• IMA sites
  • IMAJ services
  • IMA journals
  • Follow us
  • Alternate Text Alternate Text
עמוד בית
Tue, 17.11.26

November 2026


OPENING
Alex Mintz PhD, Arieh Riskin MD PhD, Yehuda Shoenfeld MD FRCP MaACR

Navigating a complex clinical case is among the greatest challenges in medicine. Previous issues of the Israel Medical Association Journal (IMAJ) have published two accounts that show why. In the first account, Peled et al. [1] followed a single patient for 26 years. The patient presented with unstable angina and three-vessel disease treated by bypass grafting in 1993. He presented with a type A aortic dissection in 1999. In addition, He underwent a transcatheter aortic valve implantation. Later she presented with progressive ventricular dilatation and heart failure.

The authors reviewed ethical principles against the background of the realities of practicing medicine and of the enormous advances in therapeutics with the narrative deliberately organized around clauses of the Hippocratic Oath [1]. One of those clauses is important for our purposes: "I will not be ashamed to say, 'I know not,' nor will I fail to call in my colleagues when the skills of another are needed for a patient’s recovery". Written somewhere between the 5th and 3rd centuries BCE, this quote is a remarkably exact statement of a problem in modern terms: clinicians' capacity to recognize the limits of their own judgment and to act on that recognition.

REVIEWS
Alex Mintz PhD, Arieh Riskin MD PhD

Cognitive biases–systematic deviations from rational judgment–are implicated in most diagnostic errors in medicine. They arise from heuristics: mental shortcuts that are often beneficial in clinical settings and adaptive in most contexts but produce predictable errors when applied inappropriately. In this review, we integrate cognitive psychology, research on heuristics, dual-process theory, empirical studies of diagnostic error in medicine, and educational research on debiasing interventions to examine the theoretical foundations of cognitive heuristics and biases. Its uniqueness is in the clinical manifestations and mechanisms of the most consequential biases, and the evidence-based debiasing strategies available at the individual and system levels.

Heuristics and cognitive biases arise from System 1 reasoning (fast, automatic, pattern-based) and are amplified when System 2 monitoring fails to detect and correct erroneous outputs. Key biases that are analyzed include anchoring bias, availability bias, confirmation bias, premature closure, the representativeness heuristic, overconfidence bias, framing bias, commission and omission bias, status quo bias, loss aversion, ignoring critical information, locking on a preferred alternative, shooting from the hip, and wishful thinking. Each bias has distinct clinical manifestations with documented consequences for diagnostic accuracy and patient safety.

Yarden Riskin MSc, Arieh Riskin MD PhD

Metacognition, which is the capacity to monitor, evaluate, and regulate one’s own cognitive processes, is an important determinant of clinical decision quality. Clinicians who fail to monitor their reasoning are more susceptible to the systematic cognitive biases that contribute to many diagnostic errors. In this narrative review, we integrated theoretical models of metacognition, presented research on intuitive and analytical clinical reasoning, showed empirical studies of cognitive bias in medicine, and described educational interventions targeting metacognitive skills. The review is built on the monitoring–control framework in which monitoring is what clinicians notice about their own reasoning and control is what they do in response. Two monitoring signals are most directly relevant to diagnosis: confidence, which is often poorly calibrated to actual accuracy, and the feeling of rightness, which determines whether an initial diagnosis is reconsidered. Using clinical examples of anchoring, premature closure, availability bias, overconfidence, and confirmation bias, we specified what the clinician needed to monitor and the control action for each bias.

We described a diagnostic time-out, an active search for disconfirming evidence, a discriminating test, or consultation, that monitoring should have triggered. Structured reflection and calibration feedback have shown benefit, whereas bias education alone has shown limited effect on diagnostic accuracy. Metacognitive capacity is a modifiable, teachable competency. Its development is a patient safety imperative, and medical education programs should incorporate explicit metacognitive training, which are linked to concrete clinical cues and actions, as a core component of clinical reasoning curricula.

Arieh Riskin MD PhD, Ayala Gover MD

Sociodemographic biases, which are related to gender, race and ethnicity, socioeconomic status, age, weight, and sexual orientation, permeate medical decision-making and contribute to health disparities that affect diagnosis, treatment allocation, and patient outcomes. In this narrative review, we address sociodemographic bias in general, treating implicit bias as one of several mechanisms. Drawing on peer-reviewed studies, systematic reviews, and meta-analyses published mainly between 2010 and 2025, we examined evidence for these biases, their mechanisms, bias in artificial intelligence (AI) and large language model (LLM) clinical tools, and interventions for bias reduction. We integrated literature usually reviewed separately and distinguished evidence of outcome disparities from evidence that directly measures clinician bias. Robust evidence documents gender disparities in cardiovascular care and pain management, racial and ethnic disparities in procedural access and analgesia, and socioeconomic gradients in treatment intensity.

Experimental studies that hold the clinical presentation constant confirm that these patient characteristics can influence clinicians’ decisions. LLMs replicate and sometimes amplify these biases. Up to 91.7% of reviewed studies identified biases in medical AI, with LGBTQ+ patients receiving mental health referrals six to seven times more frequently than clinically warranted. Mechanisms operate at the clinician level (prototype-based reasoning, implicit and explicit attitudes, cognitive load), the knowledge level (non-representative research and training data), and the system level (structural inequity). Multilevel interventions, such as individual bias awareness, structural protocols, and algorithmic fairness techniques, show promise but require sustained institutional commitment. Equity in healthcare is not only an ethical imperative but a clinical and scientific one.

Maxim Glebov MD, Alex Mintz PhD

Generative artificial intelligence (GenAI), particularly large language models, is increasingly used in medicine to support diagnosis, risk prediction, treatment planning, workflow optimization, documentation, and clinical decision-making. Its ability to process large amounts of data and identify complex patterns may greatly improve patient care and assist clinicians in increasingly data-rich healthcare environments. However, GenAI systems are not inherently neutral, and their use in medicine creates a broad spectrum of clinical risks.

These risks include biased or unrepresentative data, poor generalizability, inconsistent recommendations, inaccurate labels, inappropriate outcome proxies, opacity, automation bias, erosion of clinical expertise, hallucinations, fabricated evidence, cybersecurity vulnerabilities, privacy concerns, and unclear accountability. In clinical settings, these risks may lead to misdiagnosis, over- or undertreatment, unequal allocation of resources, propagation of false evidence, inappropriate reliance on automated outputs, and loss of trust. In this review, we discuss the major clinical risks and mechanisms of harm associated with GenAI in medicine, provide clinical examples, and suggest strategies to reduce harm. GenAI should be viewed as a powerful clinical tool, but not as a replacement for human judgment, professional accountability, or compassionate medical care.

Moshe Y. Flugelman MD, Arieh Riskin MD PhD

Clinical reasoning, the cognitive process by which physicians gather, interpret, and synthesize information to reach diagnostic and therapeutic decisions, is the cornerstone of medical practice. However, it is rarely taught explicitly in medical curricula and the consequences of this omission are substantial. In this narrative review, we integrate theoretical models of clinical cognition, the history and limitations of evidence-based guidelines, and emerging data on artificial intelligence (AI)-assisted clinical reasoning. Clinical reasoning operates through dual cognitive pathways: rapid, pattern-based System 1 and deliberate, analytical System 2 thinking.

This thinking is vulnerable to well-characterized cognitive biases responsible for most diagnostic errors. Guidelines, while representing a systematic approach to codifying evidence, carry inherent limitations when applied to individual patients. Large language models (LLMs) demonstrate diagnostic capabilities comparable to experienced clinicians in structured settings; however, they introduce new risks including automation bias and hallucination. The medical interview remains the foundational act of clinical reasoning and the primary vehicle for building the therapeutic trust that underpins all clinical decisions. In the era of AI, strong clinical reasoning skills are not diminished in importance; they are indispensable. Situation awareness, the integration of perception, comprehension, and projection, is the cognitive capacity that distinguishes effective clinical decision-making, and it is a capacity that neither algorithms nor guidelines can replace. Formal, longitudinal education in clinical reasoning must become a core component of medical training at all levels.

Varda Liberman PhD

Clinical decisions are influenced by how physicians represent uncertainty, interpret disagreement, combine assessments, and evaluate new evidence. These psychological processes can support effective decisions, but they can also cause plausible alternatives to be overlooked or discounted. In this conceptual framework, the author developed a different conceptual framework that brought together findings from cognitive and social psychology, linking them through how clinicians considered alternative explanations and judgments.

Support theory explains why possibilities receive different probability estimates depending on whether they are explicitly named or left unspecified. Naïve realism, the experience of one’s own judgments as objective reflections of reality, helps explain why clinicians may dismiss disagreement rather than examine its basis. Research on collaboration within dyads shows how independently formed assessments can be integrated, whereas groupthink illustrates how pressure for consensus may suppress alternative views. Biased assimilation explains how prior commitments can influence the evaluation of subsequent evidence.

Ayala Gover MD, Arieh Riskin MD PhD

Medical decision-making has historically been modeled as a rational, cognitive process. Substantial evidence now challenges this view, demonstrating that emotions, both those arising from the clinical encounter itself and those carried over from unrelated contexts, substantially influence clinical judgment, risk assessment, and treatment choices.

In this narrative review of foundational psychological theory and empirical studies in emotion science and clinical medicine, with emphasis on work published since 2000 and on recent systematic reviews, we examine how emotions affect medical decision-making, the psychological mechanisms involved as described by the appraisal tendency framework, and the evidence-based strategies for emotion regulation that can improve clinical reasoning. Emotions influence clinical decisions through multiple pathways: integral emotions, arising from the decision itself, directly shape risk perception and treatment intensity, while incidental emotions, carried over from unrelated events, systematically bias judgment.

Arieh Riskin MD PhD, Alex Mintz PhD

Beyond the individual cognitive biases of single clinicians, medical teams are subject to distinct group-level biases that emerge from social dynamics, hierarchical structures, and interpersonal interactions. These biases, including groupthink, group polarization, and conformity, as well as the less known polythink bias, are systematically amplified or attenuated by environmental factors such as incivility, psychological safety, and gratitude. In this narrative review, we integrate group psychology, organizational behavior, and empirical studies of medical team performance.

This group includes experimental randomized research on the effects of rudeness and gratitude on clinical outcomes to examine the theoretical foundations of group decision-making dynamics for medical teams, the major group-level biases and their clinical manifestations, the evidence on the social environment of clinical work, and the strategies available for reducing group biases in practice. Medical teams are structurally predisposed to groupthink, which is the premature consensus at the expense of critical analysis, because of hierarchical authority gradients, team cohesion, and time pressure.

Legal Disclaimer: The information contained in this website is provided for informational purposes only, and should not be construed as legal or medical advice on any matter.
The IMA is not responsible for and expressly disclaims liability for damages of any kind arising from the use of or reliance on information contained within the site.
© All rights to information on this site are reserved and are the property of the Israeli Medical Association. Privacy policy

2 Twin Towers, 35 Jabotinsky, POB 4292, Ramat Gan 5251108 Israel