Table 4.
Overview of the identified ethical barriers and facilitators.
| Topic | ||
|---|---|---|
| Barriers | Facilitators | |
| Consent (n = 16 studies) | Ethical concerns appear when patients are compelled to use a specific eHealth technology due to a lack of alternatives (47, 56) | Ensure transparent information disclosure to users (42, 43, 45, 49, 56, 58, 63, 65) |
| Deleting data from users who have withdrawn consent may be impossible, especially when it is anonymized (43, 63) | Obtain explicit consent for data sharing (36, 45, 47, 49, 56, 63) | |
| Patient data may be shared with other providers or third parties without the patient's explicit consent (47) | Uphold users’ right to withdraw consent (43, 47) | |
| Ensuring the acceptance and consent of both patients and medical professionals is challenging (70) | Streamline the process of obtaining and managing consent (36, 53) | |
| Patients may lack the ability to provide consent in emergency situations (45) | Empower patients to control access to their data (47) | |
| Privacy concerns can arise when parents have access to their children's records (45) | Supply age-appropriate information to inform children, even if they lack legal capacity for consent (43) | |
| Concerns arise about obtaining patient consent for various treatment options (59) | Present information to patients in an easily understandable format to aid in data interpretation (45) | |
| The process of obtaining consent can be resource intensive (40) | ||
| Users often overlook the fine print or simply click on “agree” without reading it carefully (51) | ||
| Transparency of data (n = 14 studies) | Opacity in the functioning of AI algorithms (39, 45, 56, 63) | Ensure transparency in data quality assessment (63, 65) |
| Lack of awareness among patients about the storage and sharing of their (sensitive) data (51, 66) | Ensure transparency in the decision-making process based of AI data (39, 56) | |
| Insufficient methodological transparency in deep learning models (70) | Promote the development of “open source” health technologies (58, 65) | |
| The increasing complexity of algorithms leads to decreased decision support precision in earlier (older) models (39) | Engage all relevant stakeholders in decision-making, potential adoption, and discussions regarding data usage boundaries (53, 60) | |
| Enhance transparency in data infrastructure and data flow (67) | ||
| Identify key stakeholders in the decision-making process for system and data-related matters (45) | ||
| Inclusiveness and diversity (n = 13 studies) | AI may contain biases that can unintentionally exclude or harm individuals (39, 56, 70) | Develop technologies that do not discriminate (43, 47, 49, 65) |
| Inequity in access and use of healthcare technology (60, 69) | Create user-friendly software to enhance ease of use (47, 58, 60) | |
| eHealth technologies could potentially be used as an excuse to reduce the provision of high-quality care by trained health professionals (43) | Ensure that individuals facing particular needs or risks are encompassed by the social security system's protection (51) | |
| Favoring users who willingly share health-related data over those who do not share (51) | ||
| Ongoing monitoring and privacy violations can lead to increased stigma around patients (38) | ||
| Algorithms may not consider patient preferences (39) | ||
| Balancing individual responsibility with communal solidarity can be challenging (51) | ||
| Striking a balance between societal benefits and potential harms is difficult (45) | ||
| Responsibility (n = 12 studies) | Ambiguity regarding the accountable party for collected data (41, 52, 58, 62, 67, 70) | Clarify the responsible party for technology validation and outline potential consequences in case of any harm (45, 52, 65, 66) |
| Lack of regulatory and ethical clarity regarding accountability, moral responsibility, and legal liability (56) | Support patient autonomy and respect their decision-making (43) | |
| Role confusion among healthcare professionals using AI for decision-making, necessitating a balance with their own judgments (45) | Incorporate human agency and oversight (49) | |
| Risk of excessive reliance or complacency induced by AI tools (56) | Establish clear agreements with IT providers regarding update and security responsibilities (67) | |
| Informant patients that the data they generate at home will influence their physician's clinical decisions (62) | ||
| Ensure patients are aware of the extent of access they have to the technology and the associated responsibilities (62) | ||
| Validation of eHealth (n = 10 studies) | Lack of clear certification systems or transparent guides for assessment (42, 53) | Ensure legal clarity and ethical soundness in technology validation (42, 49) |
| Uncertainty regarding the type of clinical and socio-economic evidence required from manufacturers (53) | Establish certification of medical devices and align the required clinical evidence with European Medical Device Regulation Standards (53, 63) | |
| Limited availability of high-quality evidence for eHealth (62) | Continuously validate eHealth technologies through clinical assessments (58) | |
| Difficulty in accessing complete and generalizable evidence for efficacy and effectiveness (62) | Develop a comprehensive framework with balanced regulations and innovation-friendly criteria for health technology assessment (53) | |
| Challenges faced by medical ethical committees in assessing novel eHealth solutions due to their unknown impact or burden (55) | Mandate manufacturers to conduct clinical safety evaluations before market release or deployment (37) | |
| The significant withdrawal of patients from studies can diminish the value of data analysis (59) | Base all technology components on evidence-based principles (58) | |
| Promote eHealth technologies with shared benefits and measurable outcomes (66) | ||
| Utilize real-world datasets from clinical trials for evidence generation and impact assessment (53) | ||
| Monitoring and Follow Up of Data Output (n = 7 studies) | The abundance of health technology choices and rapid innovation poses challenges among healthcare professionals (62, 69) | Emphasize that AI should complement rather than replace healthcare professionals, shifting their roles from processors to expert overseers (56) |
| Use of technologies that upload and share data may give individuals a sense of being under surveillance (51) | Ensure healthcare professionals have prompt access to information to enhance the speed and quality of their care decisions (60) | |
| Technologies may result in healthcare professionals feeling obligated to be available or responsible all the time (52) | Establish a technology that issues warnings at the organizational level rather than targeting individual healthcare professionals (52) | |
| False alerts generated by health technology (43) | ||
| Liability (n = 6 studies) | Concerns about the potential legal liability for harm to a patient's health (62) | Promote transparency regarding accountability (39, 65) |
| Lack of clarity in legislation and regulations concerning liability and accountability for both producers and healthcare providers (57) | Provide guidance on responsibilities and liabilities when different components interact with each other (48) | |
| Manufacturers’ concerns about potential liability due to external communication infrastructure vulnerabilities that could lead to damages (57) | ||
| Absence of accountability for the accuracy and correctness of shared data (51) | ||
| Implementation and Compliance with ethical policy, guidelines, and frameworks (n = 3 studies) | The typical industry practice of rapid prototype development with iterative cycles may not align with ethical standards (40) | Enhance the adaptability of ethical frameworks (40) |
| Develop regulatory and ethical frameworks for public-private partnerships (60) | ||
| Supply specialized guidance to specify the role of digital health in clinical practice (62) | ||