AI pitfalls and what not to do: mitigating bias in AI.

Various forms of artificial intelligence (AI) applications are being deployed and used in many healthcare systems. As the use of these applications increases, we are learning the failures of these models and how they can perpetuate bias. With these new lessons, we need to prioritize bias evaluation and mitigation for radiology applications; all the while not ignoring the impact of changes in the larger enterprise AI deployment which may have downstream impact on performance of AI models. In this paper, we provide an updated review of known pitfalls causing AI bias and discuss strategies for mitigating these biases within the context of AI deployment in the larger healthcare enterprise. We describe these pitfalls by framing them in the larger AI lifecycle from problem definition, data set selection and curation, model training and deployment emphasizing that bias exists across a spectrum and is a sequela of a combination of both human and machine factors.

Characteristics of Radiologists Serving as Medical Malpractice Expert Witnesses for Defense Versus Plaintiff.

PURPOSE: Previous studies have reported higher qualification characteristics for anesthesiologists, neurosurgeons, orthopedic surgeons, and otolaryngologists serving as defense (versus plaintiff) medical malpractice expert witnesses. We assessed such characteristics for radiologist expert witnesses. METHODS: Using the Westlaw legal research database, we identified radiologists serving as experts in all indexed medical malpractice cases between 2010 and 2019. Online databases were used to identify years of practice experience and scholarly bibliometrics. Using Medicare claims, individual radiologist practice types and mixes were ascertained. Radiologists testifying at least once each for defense and plaintiff were excluded from our defense-only versus plaintiff-only comparative analysis. RESULTS: Initial Boolean searches yielded 1,042 potential cases; subsequent manual review identified 179 radiologists testifying in 231 lawsuits: 143 testified in one case (58 defense, 85 plaintiff) and 36 testified in multiple cases (10 defense-only, 14 plaintiff-only, 12 both). The 68 defense-only experts had fewer years of practice experience than the 99 plaintiff-only experts (28.3 versus 31.8 years, P = .02), but the two groups were otherwise similar in both practice type (44.6% versus 54.9% academic, P = .62) and mix (63.8% versus 65.8% practiced as subspecialists, P = .37) and as well as numbers of publications (60.5 versus 62.8, P = .86), citations (1,994.1 versus 2,309.2, P = .56), and h-indices (17.2 versus 16.8, P = .89). CONCLUSIONS: In contrast to other specialists, radiologists serving as medical malpractice expert witnesses for defense and plaintiff display similar qualifications across various characteristics. Published practice parameter guidelines and experts' ability to blindly review archived original images might together explain this interspecialty discordance.

When Artificial Intelligence Models Surpass Physician Performance: Medical Malpractice Liability in an Era of Advanced Artificial Intelligence.

It seems inevitable that diagnostic and recommender artificial intelligence models will ultimately reach a point when they outperform human clinicians. Just as antibiotics displaced a host of medicinals for treating infections, the superior performance of such models will force their adoption. This article contemplates certain ethical and legal implications bearing on that adoption, especially because they involve a clinician's exposure to allegations of malpractice. The article discusses four relevant considerations: (1) the imperative of using explainable artificial intelligence models in clinical care, (2) specific strategies for diminishing liability when a clinician agrees or disagrees with a model's findings or recommendations but the patient nevertheless experiences a poor outcome, (3) relieving liability through legislation or regulation, and (4) comprehending such models as "persons" and therefore as potential defendants in legal proceedings. We conclude with observations on clinician-vendor relationships and argue that, although advanced artificial intelligence models have not yet arrived, clinicians must begin considering their implications now.

Ethical considerations in artificial intelligence.

With artificial intelligence (AI) precipitously perched at the apex of the hype curve, the promise of transforming the disparate fields of healthcare, finance, journalism, and security and law enforcement, among others, is enormous. For healthcare - particularly radiology - AI is anticipated to facilitate improved diagnostics, workflow, and therapeutic planning and monitoring. And, while it is also causing some trepidation among radiologists regarding its uncertain impact on the demand and training of our current and future workforce, most of us welcome the potential to harness AI for transformative improvements in our ability to diagnose disease more accurately and earlier in the populations we serve.

Failures of foreseeability: Risk management considerations in reducing allegations of sexual violence in psychiatric units.

This article begins with a brief discussion of findings on causal factors leading to allegations of sexual violence in health care facilities and then offers the author's account of 4 such cases that he reviewed, 3 of which occurred in psychiatric units. These cases show remarkably similar variables, especially involving decisions to allow male and female patients to commingle, the inadequate physical layout of the units, poor or absent video surveillance, and staff unacquainted with institutional policies on patient safety or refusing to enforce relevant rules. These variables arguably amount to "failures of foreseeability" that reasonably cautious health care personnel should recognize as facilitating or enabling sexual violence. As such, the proactive message of this article for health care risk management urges critical and robust attention paid to a unit's environmental/physical design as well as to performance factors among personnel so as to prevent sexual attacks and diminish the probability of malpractice actions.

Preventing sexual attacks in healthcare facilities: risk management considerations.

Reports or allegations of sexual attacks in healthcare facilities are extremely upsetting and sometimes not given the attention they deserve. In June 2011, the United States Government Accountability Office (GAO) issued a remarkable report on sexual attacks occurring in Veterans Affairs (VA) health facilities that not only raised awareness of the magnitude of the problem but that detailed numerous system weaknesses in VA facilities that might have enabled such attacks. This article discusses some of the GAO's findings as well as other instances of sexual attacks, such as occurred in the criminal prosecution of Paul Serdula, a former health professional who might have sexually assaulted hundreds of women. Some of Serdula's victims have subsequently sued in civil court, charging Serdula's employers with lack of supervision and raising the possibility of serial sexual attacks such as his evolving into large-scale patient safety disasters. This article will review certain ethical and legal considerations bearing on the liability of healthcare facilities in which sexual attacks are alleged to have occurred. Following a discussion of how two courts have used the legal construct of "foreseeability" in determining a healthcare facility's liability when an employee is charged with sexual assault, the article will conclude with a host of patient safety recommendations aimed at discouraging or deterring the occurrence of sexual attacks.

Ethical perspectives on knowledge translation in rehabilitation.

Although the literature on the ethical dimensions of knowledge creation, use, and dissemination is voluminous, it has not particularly examined the ethical dimensions of knowledge translation in rehabilitation. Yet, whether research is done in a wet lab or treatments are provided to patients in therapeutic settings, rehabilitation professionals commonly use (as well as create) knowledge and disseminate it to peers, patients, and various others. This article will refer to knowledge creation, use, and transfer as knowledge translation and examine some of its numerous ethical challenges. Three ethical dimensions of knowledge translation will particularly attract our attention: (1) the quality of knowledge disseminated to rehabilitationists; (2) ethical challenges in being too easily persuaded by or unreasonably resistant to putative knowledge; and (3) organizational barriers to knowledge translation. We will conclude with some recommendations on facilitating the ethical soundness of knowledge translation in rehabilitation.

Stroke rehabilitation and the phenomenological reconstitution of the self.

This commentary will apply the notions of constitution and "phenomenological introspection" developed by phenomenology's founder Edmund Husserl to certain themes in Sharon Kaufman's 1988 essay, "Toward a Phenomenology of Boundaries in Medicine: Chronic Illness Experience in the Case of Stroke." The article will discuss how phenomenological analysis can provide important therapeutic insights about the lived experiences of stroke patients and their caregivers, especially as that experience is shaped in the immediate aftermath of a serious stroke. This article will also argue that phenomenology in and by itself is woefully inadequate for producing the kind of self-knowledge and political will needed to produce a socioeconomic environment that reasonably accommodates the needs of stroke patients. The article will end with a brief discussion of how an Eastern, particularly Buddhist, conception of the self is considerably more disability friendly than the one Westerners (phenomenologically) "constitute" and how the former's more realistic understanding of the trajectory of human functioning and its inevitable decline over a lifespan offers a superior platform for developing disability policy and care than its Western counterpart.

Speaking up in case management, part II: implementing speaking up behaviors.

OBJECTIVES: Case managers will occasionally witness colleagues from their own or other healthcare disciplines providing care in ways that frankly deviate from or violate standards of care, rules, regulations, policies, and procedures. This article discusses the case manager's ethical obligation to speak up in such instances, as well as lists strategies and techniques that facilitate and enhance professional communications around "speaking up." PRIMARY PRACTICE SETTINGS: All case management work environments. FINDINGS/CONCLUSIONS: When healthcare professionals practice in ways that deviate from established policies, procedures, rules, regulations, and standards of care, they jeopardize the safety and welfare of their clients and, often, their employing institutions. Unfortunately, professionals often remain silent about such errors and deviations for fear that they might suffer retaliation if they speak up or because they do not anticipate a constructive or positive organizational response. Nevertheless, organizations and employees who recognize the value of speaking up can use a variety of strategies described in this article that enable speaking up to become a professional and organizational reality. IMPLICATIONS FOR CASE MANAGEMENT PRACTICE: Because they practice in the midst of multidisciplinary care, case managers have a marked opportunity to witness any number of practice standards being ignored or violated. Case managers need to develop communication skills and organizational savvy regarding how to approach and remedy such situations, especially when deviations from standards of care have become routine or "normalized." Critical elements in speaking up include cultivating a method to discern the moral perils of remaining silent, learning communication techniques to conduct emotionally challenging conversations, and building organizational leadership that recognizes the value of speaking up and supports safe harbors for employees who speak up appropriately.

Speaking up in case management, part I: ethical and professional considerations.

OBJECTIVES: Case managers will occasionally witness colleagues from their own or other healthcare disciplines providing care in ways that frankly deviate from or violate standards of care, rules, regulations, policies, and procedures. This article will discuss the case manager's ethical obligation to speak up in such instances as well as list strategies illustrated in this article by a three-color flag system that classifies poor, better, and best responses to ethically challenging situations. PRIMARY PRACTICE SETTINGS: All case management work environments. FINDINGS/CONCLUSIONS: When health professionals practice in ways that deviate from established policies, procedures, rules, regulations, and standards of care, they jeopardize the safety and welfare of their clients and, often, their employing institutions. Although speaking up is often difficult due to fears of retaliation or an organizational nonresponse, the fundamental ethical obligation of case managers is to protect and further their client's health and welfare rather than their professional self-interests. Consequently, the ethically conscientious case manager will pursue strategies whereby weaknesses or latent hazards that might compromise a client's care can be remediated. IMPLICATIONS FOR CASE MANAGEMENT PRACTICE: Because they practice in the midst of multidisciplinary care, case managers have a marked opportunity to witness any number of practice standards being ignored or violated. Case managers need to cultivate a heightened moral awareness and response to such situations, especially when deviations from standards of care have become routine or "normalized." Critical elements in speaking up include cultivating a method to discern the moral perils of remaining silent, learning communication techniques to conduct emotionally challenging conversations, and building organizational leadership that recognizes the value of speaking up and supports safe harbors for employees who speak up appropriately.

Overriding the Jehovah's Witness patient's refusal of blood: a reply to Cahana, Weibel, and Hurst.

This article is a response to a survey on moral reasoning among Swiss health professionals that appeared in a recent issue of this journal. The authors of that survey inquired whether or not their respondents would give a blood transfusion to a Jehovah's Witness patient who clearly refused it. A substantial number of the respondents answered that they would override the patient's refusal and give the transfusion. The present article examines the two ethical rationales that were offered to explain the overriding respondents' answers and argues that neither one is ethically acceptable. It concludes with an account of the phenomenon of "motivated reasoning" that, so it is argued, better explains why the overriders would refuse to honor the Jehovah's Witness patient's transfusion refusal.

Toward a more empathic relationship in pain medicine.

Specialists in pain medicine commonly experience psychological assaults on their self-esteem, especially from patients who seem unreasonably demanding, overly critical, or threatening. This article will discuss how these challenges can trigger a professional's self-protective and defensive coping mechanisms that, in turn, can provoke decidedly unempathic responses. Situations that compromise empathy, however, can be particularly worrisome in health delivery practices like pain medicine that are highly relational and that seek to use relationships therapeutically. The article will therefore conclude with two sets of strategies that might be useful in managing these uncomfortable situations more empathically. The first set will focus on certain pragmatics of empathy skill development. The second will discuss the Eastern notion of "bare attention" as an ideal form of empathic engagement that can also counteract an unhealthy degree of defensiveness when self-esteem is threatened.