure Notification Framework as exemplified in the German Corona Warn App and the Japanese approach of Computation of Infection Risk via Confidential Locational Entries (CIRCLE) method. The methodology was based on a postphenomenological perspective, combined with empirical investigations of the technological artifacts within their context of use. An ethics of disclosure approach was used to focus on the social ontologies created by the algorithms and highlight their connection to the question about privacy. RESULTS: Both algorithms use the idea of representing a social encounter of 2 subjects. These subjects gain significance in terms of risk against the background of a representation of their temporal and spatial properties. However, the comparative analysis reveals 2 major differences. Google Apple Exposure Notification Framework prioritizes temporality over spatiality. In contrast, the representation of spatiality is reduced to distance without any direction or orientation. However, the CIRCLE framework prioritizes spatiality over temporality. These different concepts and prioritizations can be seen to align with important cultural differences in considering basic concepts such as subject, time, and space in Eastern and Western thought. CONCLUSIONS: The differences noted in this study essentially lead to 2 different ethical questions about privacy that are raised against the respective backgrounds. These findings have important implications for the ethical evaluation of DCTAs, suggesting that a culture-sensitive assessment is required to ensure that technologies fit into their context and create less concern regarding their ethical acceptability. Methodologically, our study provides a basis for an intercultural approach to the ethics of disclosure, allowing for cross-cultural dialogue that can overcome mutual implicit biases and blind spots based on cultural differences.

Socially Assistive Devices in Healthcare-a Systematic Review of Empirical Evidence from an Ethical Perspective.

Socially assistive devices such as care robots or companions have been advocated as a promising tool in elderly care in Western healthcare systems. Ethical debates indicate various challenges. An important part of the ethical evaluation is to understand how users interact with these devices and how interaction influences users' perceptions and their ability to express themselves. In this review, we report and critically appraise findings of non-comparative empirical studies with regard to these effects from an ethical perspective.Electronic databases and other sources were queried using a comprehensive search strategy generating 9851 records. Studies were screened independently by two authors. Methodological quality of studies was assessed. For 22 reports on 21 datasets using a non-comparative design a narrative synthesis was performed.Data shows positive findings in regard to attitudes and emotional reactions of users. Varying perception of a social relation and social presence are the most commonly observed traits of interaction. Users struggle with understanding technical complexities while functionality of the devices is limited. This leads to a behavioral alignment of users towards the requirements of the devices to be able to make use of them.This evidence adds to three important ethical debates on the use of socially assistive devices in healthcare in regard to (1) reliability of existing empirical evidence to inform normative judgements, (2) ethical significance of the social presence of devices and (3) user autonomy in regard to behavioral alignment.

How to derive ethically appropriate recommendations for action? A methodology for applied ethics.

Researchers in applied ethics, and some areas of bioethics particularly, aim to develop concrete and appropriate recommendations for action in morally relevant real-world situations. When proceeding from more abstract levels of ethical reasoning to such concrete recommendations, however, even with regard to the very same normative principle or norm, it seems possible to develop divergent or even contradictory recommendations for action regarding a certain situation. This may give the impression that such recommendations would be arbitrary and, hence, not well justified. Against this background, we, first, aim at showing that ethical recommendations for action, although being contingent in some sense, are not arbitrary if developed appropriately. For this purpose, we examine two types of contingencies arising in applied ethics reasoning based on recent examples of recommendations for action in the context of the COVID-19 pandemic. In doing so, we refer to a three-step model of ethical reasoning towards recommendations for actions. This, however, leaves open the question of how applied ethics may cope with contingent recommendations for action. Therefore, in a second step, we analyze the role of bridge principles for developing ethically appropriate recommendations for action, i.e., principles which connect normative claims with relevant empirical information to justify certain recommendations for action in a given morally relevant situation. Finally, we discuss some implications for reasoning and reporting in empirically informed ethics.

'Can you hear me?': communication, relationship and ethics in video-based telepsychiatric consultations.

Telepsychiatry has long been discussed as a supplement to or substitute for face-to-face therapeutic consultations. The current pandemic crisis has fueled the development in an unprecedented way. More and more psychiatric consultations are now carried out online as video-based consultations. Treatment results appear to be comparable with those of face-to-face care in terms of clinical outcome, acceptance, adherence and patient satisfaction. However, evidence on videoconferencing in a variety of different fields indicates that there are extensive changes in the communication behaviour in online conversations. We hypothesise that this might impact ethically relevant aspects of the therapeutic relationship, which plays a prominent role in psychiatry. In this paper, we review effects of video-based consultations on communication between therapists and patients in psychiatry. Based on a common understanding of video-based consultations as changing the lived experience of communication, we categorise these effects according to sensory, spatial and technical aspects. Departing from a power-based model of therapeutic relationships, we then discuss the ethical significance of this changed communication situation, based on dimensions of respect for autonomy, lucidity, fidelity, justice and humanity. We conclude that there is evidence for ethically relevant changes of the therapeutic relationship in video-based telepsychiatric consultations. These changes need to be more carefully considered in psychiatric practice and future studies.

Framework for evaluation research on clinical ethical case interventions: the role of ethics consultants.

Evaluation of clinical ethical case consultations has been discussed as an important research task in recent decades. A rigid framework of evaluation is essential to improve quality of consultations and, thus, quality of patient care. Different approaches to evaluate those services appropriately and to determine adequate empirical endpoints have been proposed. A key challenge is to provide an answer to the question as to which empirical endpoints-and for what reasons-should be considered when evaluating the quality of a service. In this paper, we argue for an approach that adopts the role of ethics consultants as its point of departure. In a first step, we describe empirical and ethical characteristics of evaluating clinical ethical case. We show that the mode of action and the explicit normative character of the interventions constitute two characteristics which pose challenges to the selection of appropriate quality criteria and require special attention. In a second step, we outline the way in which an analysis of the role of ethics consultants in the context of a clinical ethical case consultation services can account for the existing challenges by linking empirically measurable endpoints with normative theory. Finally, we discuss practical implications of our model for evaluation research.

Types of Ethical Problems and Expertise in Clinical Ethics Consultation in Psychiatry - Insights From a Qualitative Empirical Ethics Study.

Background: Ethics consultation has been advocated as a valuable tool in ethically challenging clinical situations in healthcare. It is paramount for the development and implementation of clinical ethics support services (CESS) in psychiatry that interventions can address the moral needs of mental health professionals adequately and communicate the nature of the services clearly. This study explores types of ethical problems and concepts of ethical expertise as core elements of CESS in mental healthcare with the aim of contributing to the further development of ethical support in psychiatry. Methods: We conducted 13 semi-structured interviews with mental health professionals and CESS members and triangulated them with four non-participant observations of ethical case consultations in psychiatry. Data were analyzed according to principles of grounded theory and are discussed from a normative perspective. Results: The analysis of the empirical data reveals a typology of three different ethical problems professionals want to refer to CESS: (1) Dyadic problems based on the relationship between patients and professionals, (2) triangular problems, where a third party is involved and affected as a side effect, and (3) problems on a systemic level. However, CESS members focus largely on types (1) and (2), while the third remains unrecognized or members do not feel responsible for these problems. Furthermore, they reflect a strong inner tension connected to their role as ethical experts which can be depicted as a dilemma. On the one hand, as ethically trained people, they reject the idea that their judgments have expert status. On the other hand, they feel that mental health professionals reach out for them to obtain guidance and that it is their responsibility to offer it. Conclusion: CESS members and professionals in mental healthcare have different ideas of the scope of responsibility of CESS. This contains the risk of misunderstandings and misconceptions and may affect the quality of consultations. It is necessary to adapt concepts of problem solving to practitioners' needs to overcome these problems. Secondly, CESS members struggle with their role as ethical experts. CESS members in psychiatry need to develop a stable professional identity. Theoretical clarification and practical training are needed.

Outcomes of clinical ethics support near the end of life: A systematic review.

BACKGROUND: Clinical ethics support services have been advocated in recent decades. In clinical practice, clinical ethics support services are often requested for difficult decisions near the end of life. However, their contribution to improving healthcare has been questioned and demands for evaluation have been put forward. Research indicates that there are considerable challenges associated with defining adequate outcomes for clinical ethics support services. In this systematic review, we report findings of qualitative studies and surveys, which have been conducted to evaluate clinical ethics support services near the end of life. METHODS: Electronic databases and other sources were queried from 1970 to May 2018. Two authors screened studies independently. Methodological quality of studies was assessed. For each arm of the review, an individual synthesis was performed. Prospero ID: CRD42016036241. ETHICAL CONSIDERATIONS: Ethical approval is not needed as it is a systematic review of published literature. RESULTS: In all, 2088 hits on surveys and 2786 on qualitative studies were found. After screening, nine surveys and four qualitative studies were included. Survey studies report overall positive findings using a very wide and heterogeneous range of outcomes. Negative results were reported only occasionally. However, methodological quality and conceptual justification of used outcomes was often weak and limits generalizability of results. CONCLUSION: Evidence points to positive outcomes of clinical ethics support services. However, methodological quality needs to be improved. Further qualitative or mixed-method research on evaluating clinical ethics support services may contribute to the development of evaluating outcomes of clinical ethics support services by means of broaden the range of appropriate (process-oriented) outcomes of (different types of) clinical ethics support services.

Ethical case interventions for adult patients.

BACKGROUND: Decisions in clinical medicine can be associated with ethical challenges. Ethical case interventions (e.g. ethics committee, moral case deliberation) identify and analyse ethical conflicts which occur within the context of care for patients. Ethical case interventions involve ethical experts, different health professionals as well as the patient and his/her family. The aim is to support decision-making in clinical practice. This systematic review gathered and critically appraised the available evidence of controlled studies on the effectiveness of ethical case interventions. OBJECTIVES: To determine whether ethical case interventions result in reduced decisional conflict or moral distress of those affected by an ethical conflict in clinical practice; improved patient involvement in decision-making and a higher quality of life in adult patients. To determine the most effective models of ethical case interventions and to analyse the use and appropriateness of the outcomes in experimental studies. SEARCH METHODS: We searched the following electronic databases for primary studies to September 2018: CENTRAL, MEDLINE, Embase, CINAHL and PsycINFO. We also searched CDSR and DARE for related reviews. Furthermore, we searched Clinicaltrials.gov, International Clinical Trials Registry Platform Search Portal and conducted a cited reference search for all included studies in ISI WEB of Science. We also searched the references of the included studies. SELECTION CRITERIA: We included randomised trials, non-randomised trials, controlled before-after studies and interrupted time series studies which compared ethical case interventions with usual care or an active control in any language. The included population were adult patients. However, studies with mixed populations consisting of adults and children were included, if a subgroup or sensitivity analysis (or both) was performed for the adult population. DATA COLLECTION AND ANALYSIS: We used standard methodological procedures expected by Cochrane and the Effective Practice and Organisation of Care review group. We used meta-analysis based on a random-effects model for treatment costs and structured analysis for the remaining outcomes, because these were heterogeneously reported. We used the GRADE approach to assess the certainty of the evidence. MAIN RESULTS: We included four randomised trials published in six articles. The publication dates ranged from 2000 to 2014. Three studies were conducted in the USA, and one study in Taiwan. All studies were conducted on intensive care units and included 1165 patients. We judged the included studies to be of moderate or high risk of bias. It was not possible to compare different models of the intervention regarding effectiveness due to the diverse character of the interventions and the small number of studies. Included studies did not directly measure the main outcomes. All studies received public funding and one received additional funding from private sources.We identified two models of ethical case interventions: proactive and request-based ethics consultation. Three studies evaluated proactive ethics consultation (n = 1103) of which one study reported findings on one key outcome criterion. The studies did not report data on decisional conflict, moral distress of participants of ethical case interventions, patient involvement in decision-making, quality of life or ethical competency for proactive ethics consultation. One study assessed satisfaction with care on a 5-point Likert scale (1 = lowest rating, 5 = highest rating). The healthcare providers (nurses and physicians, n = 365) scored a value of 4 or 5 for 81.4% in the control group and 86.1% in the intervention group (P > 0.05). The patients or their surrogates (n = 275) scored a value of 4 or 5 for 83.6% in the control group and for 74.8% in the intervention group (P > 0.05). It was uncertain whether proactive ethics consultation led to high satisfaction with care, because the certainty of evidence was very low.One study evaluated request-based ethics consultation (n = 62). The study indirectly measured decisional conflict by assessing consensus regarding patient care. The risk (increase in consensus, reduction in decisional conflict) increased by 80% as a result of the intervention. The risk ratio was 0.20 (95% confidence interval 0.09 to 0.46; P < 0.01). It was uncertain whether request-based ethics consultation reduced decisional conflict, because the certainty of evidence was very low. The study did not report data on moral distress of participants of ethical case interventions, patient involvement in decision-making, quality of life, or ethical competency or satisfaction with care for request-based ethics consultation. AUTHORS' CONCLUSIONS: It is not possible to determine the effectiveness of ethical case interventions with certainty due to the low certainty of the evidence of included studies in this review. The effectiveness of ethical case interventions should be investigated in light of the outcomes reported in this systematic review. In addition, there is need for further research to identify and measure outcomes which reflect the goals of different types of ethical case intervention.

Do we understand the intervention? What complex intervention research can teach us for the evaluation of clinical ethics support services (CESS).

BACKGROUND: Evaluating clinical ethics support services (CESS) has been hailed as important research task. At the same time, there is considerable debate about how to evaluate CESS appropriately. The criticism, which has been aired, refers to normative as well as empirical aspects of evaluating CESS. MAIN BODY: In this paper, we argue that a first necessary step for progress is to better understand the intervention(s) in CESS. Tools of complex intervention research methodology may provide relevant means in this respect. In a first step, we introduce principles of "complex intervention research" and show how CESS fulfil the criteria of "complex interventions". In a second step, we develop a generic "conceptual framework" for "ethics consultation on request" as standard for many forms of ethics consultation in clinical ethics practice. We apply this conceptual framework to the model of "bioethics mediation" to make explicit the specific structural and procedural elements of this form of ethics consultation on request. In a final step we conduct a comparative analysis of two different types of CESS, which have been subject to evaluation research: "proactive ethics consultation" and "moral case deliberation" and discuss implications for evaluating both types of CESS. CONCLUSION: To make explicit different premises of implemented CESS interventions by means of conceptual frameworks can inform the search for sound empirical evaluation of CESS. In addition, such work provides a starting point for further reflection about what it means to offer "good" CESS.

"I need to know what makes somebody tick ": Challenges and Strategies of Implementing Shared Decision-Making in Individualized Oncology.

BACKGROUND: Shared decision-making (SDM) has been advocated as an ethical framework for decision-making in cancer care. According to SDM, patients make decisions in light of their values and based on the available evidence. However, SDM is difficult to implement in cancer care. A lack of applicability in practice is often reported. This empirical-ethical study explores factors potentially relevant to current difficulties in translating the concept of SDM into clinical practice. METHODS: This study was conducted with nonparticipant observation of the decision-making process in patients with gastrointestinal cancers for whom the benefit of adjuvant chemotherapy was uncertain according to clinical guidelines. Triangulation of qualitative data analysis was conducted by means of semistructured interviews subsequent to the observation. Observation notes and interview transcripts were analyzed according to the principles of grounded theory. RESULTS: Deviating from the concept of SDM, oncologists initiated a process of eliciting values and medical information prior to conveying information. The purpose of this approach was to select and individualize information relevant to the treatment decision. In doing so, the oncologists observed used two strategies: "biographical communication" and a "metacommunicative approach." Both strategies could be shown to be effective or to fail depending on patients' characteristics such as their view of the physicians' role and the relevance of value-related information for medical decision-making. CONCLUSION: In contrast to the conceptual account of SDM, oncologists are in need of patient-related information prior to conveying information. Both strategies observed to elicit such information are in principle justifiable but need to be adapted in accordance with patient preferences and decision-making styles. IMPLICATIONS FOR PRACTICE: This study showed that knowledge of patients' values and preferences is very important to properly adapt the giving of medical information and to further the process of shared decision-making. Shared decision-making (SDM) trainings should consider different strategies of talking about values. The right strategy depends largely on the patient's preferences in communication. To be aware of the role of values in SDM and to be able to switch communicative strategies might prove to be of particular value. A more systematic evaluation of the patient's decision-making preferences as part of routine procedures in hospitals might help to reduce value-related barriers in communication.