Artificial intelligence and gender equity: An integrated approach for health professional education

Medical Education, Journal Year: 2025, Volume and Issue: unknown

Published: March 10, 2025

Abstract Introduction As artificial intelligence (AI) increasingly integrates into health workplaces, evidence suggests AI can exacerbate gender inequity. Health professional programmes have a role to play in ensuring graduates grasp the challenges facing working an AI‐mediated world. Approach Drawing from feminist scholars and empirical evidence, this conceptual paper synthesises current future ways which compounds inequities and, response, proposes foci for integrated approach teaching about equity. Analysis We propose three concerns. Firstly, multiple literature reviews suggest that divide is embedded within technologies both process (AI development) product output) perspectives. Next, there emerging reinforcing already entrenched workforce inequities, where certain types of roles are seen as being domain genders. Finally, may disassociate professionals' interactions with embodied, agentic patient by diverting attention gendered digital twin. Implications Responding these concerns not simply matter bias but needs promote understanding sociotechnical phenomenon. Healthcare curricula could usefully provide clinically relevant educational experiences illustrate how intersects inequitable knowledge practices. Students be directed to: (1) explore doubts when AI‐generated data or decisions; (2) refocus on caring through prioritising embodied connections; (3) consider negotiate workplaces time AI. Conclusion The intersection equity provides accessible, illustrative case changing practices potential embed inequity education might respond.

Language: Английский

---

When I say … artificial intelligence

Medical Education, Journal Year: 2024, Volume and Issue: unknown

Published: April 20, 2024

Artificial intelligence (AI) looms large in popular imagination, from Shelley's Frankenstein to Kubrick's HAL9000. But AI also has been a significant topic of research for computer scientists, health informaticians, educational technologists and social scientists over many decades. This broad engagement means that it is often unclear what the term refers to; definitions vary markedly across fields endeavour.1 While generative other language models (LLMs) have grabbed headlines, there lack clarity about might do, both broadly society specifically within professional settings. In order clarify different modes thinking AI, we categorise conceptualisations associated definitions. These are not exclusive nor one approach better or more correct. Rather, they serve purposes. We start with technical (what is) move capability does), before exploring relational (how works system). end by introducing 'AI interaction', conceptualisation may be particularly valuable practice education. Technical (describing an provide most straightforward approach. For example, two main algorithmic labels AI: pre-set approaches ('expert systems', which rely on known rules) pattern recognition ('machine learning' trained existing datasets). The latter use LLMs like ChatGPT; employ machine learning, where statistical weighting allows software predict users' desired patterns text, image audio. those who feel these technologies too mysterious (and indeed magical), you wish think them as highly sophisticated predictive text generators, ones completing your sentences smartphone. allow people understand how work, but limited defining contributes particular task situation. From its early inceptions, defined capabilities.2 1980, Searle3 classically separated out 'weak' acts technological tool under control humans, 'strong' 'can literally said cognitive states'. Strong AIs—or conscious machines—remain stuff science fiction. Current AIs generally designed tools therefore capable doing (rather than underlying algorithms). Medical education scholars tend towards Tolsgaard et al4 cite Oxford dictionary, describing capabilities '… perform tasks normally requiring human intelligence, such visual perception, speech recognition, [and] decision-making …'. Indeed, focus decision-making. A classic undergraduate textbook defines technology seeks identify 'best possible action situation'.5 Moreover, their specific capabilities, classifiers pathology driverless cars. useful because clearly delimit AI's scope, decontextualised manner. Relational address work together situations. draw theoretical foundations studies (STS), position all being actors. this perspective, can understood relational, sense meaning function found ways put use.6 Along line, Johnson Verdicchio1 propose system' sociotechnical ensembles [which are] combinations artefacts, behaviour, arrangements meaning'. type definition makes when pedagogy medical Our students must learn engage complex messy world care—full ambiguity, flesh-and-blood experiences rife emotions—and simultaneously around AI.6 framing conceptualises situational dynamic rather fixed artefact. Importantly, foregrounds issue ambiguity. And surprisingly ambiguous. 2019 ethnography radiologists' responses introduction bone age assessments, suggests increased doctors' uncertainty amidst burden deep care patients. One participant said: 'Sometimes (the AI) would give me ages make re-think I go, "Ok, maybe". adjust closer assessment). sometimes, think, "This way off". So don't know. just know …'.7 isolated: 2024 similar challenges underlines need prepare our graduates reality patient promise.8 interaction'9 concept focuses indeterminate relationship between AI. concerned happens moment help managing realities practice. To calculator considered 4-year-old trust calculator's outputs without any knowing whether right wrong. suggest, therefore, child uses calculator, interaction. adult not. Thus, 'AI' dependent specifications even technology. formal interaction when: 'in context interaction, computational artefact provides judgement inform optimal course cannot traced'.9 matter person producing answer trace way, much using it, are. further if doctor asks LLM treatment expert, will already familiar sources drawing from. layperson, take trust; at moment, no knowing, manual hand, looking inside 'black box'. words, involve leap faith. uncomfortable. How contribute interaction? thinks vice versa? argue immaterial: centres faith time. It does not—or do. how, something trust. ethnographic examples describe radiologists pathologists alike practices were fundamentally altered unexpected was (or distrust). helpful understanding practice, learning software, reliability underpinnings. interaction' introduces definitional level, role doubt,9 contextualised nature grapple ethical implications working eschews unanswerable questions 'is accurate?' ask ourselves critically consider meaningful, harmful. attunes us develop distinctly part interactions. Thinking interactions highlight curricula emphasise discriminate quality9 underline compassionate AI-mediated world.6 direct types hence fundamental problems Such things done through simulations case complexity, ambiguity clinical responsibility. counter characterisation necessarily superior rational enable critical foregrounding co-produced humans machines situated messiness Margaret Beaman led writing primary draft; Rola Ajjawi contributed reviewed edited final version. Open access publishing facilitated Deakin University, Wiley - University agreement via Council Australian Librarians. Data sharing applicable article new data created analyzed study.

Language: Английский

---

Advancements in pathology: Digital transformation, precision medicine, and beyond

Journal of Pathology Informatics, Journal Year: 2024, Volume and Issue: 16, P. 100408 - 100408

Published: Nov. 19, 2024

Pathology, a cornerstone of medical diagnostics and research, is undergoing revolutionary transformation fueled by digital technology, molecular biology advancements, big data analytics. Digital pathology converts conventional glass slides into high-resolution images, enhancing collaboration efficiency among pathologists worldwide. Integrating artificial intelligence (AI) machine learning (ML) algorithms with improves diagnostic accuracy, particularly in complex diseases like cancer. Molecular pathology, facilitated next-generation sequencing (NGS), provides comprehensive genomic, transcriptomic, proteomic insights disease mechanisms, guiding personalized therapies. Immunohistochemistry (IHC) plays pivotal role biomarker discovery, refining classification prognostication. Precision medicine integrates pathology's findings individual genetic, environmental, lifestyle factors to customize treatment strategies, optimizing patient outcomes. Telepathology extends services underserved areas through remote pathology. Pathomics leverages analytics extract meaningful from advancing our understanding therapeutic targets. Virtual autopsies employ non-invasive imaging technologies revolutionize forensic These innovations promise earlier diagnoses, tailored treatments, enhanced care. Collaboration across disciplines essential fully realize the transformative potential these advancements practice research.

Language: Английский

---