Biomedical Data Science, Artificial Intelligence, and Ethics: Navigating Challenges in the Face of Explosive Growth.

Advances in biomedical data science and artificial intelligence (AI) are profoundly changing the landscape of healthcare. This article reviews the ethical issues that arise with the development of AI technologies, including threats to privacy, data security, consent, and justice, as they relate to donors of tissue and data. It also considers broader societal obligations, including the importance of assessing the unintended consequences of AI research in biomedicine. In addition, this article highlights the challenge of rapid AI development against the backdrop of disparate regulatory frameworks, calling for a global approach to address concerns around data misuse, unintended surveillance, and the equitable distribution of AI's benefits and burdens. Finally, a number of potential solutions to these ethical quandaries are offered. Namely, the merits of advocating for a collaborative, informed, and flexible regulatory approach that balances innovation with individual rights and public welfare, fostering a trustworthy AI-driven healthcare ecosystem, are discussed.

Developer Perspectives on Potential Harms of Machine Learning Predictive Analytics in Health Care: Qualitative Analysis.

BACKGROUND: Machine learning predictive analytics (MLPA) is increasingly used in health care to reduce costs and improve efficacy; it also has the potential to harm patients and trust in health care. Academic and regulatory leaders have proposed a variety of principles and guidelines to address the challenges of evaluating the safety of machine learning-based software in the health care context, but accepted practices do not yet exist. However, there appears to be a shift toward process-based regulatory paradigms that rely heavily on self-regulation. At the same time, little research has examined the perspectives about the harms of MLPA developers themselves, whose role will be essential in overcoming the "principles-to-practice" gap. OBJECTIVE: The objective of this study was to understand how MLPA developers of health care products perceived the potential harms of those products and their responses to recognized harms. METHODS: We interviewed 40 individuals who were developing MLPA tools for health care at 15 US-based organizations, including data scientists, software engineers, and those with mid- and high-level management roles. These 15 organizations were selected to represent a range of organizational types and sizes from the 106 that we previously identified. We asked developers about their perspectives on the potential harms of their work, factors that influence these harms, and their role in mitigation. We used standard qualitative analysis of transcribed interviews to identify themes in the data. RESULTS: We found that MLPA developers recognized a range of potential harms of MLPA to individuals, social groups, and the health care system, such as issues of privacy, bias, and system disruption. They also identified drivers of these harms related to the characteristics of machine learning and specific to the health care and commercial contexts in which the products are developed. MLPA developers also described strategies to respond to these drivers and potentially mitigate the harms. Opportunities included balancing algorithm performance goals with potential harms, emphasizing iterative integration of health care expertise, and fostering shared company values. However, their recognition of their own responsibility to address potential harms varied widely. CONCLUSIONS: Even though MLPA developers recognized that their products can harm patients, public, and even health systems, robust procedures to assess the potential for harms and the need for mitigation do not exist. Our findings suggest that, to the extent that new oversight paradigms rely on self-regulation, they will face serious challenges if harms are driven by features that developers consider inescapable in health care and business environments. Furthermore, effective self-regulation will require MLPA developers to accept responsibility for safety and efficacy and know how to act accordingly. Our results suggest that, at the very least, substantial education will be necessary to fill the "principles-to-practice" gap.

Stronger regulation of AI in biomedicine.

Regulatory agencies need to ensure the safety and equity of AI in biomedicine, and the time to do so is now.

Not in my AI: Moral engagement and disengagement in health care AI development.

Machine learning predictive analytics (MLPA) are utilized increasingly in health care, but can pose harms to patients, clinicians, health systems, and the public. The dynamic nature of this technology creates unique challenges to evaluating safety and efficacy and minimizing harms. In response, regulators have proposed an approach that would shift more responsibility to MLPA developers for mitigating potential harms. To be effective, this approach requires MLPA developers to recognize, accept, and act on responsibility for mitigating harms. In interviews of 40 MLPA developers of health care applications in the United States, we found that a subset of ML developers made statements reflecting moral disengagement, representing several different potential rationales that could create distance between personal accountability and harms. However, we also found a different subset of ML developers who expressed recognition of their role in creating potential hazards, the moral weight of their design decisions, and a sense of responsibility for mitigating harms. We also found evidence of moral conflict and uncertainty about responsibility for averting harms as an individual developer working in a company. These findings suggest possible facilitators and barriers to the development of ethical ML that could act through encouragement of moral engagement or discouragement of moral disengagement. Regulatory approaches that depend on the ability of ML developers to recognize, accept, and act on responsibility for mitigating harms might have limited success without education and guidance for ML developers about the extent of their responsibilities and how to implement them.

Ethical and epistemic issues in the design and conduct of pragmatic stepped-wedge cluster randomized clinical trials.

Stepped-wedge cluster randomized trial (SW-CRT) designs are increasingly employed in pragmatic research; they differ from traditional parallel cluster randomized trials in which an intervention is delivered to a subset of clusters, but not to all. In a SW-CRT, all clusters receive the intervention under investigation by the end of the study. This approach is thought to avoid ethical concerns about the denial of a desired intervention to participants in control groups. Such concerns have been cited in the literature as a primary motivation for choosing SW-CRT design, however SW-CRTs raise additional ethical concerns related to the delayed implementation of an intervention and consent. Yet, PCT investigators may choose SW-CRT designs simply because they are concerned that other study designs are infeasible. In this paper, we examine justifications for the use of SW-CRT study design, over other designs, by drawing on the experience of the National Institutes of Health's Health Care Systems Research Collaboratory (NIH Collaboratory) with five pragmatic SW-CRTs. We found that decisions to use SW-CRT design were justified by practical and epistemic reasons rather than ethical ones. These include concerns about feasibility, the heterogeneity of cluster characteristics, and the desire for simultaneous clinical evaluation and implementation. In this paper we compare the potential benefits of SW-CRTs against the ethical and epistemic challenges brought forth by the design and suggest that the choice of SW-CRT design must balance epistemic, feasibility and ethical justifications. Moreover, given their complexity, such studies need rigorous and informed ethical oversight.

A systematic review and meta-analysis of pregabalin preclinical studies.

Despite large efforts to test analgesics in animal models, only a handful of new pain drugs have shown efficacy in patients. Here, we report a systematic review and meta-analysis of preclinical studies of the commercially successful drug pregabalin. Our primary objective was to describe design characteristics and outcomes of studies testing the efficacy of pregabalin in behavioral models of pain. Secondarily, we examined the relationship between design characteristics and effect sizes. We queried MEDLINE, Embase, and BIOSIS to identify all animal studies testing the efficacy of pregabalin published before January 2018 and recorded experimental design elements addressing threats to validity and all necessary data for calculating effect sizes, expressed as the percentage of maximum possible effect. We identified 204 studies (531 experiments) assessing the efficacy of pregabalin in behavioral models of pain. The analgesic effect of pregabalin was consistently robust across every etiology/measure tested, even for pain conditions that have not responded to pregabalin in patients. Experiments did not generally report using design elements aimed at reducing threats to validity, and analgesic activity was typically tested in a small number of model systems. However, we were unable to show any clear relationships between preclinical design characteristics and effect sizes. Our findings suggest opportunities for improving the design and reporting of preclinical studies in pain. They also suggest that factors other than those explored in this study may be more important for explaining the discordance between outcomes in animal models of pain and those in clinical trials.

Assessment of Pregabalin Postapproval Trials and the Suggestion of Efficacy for New Indications: A Systematic Review.

Importance: After a drug receives regulatory approval, researchers often pursue small, underpowered trials, called exploratory trials, aimed at testing additional indications. If favorable early findings from exploratory trials are not promptly followed by confirmatory trials, then physicians, patients, and payers can be left uncertain about a drug's clinical value (clinical agnosticism). Such findings may encourage the off-label use of ineffective drugs. Objective: To characterize the relationship between exploratory and confirmatory postapproval trials for the blockbuster drug, pregabalin (Lyrica). Evidence Review: Ovid MEDLINE and Embase databases were used to identify clinical trials published prior to January 2018 and that tested the efficacy of pregabalin for nonapproved indications. Indications, trial outcomes, publication dates, and trial design elements were recorded. Time elapsed was calculated between the generation of clinical agnosticism about pregabalin (ie, publications reporting positive or inconclusive evidence of efficacy on a primary endpoint) and it being addressed (publication of at least 1 confirmatory trial in the same indication, regardless of outcome). Findings: There were 238 trials identified that tested the efficacy of pregabalin in at least 33 indications; 5 indications eventually received European Medicines Agency and/or US Food and Drug Administration marketing approval. Sixty-seven percent (22 of 33) of first publications for new indications may have generated clinical agnosticism. Of those indications with at least 5 years of follow-up, 63% (17 of 27) may have generated agnosticism that was not addressed in confirmatory trials within 5 years. As pregabalin development expanded from indications that received regulatory approval to other indications, the linkage of exploratory to confirmatory trial publication diminished. Conclusions and Relevance: After initial approval, exploratory evidence suggesting the value of pregabalin for new indications often went unconfirmed for extended periods of time. Poor coordination between exploratory and confirmatory testing may represent an important vehicle through which off-label prescription is recommended in clinical practice guidelines and encouraged in the absence of confirmatory trial evidence.

A meta-analysis of threats to valid clinical inference in preclinical research of sunitinib.

Poor study methodology leads to biased measurement of treatment effects in preclinical research. We used available sunitinib preclinical studies to evaluate relationships between study design and experimental tumor volume effect sizes. We identified published animal efficacy experiments where sunitinib monotherapy was tested for effects on tumor volume. Effect sizes were extracted alongside experimental design elements addressing threats to valid clinical inference. Reported use of practices to address internal validity threats was limited, with no experiments using blinded outcome assessment. Most malignancies were tested in one model only, raising concerns about external validity. We calculate a 45% overestimate of effect size across all malignancies due to potential publication bias. Pooled effect sizes for specific malignancies did not show apparent relationships with effect sizes in clinical trials, and we were unable to detect dose-response relationships. Design and reporting standards represent an opportunity for improving clinical inference.

Patient time costs and out-of-pocket costs in hepatitis C.

BACKGROUND: Hepatitis C virus (HCV) infection is associated with substantial costs to patients, their caregivers and society. AIMS: We evaluated time costs (time spent seeking healthcare) and out-of-pocket (OOP) costs for patients with HCV and their caregivers. METHODS: We measured costs for 738 HCV outpatients in a tertiary-care clinic using a patient-completed questionnaire. Time and OOP costs were compared across disease stages and sociodemographic categories. We examined the association between cost and disease stage using linear regression adjusting for age, gender, marital status, education, income and Index of Coexistent Disease (ICED) comorbidity score. Costs were expressed in 2007 Canadian dollars. RESULTS: The mean annual time cost per patient was $2136 (98 h), and ranged from $281 (18 h) in individuals who had cleared the virus to $9416 in transplant recipients (420 h). Caregiver costs were reported in 10% of patients. The mean annual OOP cost per patient was $1326. Patients receiving active treatment and those with late-stage disease spent $2500-2800 per year on HCV-related healthcare, approximately 7% of their annual income. Patients who had cleared the virus had the lowest time and OOP costs. Low income and unemployed patients had higher costs. CONCLUSIONS: In HCV-infected individuals, OOP and time costs represent a significant economic burden and fall disproportionately upon those least able to afford them. The lower cost burden among those who were successfully treated suggests that wider use of antiviral therapy may reduce economic burden in addition to improving health outcomes.

Health utilities and psychometric quality of life in patients with early- and late-stage hepatitis C virus infection.

BACKGROUND AND AIM: Hepatitis C virus (HCV) infection is associated with impairment in health-related quality of life (HRQOL). The purpose of this study was to evaluate HRQOL across the HCV disease spectrum using preference-based (utility) and non-preference-based (psychometric) methods, adjusting for sociodemographic factors and co-morbidity. METHODS: Hepatitis C virus patients (n = 751) were recruited from several tertiary care settings in Vancouver, Canada for this observational, cross-sectional cohort study. Patients completed the Health Utilities Index Mark 2/3, a self-administered time trade-off utility instrument, and the Hepatitis Quality of Life Questionnaire (SF-36 with HCV-specific items). We examined the association between HRQOL and disease stage using linear regression adjusting for age, education, marital status, income, and co-morbidities. RESULTS: Utility scores were low across disease stage and instrument, ranging from 0.51 to 0.80. On the SF-36, the mean Physical Component Summary score ranged from 37.2 to 49.2 across disease stage, and the Mental Component Summary score ranged from 39.7 to 45.7 (United States norms = 50). In general, patients with viral clearance had the highest scores, and those with late-stage disease (cirrhosis, liver cancer) had the lowest. Multivariable linear regression showed that the effect of disease stage was modest overall. Increasing age, lower income, unattached marital status, and high comorbidity were strongly associated with impairment in HRQOL. CONCLUSIONS: The effect of stage of disease on HRQOL is modest, although viral clearance is associated with higher HRQOL. HCV patients' HRQOL is strongly associated with concomitant illness and sociodemographic factors.

Untapped ethical resources for neurodegeneration research.

BACKGROUND: The research community has a mandate to discover effective treatments for neurodegenerative disorders. The ethics landscape surrounding this mandate is in a constant state of flux, and ongoing challenges place ever greater demands on investigators to be accountable to the public and to answer questions about the implications of their work for health care, society, and policy. METHODS: We surveyed US-based investigators involved in neurodegenerative diseases research about how they value ethics-related issues, what motivates them to give consideration to those issues, and the barriers to doing so. Using the NIH CRISP database we identified 1,034 researchers with relevant, active grants and invited them to complete an online questionnaire. We received 193 responses. We used exploratory factor analysis to transform individual survey questions into a smaller set of factors, and linear regression to understand the effect of key variables of interest on the factor scores. RESULTS: Ethics-related issues clustered into two groups: research ethics and external influences. Heads of research groups viewed issues of research ethics to be more important than the other respondents. Concern about external influences was related to overall interest in ethics. Motivators clustered into five groups: ensuring public understanding, external forces, requirements, values, and press and public. Heads of research groups were more motivated to ensure public understanding of research than the other respondents. Barriers clustered into four groups: lack of resources, administrative burden, relevance to the research, and lack of interest. Perceived lack of ethics resources was a particular barrier for investigators working in drug discovery. CONCLUSIONS: The data suggest that senior level neuroscientists working in the field of neurodegeneration (ND), and drug discovery specifically, are motivated to consider ethics issues related to their work, but the perceived lack of ethics resources thwarts their efforts. With bioethics centres at more than 50% of the institutions at which these respondents reside, the neuroscience and bioethics communities appear to be disconnected. Dedicated ethical, legal and social implications (ELSI) programs, such as those fully integrated into genetics and regenerative medicine, provide models for achieving meaningful partnerships not yet adequately realized for scholars and trainees interested in drug discovery for ND.

Reducing barriers to ethics in neuroscience.

Ethics is a growing interest for neuroscientists, but rather than signifying a commitment to the protection of human subjects, care of animals, and public understanding to which the professional community is engaged in a fundamental way, interest has been consumed by administrative overhead and the mission creep of institutional ethics reviews. Faculty, trainees, and staff (n = 605) in North America whose work involves brain imaging and brain stimulation completed an online survey about ethics in their research. Using factor analysis and linear regression, we found significant effects for invasiveness of imaging technique, professional position, gender, and local presence of bioethics centers. We propose strategies for improving communication between the neuroscience community and ethics review boards, collaborations between neuroscientists and biomedical ethicists, and ethics training in graduate neuroscience programs to revitalize mutual goals and interests.