Amelia Fiske
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Research Interests
Amelia Fiske has a total of 195 co-authors. Here are the top co-authors by number of shared publications:
- Stuart McLennan
Institute of History and Ethics in Medicine, Department of Preclinical Medicine, TUM School of Medicine and Health Technical University of Munich Munich Germany
17 shared publications - Alena Buyx
Institute of History and Ethics in Medicine, TUM School of Medicine and Health, Technical University of Munich, Munich, Germany
16 shared publications - Barbara Prainsack
University of Vienna
11 shared publications Alena Buyx
Institute of History and Ethics in Medicine, Department of Preclinical Medicine, TUM School of Medicine and Health, Technical University of Munich, Munich, Germany
9 shared publications- Leo Anthony Celi
Laboratory for Computational Physiology, Massachusetts Institute of Technology
9 shared publications - Bettina Maria Zimmermann
Institute of History and Ethics in Medicine, TUM School of Medicine and Health, Technical University of Munich, Munich, Germany
7 shared publications
Latest Publications
- Open Access
- 17 September 2026
- Published by Public Library of Science (PLoS) in PLOS Digital Health
- Vol. 5 (9) , e0001671
M3: Conversational LLMs simplify secure clinical data access, understanding, and analysis
Large-scale clinical databases offer opportunities for medical research, but their complexity creates barriers to effective use. The Medical Information Mart for Intensive Care (MIMIC-IV), one of the world’s largest open-source electronic health record databases, traditionally requires both SQL proficiency and clinical domain expertise. We introduce M3, a system that enables natural language querying of MIMIC-IV data through the Model Context Protocol. With a single command, M3 retrieves MIMIC-IV from PhysioNet, launches a local SQLite instance or connects to hosted BigQuery, and allows researchers to pose clinical questions in plain English. We evaluated M3 using samples from the EHRSQL 2024 benchmark with two language models. On one hundred answerable questions, the proprietary Claude Sonnet 4 achieved 94% accuracy and the open-weights gpt-oss-20B (deployable locally on consumer hardware) achieved 93%; on a matched sample of one hundred unanswerable questions, where correct behavior is to abstain rather than produce SQL, gpt-oss-20B correctly abstained on 69%. Both models translate natural language into SQL, execute queries against MIMIC-IV, and return structured results alongside the underlying query for verification. Error analysis revealed that most failures stemmed from complex temporal reasoning or ambiguous question phrasing rather than fundamental architectural limitations. The comparable performance of a smaller open-weights model demonstrates that privacy-preserving local deployment is viable for sensitive clinical data analysis. M3 lowers technical barriers to critical care data analysis and is designed with security measures including OAuth2 authentication, query validation, and audit logging.
- 17 September 2026
- Published by Public Library of Science (PLoS) in PLOS Digital Health
- Vol. 5 (9) , e0001671
Open AccessM3: Conversational LLMs simplify secure clinical data access, understanding, and analysis
Large-scale clinical databases offer opportunities for medical research, but their complexity creates barriers to effective use. The Medical Information Mart for Intensive Care (MIMIC-IV), one of the world’s largest open-source electronic health record databases, traditionally requires both SQL proficiency and clinical domain expertise. We introduce M3, a system that enables natural language querying of MIMIC-IV data through the Model Context Protocol. With a single command, M3 retrieves MIMIC-IV from PhysioNet, launches a local SQLite instance or connects to hosted BigQuery, and allows researchers to pose clinical questions in plain English. We evaluated M3 using samples from the EHRSQL 2024 benchmark with two language models. On one hundred answerable questions, the proprietary Claude Sonnet 4 achieved 94% accuracy and the open-weights gpt-oss-20B (deployable locally on consumer hardware) achieved 93%; on a matched sample of one hundred unanswerable questions, where correct behavior is to abstain rather than produce SQL, gpt-oss-20B correctly abstained on 69%. Both models translate natural language into SQL, execute queries against MIMIC-IV, and return structured results alongside the underlying query for verification. Error analysis revealed that most failures stemmed from complex temporal reasoning or ambiguous question phrasing rather than fundamental architectural limitations. The comparable performance of a smaller open-weights model demonstrates that privacy-preserving local deployment is viable for sensitive clinical data analysis. M3 lowers technical barriers to critical care data analysis and is designed with security measures including OAuth2 authentication, query validation, and audit logging.
- 3 July 2026
- Published by Taylor & Francis in Medicine, Conflict and Survival
- Vol. 42 (3) , 559-568
Global health at risk because of organized disinformation
- 3 July 2026
- Published by Taylor & Francis in Medicine, Conflict and Survival
- Vol. 42 (3) , 559-568
Global health at risk because of organized disinformation
- Open Access
- 20 May 2026
- Published by Elsevier in SSRN Electronic Journal
Not a Private Consultation: Third-Party Network Traffic in Commercial LLMs
Objective: This study presents an analysis of network traffic across four commercial LLM platforms to document which external domains they contact during routine use and to classify those domains by functional purpose.Design: Cross-sectional analysis of HTTP/HTTPS network traffic captured from four commercial LLM platforms (Claude, ChatGPT, Gemini, and Grok) using an intercepting proxy. SettingTwo researchers independently executed standardized interaction protocols using newly created free-tier accounts from US-based locations. Data were collected from October 7 to November 14, 2025.Results: A total of 1,839 HTTP/HTTPS requests were captured. Third-party request rates varied substantially across platforms: Grok, 35.9% (123 of 343 requests); Claude, 14.0% (87 of 623); ChatGPT, 13.7% (54 of 395); and Gemini, 9.0% (43 of 478). The login action generated the highest third-party request proportions across all platforms; Claude transmitted 64.0% of login requests to external domains. Of 307 third-party requests, 301 (98.0%) were classified. Analytics and tracking services were contacted by all four platforms. Functional breakdown revealed that 62% of Claude’s third-party traffic comprised content delivery network (CDN) requests; Grok’s third-party traffic included 38 analytics requests (Mixpanel and Google Analytics). Gemini’s comparatively low aggregate third-party rate partly reflects Google’s vertical infrastructure ownership rather than categorically different data-sharing practices.Conclusions: All four commercial LLM platforms maintained connections to third-party analytics and tracking infrastructure during routine use. These findings identify a regulatory void in which platforms increasingly used for health-related queries operate under consumer technology data governance frameworks without the privacy protections applicable to healthcare settings. Independent monitoring, transparent disclosure to users, and updated regulatory frameworks are warranted.
- 20 May 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessNot a Private Consultation: Third-Party Network Traffic in Commercial LLMs
Objective: This study presents an analysis of network traffic across four commercial LLM platforms to document which external domains they contact during routine use and to classify those domains by functional purpose.Design: Cross-sectional analysis of HTTP/HTTPS network traffic captured from four commercial LLM platforms (Claude, ChatGPT, Gemini, and Grok) using an intercepting proxy. SettingTwo researchers independently executed standardized interaction protocols using newly created free-tier accounts from US-based locations. Data were collected from October 7 to November 14, 2025.Results: A total of 1,839 HTTP/HTTPS requests were captured. Third-party request rates varied substantially across platforms: Grok, 35.9% (123 of 343 requests); Claude, 14.0% (87 of 623); ChatGPT, 13.7% (54 of 395); and Gemini, 9.0% (43 of 478). The login action generated the highest third-party request proportions across all platforms; Claude transmitted 64.0% of login requests to external domains. Of 307 third-party requests, 301 (98.0%) were classified. Analytics and tracking services were contacted by all four platforms. Functional breakdown revealed that 62% of Claude’s third-party traffic comprised content delivery network (CDN) requests; Grok’s third-party traffic included 38 analytics requests (Mixpanel and Google Analytics). Gemini’s comparatively low aggregate third-party rate partly reflects Google’s vertical infrastructure ownership rather than categorically different data-sharing practices.Conclusions: All four commercial LLM platforms maintained connections to third-party analytics and tracking infrastructure during routine use. These findings identify a regulatory void in which platforms increasingly used for health-related queries operate under consumer technology data governance frameworks without the privacy protections applicable to healthcare settings. Independent monitoring, transparent disclosure to users, and updated regulatory frameworks are warranted.
- 18 May 2026
- Published by Taylor & Francis in The American Journal of Bioethics
- Vol. 26 (6) , 41-43
Planetary Health Ethics Needs to Consider the Emerging Role of Artificial Intelligence
Sabine Salloch’s call for a planetary health research ethics is a timely and illuminating conceptual proposal (Salloch 2026). She successfully showcases why environmental concerns in research ethic...
- 18 May 2026
- Published by Taylor & Francis in The American Journal of Bioethics
- Vol. 26 (6) , 41-43
Planetary Health Ethics Needs to Consider the Emerging Role of Artificial Intelligence
Sabine Salloch’s call for a planetary health research ethics is a timely and illuminating conceptual proposal (Salloch 2026). She successfully showcases why environmental concerns in research ethic...
- Open Access
- 20 January 2026
- Published by Edward Elgar Publishing
- p. 19-36
Health under climate pressure: An emerging research agenda
- 20 January 2026
- Published by Edward Elgar Publishing
- p. 19-36
Open AccessHealth under climate pressure: An emerging research agenda
- Top Cited PapersOpen Access
- 2 May 2025
- Published by Springer Nature in BMC Global and Public Health
- Vol. 3 (1) , 1-11
Building health systems capable of leveraging AI: applying Paul Farmer’s 5S framework for equitable global health
The development of artificial intelligence (AI) applications in healthcare is often positioned as a solution to the greatest challenges facing global health. Advocates propose that AI can bridge gaps in care delivery and access, improving healthcare quality and reducing inequity, including in resource-constrained settings. A broad base of critical scholarship has highlighted important issues with healthcare AI, including algorithmic bias and inequitable and inaccurate model outputs. While such criticisms are valid, there exists a much more fundamental challenge that is often overlooked in global health policy debates: the dangerous mismatch between AI’s imagined benefits and the material realities of healthcare systems globally. AI cannot be deployed effectively or ethically in contexts lacking sufficient social and material infrastructure and resources to provide effective healthcare services. Continued investments in AI within unprepared, under-resourced contexts risk misallocating resources and potentially causing more harm than good. The article concludes by providing concrete questions to assess AI systemic capacity and socio-technical readiness in global health.
- 2 May 2025
- Published by Springer Nature in BMC Global and Public Health
- Vol. 3 (1) , 1-11
Top Cited PapersOpen AccessBuilding health systems capable of leveraging AI: applying Paul Farmer’s 5S framework for equitable global health
The development of artificial intelligence (AI) applications in healthcare is often positioned as a solution to the greatest challenges facing global health. Advocates propose that AI can bridge gaps in care delivery and access, improving healthcare quality and reducing inequity, including in resource-constrained settings. A broad base of critical scholarship has highlighted important issues with healthcare AI, including algorithmic bias and inequitable and inaccurate model outputs. While such criticisms are valid, there exists a much more fundamental challenge that is often overlooked in global health policy debates: the dangerous mismatch between AI’s imagined benefits and the material realities of healthcare systems globally. AI cannot be deployed effectively or ethically in contexts lacking sufficient social and material infrastructure and resources to provide effective healthcare services. Continued investments in AI within unprepared, under-resourced contexts risk misallocating resources and potentially causing more harm than good. The article concludes by providing concrete questions to assess AI systemic capacity and socio-technical readiness in global health.
- Open Access
- 3 April 2025
- Published by Springer Nature in Zeitschrift für Ethik und Moralphilosophie
- p. 1-6
What feminist turn?
In this commentary, we would like to ask: what is meant by the feminist turn in German-speaking bioethics? We take this opportunity to reflect on how non-dominant histories in feminist thought and bioethics might offer important starting points for considering the feminist turn in German-speaking bioethics. Any feminist turn in bioethics needs to strengthen an analysis of power, intersectionality, relationality and contextuality, social responsibility, with a specific stake in the socio-political positionalities and commitments of the German-speaking context.
- 3 April 2025
- Published by Springer Nature in Zeitschrift für Ethik und Moralphilosophie
- p. 1-6
Open AccessWhat feminist turn?
In this commentary, we would like to ask: what is meant by the feminist turn in German-speaking bioethics? We take this opportunity to reflect on how non-dominant histories in feminist thought and bioethics might offer important starting points for considering the feminist turn in German-speaking bioethics. Any feminist turn in bioethics needs to strengthen an analysis of power, intersectionality, relationality and contextuality, social responsibility, with a specific stake in the socio-political positionalities and commitments of the German-speaking context.
- Open Access
- 27 March 2025
- Published by Elsevier in The Lancet Digital Health
- Vol. 7 (4) , e286-e294
Weighing the benefits and risks of collecting race and ethnicity data in clinical settings for medical artificial intelligence
- 27 March 2025
- Published by Elsevier in The Lancet Digital Health
- Vol. 7 (4) , e286-e294
Open AccessWeighing the benefits and risks of collecting race and ethnicity data in clinical settings for medical artificial intelligence
- Open Access
Care Phenotypes In Critical Care
The Social Determinants of Health (SDoH) have long been recognised as significant drivers of health inequalities. Within healthcare settings, large EHR datasets have increasingly enabled the use of machine learning (ML) to explore how patient background and demographic factors mediate and predict clinical outcomes. The intensive care unit (ICU) in particular provides a rich source of data for such research. However, major limitations with current approaches persist, including (i) overreliance on individual demographic labels or measures of difference, (ii) the impracticality of highly intersectional patient groups and (iii) that the underlying accuracy and validity of these demographic constructs is low. The main objective of this study was to take a novel approach, to first understand who within the ICU setting receives sub-standard care and use this to create new, objective labels based on quality of care and outcomes (‘Care Phenotypes’) when different patients interface with the health system. Using the MIMIC-IV database, we focused on highly protocolised, essential care procedures (turning, mouth care) in mechanically ventilated ICU patients. We performed a series of regression analyses to understand in which patients treatment deviated from these protocols. In a cohort of 8,919 ICU patients undergoing IMV, consistent patterns in sup-optimal protocol adherence for certain groups, notably heavier patients. Compared to equivalently sick peers, for every extra weight decile, a patient can expect a reduction of one percentile in frequency of turning care (0.0760 turning interval percentile per weight percentile, p<0.05). Furthermore, patients who receive fewer turnings should also expect to receive fewer mouth care procedures, in a quantile ratio of 1 to 5 (0.2055 mouth care interval quantile per turning interval quantile, p<0.05). The findings in this initial analysis provide support to the concept of first looking at the actual care delivered to patients to illuminate the relationship between patient demographics and outcomes of interest. The ‘Care Phenotypes’ approach has the potential to improve fairness evaluations for machine learning in healthcare, support causal inference research and play a larger role in research into healthcare disparities.
Open AccessCare Phenotypes In Critical Care
The Social Determinants of Health (SDoH) have long been recognised as significant drivers of health inequalities. Within healthcare settings, large EHR datasets have increasingly enabled the use of machine learning (ML) to explore how patient background and demographic factors mediate and predict clinical outcomes. The intensive care unit (ICU) in particular provides a rich source of data for such research. However, major limitations with current approaches persist, including (i) overreliance on individual demographic labels or measures of difference, (ii) the impracticality of highly intersectional patient groups and (iii) that the underlying accuracy and validity of these demographic constructs is low. The main objective of this study was to take a novel approach, to first understand who within the ICU setting receives sub-standard care and use this to create new, objective labels based on quality of care and outcomes (‘Care Phenotypes’) when different patients interface with the health system. Using the MIMIC-IV database, we focused on highly protocolised, essential care procedures (turning, mouth care) in mechanically ventilated ICU patients. We performed a series of regression analyses to understand in which patients treatment deviated from these protocols. In a cohort of 8,919 ICU patients undergoing IMV, consistent patterns in sup-optimal protocol adherence for certain groups, notably heavier patients. Compared to equivalently sick peers, for every extra weight decile, a patient can expect a reduction of one percentile in frequency of turning care (0.0760 turning interval percentile per weight percentile, p<0.05). Furthermore, patients who receive fewer turnings should also expect to receive fewer mouth care procedures, in a quantile ratio of 1 to 5 (0.2055 mouth care interval quantile per turning interval quantile, p<0.05). The findings in this initial analysis provide support to the concept of first looking at the actual care delivered to patients to illuminate the relationship between patient demographics and outcomes of interest. The ‘Care Phenotypes’ approach has the potential to improve fairness evaluations for machine learning in healthcare, support causal inference research and play a larger role in research into healthcare disparities.
- Open Access
Diversity in the medical research ecosystem: a descriptive scientometric analysis of over 49 000 studies and 150 000 authors published in high-impact medical journals between 2007 and 2022
Objectives: Health research that significantly impacts global clinical practice and policy is often published in high-impact factor (IF) medical journals. These outlets play a pivotal role in the worldwide dissemination of novel medical knowledge. However, researchers identifying as women and those affiliated with institutions in low- and middle-income countries (LMICs) have been largely under-represented in high-IF journals across multiple fields of medicine. To evaluate disparities in gender and geographical representation among authors who have published in any of five top general medical journals, we conducted scientometric analyses using a large-scale dataset extracted from the New England Journal of Medicine , Journal of the American Medical Association , The BMJ , The Lancet and Nature Medicine . Methods: Author metadata from all articles published in the selected journals between 2007 and 2022 were collected using the DimensionsAI platform. The Genderize.io Application Programming Interface was then used to infer each author’s likely gender based on their extracted first name. The World Bank country classification was used to map countries associated with researcher affiliations to the LMIC or the high-income country (HIC) category. We characterised the overall gender and country income category representation across the five medical journals. In addition, we computed article-level diversity metrics and contrasted their distributions across the journals. Results: We studied 151 536 authors across 49 764 articles published in five top medical journals, over a period spanning 15 years. On average, approximately one-third (33.1%) of the authors of a given paper were inferred to be women; this result was consistent across the journals we studied. Further, 86.6% of the teams were exclusively composed of HIC authors; in contrast, only 3.9% were exclusively composed of LMIC authors. The probability of serving as the first or last author was significantly higher if the author was inferred to be a man (18.1% vs 16.8%, p<0.01) or was affiliated with an institution in a HIC (16.9% vs 15.5%, p<0.01). Our primary finding reveals that having a diverse team promotes further diversity, within the same dimension (ie, gender or geography) and across dimensions. Notably, papers with at least one woman among the authors were more likely to also involve at least two LMIC authors (11.7% vs 10.4% in baseline, p<0.001; based on inferred gender); conversely, papers with at least one LMIC author were more likely to also involve at least two women (49.4% vs 37.6%, p<0.001; based on inferred gender). Conclusion: We provide a scientometric framework to assess authorship diversity. Our research suggests that the inclusiveness of high-impact medical journals is limited in terms of both gender and geography. We advocate for medical journals to adopt policies and practices that promote greater diversity and collaborative research. In addition, our findings offer a first step towards understanding the composition of teams conducting medical research globally and an opportunity for individual authors to reflect on their own collaborative research practices and possibilities to cultivate more diverse partnerships in their work.
Open AccessDiversity in the medical research ecosystem: a descriptive scientometric analysis of over 49 000 studies and 150 000 authors published in high-impact medical journals between 2007 and 2022
Objectives: Health research that significantly impacts global clinical practice and policy is often published in high-impact factor (IF) medical journals. These outlets play a pivotal role in the worldwide dissemination of novel medical knowledge. However, researchers identifying as women and those affiliated with institutions in low- and middle-income countries (LMICs) have been largely under-represented in high-IF journals across multiple fields of medicine. To evaluate disparities in gender and geographical representation among authors who have published in any of five top general medical journals, we conducted scientometric analyses using a large-scale dataset extracted from the New England Journal of Medicine , Journal of the American Medical Association , The BMJ , The Lancet and Nature Medicine . Methods: Author metadata from all articles published in the selected journals between 2007 and 2022 were collected using the DimensionsAI platform. The Genderize.io Application Programming Interface was then used to infer each author’s likely gender based on their extracted first name. The World Bank country classification was used to map countries associated with researcher affiliations to the LMIC or the high-income country (HIC) category. We characterised the overall gender and country income category representation across the five medical journals. In addition, we computed article-level diversity metrics and contrasted their distributions across the journals. Results: We studied 151 536 authors across 49 764 articles published in five top medical journals, over a period spanning 15 years. On average, approximately one-third (33.1%) of the authors of a given paper were inferred to be women; this result was consistent across the journals we studied. Further, 86.6% of the teams were exclusively composed of HIC authors; in contrast, only 3.9% were exclusively composed of LMIC authors. The probability of serving as the first or last author was significantly higher if the author was inferred to be a man (18.1% vs 16.8%, p<0.01) or was affiliated with an institution in a HIC (16.9% vs 15.5%, p<0.01). Our primary finding reveals that having a diverse team promotes further diversity, within the same dimension (ie, gender or geography) and across dimensions. Notably, papers with at least one woman among the authors were more likely to also involve at least two LMIC authors (11.7% vs 10.4% in baseline, p<0.001; based on inferred gender); conversely, papers with at least one LMIC author were more likely to also involve at least two women (49.4% vs 37.6%, p<0.001; based on inferred gender). Conclusion: We provide a scientometric framework to assess authorship diversity. Our research suggests that the inclusiveness of high-impact medical journals is limited in terms of both gender and geography. We advocate for medical journals to adopt policies and practices that promote greater diversity and collaborative research. In addition, our findings offer a first step towards understanding the composition of teams conducting medical research globally and an opportunity for individual authors to reflect on their own collaborative research practices and possibilities to cultivate more diverse partnerships in their work.