Stephen Pfohl
Stephen Pfohl is a research scientist in Responsible AI at Google Research on the Context in AI Research (CAIR) team. He conducts research on the design on the machine learning models that are fair, robust, and transparent, primarily for use in healthcare contexts. Prior to joining Google, he completed his PhD in Biomedical Informatics at Stanford University.
Research Areas
Authored Publications
Sort By
Artificial Intelligence, in context: The Collaboration for trANslational Artificial Intelligence tRIals (CANAIRI) and the value of local evaluation of AI
Melissa D McCradden
Alex John London
Judy Gichoya
Mark Sendak
Lauren Erdman
Ian Stedman
Lauren Oakden-Rayner
Ismail Akrout
James A Anderson
Lesley-Anne Farmer
Robert Greer
Anna Goldenberg
Yvonne Ho
Shalmali Joshi
Jennie Louise
Muhammad Mamdani
Mjaye L. Mazwi
Abdullahi Mohamud
Lyle Palmer
Antonios Peperidis
Mandy Rickard
Carolyn Semmler
Karandeep Singh
Devin Singh
Seyi Soremekun
Lana Tikhomirov
Anton H van der Vegt
Xiaoxuan Liu
Nature Medicine (2025)
Preview abstract
The literature and governance frameworks relating to the clinical translation of artificial intelligence (AI) products are coalescing around best practices. A major knowledge gap concerns demonstration of local, on-the-ground evidence of on-the-ground performance, equity, and safety. The ‘silent trial’ (also known as ‘silent evaluation’, ‘shadow trial’, ‘shadow evaluation’) refers to the integration of an AI system into the intended operational systems without affecting patient care. A silent trial enables collection of on-the-ground evidence of performance while integrating data security considerations, assessing operational feasibility, validating deployment strategy, selecting risk category thresholds, and testing workflow integration. While increasingly recognized for their value, silent trials remain under-appreciated and under-utilized both in the breadth of their uptake and in the scope of considerations they can address - for example, local fairness assessments and patient engagement, among others. This commentary introduces the Collaboration for trANslational Artificial Intelligence tRIals (CANAIRI), an international consortium initiative that aims to develop guidance around the application of holistic silent evaluations for healthcare AI tools.
View details
Understanding challenges to the validity of disaggregated evaluations for algorithmic fairness
Chirag Nagpal
David Madras
Vishwali Mhasawade
Olawale Salaudeen
Awa Dieng
Shannon Sequeira
Santiago Arciniegas
Lillian Sung
Nnamdi Ezeanochie
Heather Cole-Lewis
Sanmi Koyejo
Proceedings of the 2025 Conference on Neural Information Processing Systems (NeurIPS) (2025)
Preview abstract
Disaggregated evaluation across subgroups is critical for assessing the fairness of machine learning models, but its uncritical use can mislead practitioners. We show that equal performance across subgroups is an unreliable measure of fairness when data are representative of the relevant populations but reflective of real-world disparities. Furthermore, when data are not representative due to selection bias, both disaggregated evaluation and alternative approaches based on conditional independence testing may be invalid without explicit assumptions regarding the bias mechanism. We use causal graphical models to characterize fairness properties and metric stability across subgroups under different data generating processes. Our framework suggests complementing disaggregated evaluations with explicit causal assumptions and analysis to control for confounding and distribution shift, including conditional independence testing and weighted performance estimation. These findings have broad implications for how practitioners design and interpret model assessments given the ubiquity of disaggregated evaluation.
View details
The Case for Globalizing Fairness: A Mixed Methods Study on the Perceptions of Colonialism, AI and Health in Africa
Awa Dieng
Iskandar Haykel
Chirag Nagpal
Aisha Walcott-Bryant
Sanmi Koyejo
2024
Preview abstract
With growing machine learning (ML) and large language model applications in healthcare, there have been calls for fairness in ML to understand and mitigate ethical concerns these systems may pose. Fairness has implications for health in Africa, which already has inequitable power imbalances between the Global North and South. This paper seeks to explore fairness for global health, with Africa as a case study.
We conduct a scoping review to propose fairness attributes for consideration in the African context and delineate where they may come into play in different ML-enabled medical modalities. We then conduct qualitative research studies with 625 general population study participants in 5 countries in Africa and 28 experts in ML, Health, and/or policy focussed on Africa to obtain feedback on the proposed attributes. We delve specifically into understanding the interplay between AI, health and colonialism.
Our findings demonstrate that among experts there is a general mistrust that technologies that are solely developed by former colonizers can benefit Africans, and that associated resource constraints due to pre-existing economic and infrastructure inequities can be linked to colonialism. General population survey responses found about an average of 40% of people associate an undercurrent of colonialism to AI and this was most dominant amongst participants from South Africa. However the majority of the general population participants surveyed did not think there was a direct link between AI and colonialism.Colonial history, country of origin, National income level were specific axes of disparities that participants felt would cause an AI tool to be biased
This work serves as a basis for policy development around Artificial Intelligence for health in Africa and can be expanded to other regions.
View details
Nteasee: A qualitative study of expert and general population perspectives on deploying AI for health in African countries
Iskandar Haykel
Awa Dieng
Kerrie Kauer
Florence Ofori
Tousif Ahmad
2024
Preview abstract
Background: Artificial Intelligence for health has the potential to significantly change and improve healthcare. However in most African countries identifying culturally and contextually attuned approaches for deploying these solutions is not well understood. To bridge this gap, we conduct a qualitative study to investigate the best practices, fairness indicators and potential biases to mitigate when deploying AI for health in African countries, as well as explore opportunities where artificial intelligence could make a positive impact in health.
Methods: We used a mixed methods approach combining in-depth interviews (IDIs) and surveys. We conduct 1.5-2 hour long IDIs with 50 experts in health, policy and AI across 17 countries, and through an inductive approach we conduct a qualitative thematic analysis on expert IDI responses. We administer a blinded 30-minute survey with thought-cases to 672 general population participants across 5 countries in Africa (Ghana, South Africa, Rwanda, Kenya and Nigeria), and analyze responses on quantitative scales, statistically comparing responses by country, age, gender, and level of familiarity with AI. We thematically summarize open-ended responses from surveys.
Results and Conclusion: Our results find generally positive attitudes, high levels of trust, accompanied by moderate levels of concern among general population participants for AI usage for health in Africa. This contrasts with expert responses, where major themes revolved around trust/mistrust, AI ethics concerns, and systemic barriers to overcome, among others. This work presents the first-of-its-kind qualitative research study of the potential of AI for health in Africa with perspectives from both experts and the general population. We hope that this work guides policy makers and drives home the need for education and the inclusion of general population perspectives in decision-making around AI usage.
View details
An intentional approach to managing bias in embedding models
Atilla P. Kiraly
Jungyeon Park
Rory Pilgrim
Charles Lau
Heather Cole-Lewis
Shravya Shetty
Krish Eswaran
Leo Anthony Celi
The Lancet Digital Health, 6 (2024), E126-E130
Preview abstract
Advances in machine learning for health care have brought concerns about bias from the research community; specifically, the introduction, perpetuation, or exacerbation of care disparities. Reinforcing these concerns is the finding that medical images often reveal signals about sensitive attributes in ways that are hard to pinpoint by both algorithms and people. This finding raises a question about how to best design general purpose pretrained embeddings (GPPEs, defined as embeddings meant to support a broad array of use cases) for building downstream models that are free from particular types of bias. The downstream model should be carefully evaluated for bias, and audited and improved as appropriate. However, in our view, well intentioned attempts to prevent the upstream components—GPPEs—from learning sensitive attributes can have unintended consequences on the downstream models. Despite producing a veneer of technical neutrality, the resultant end-to-end system might still be biased or poorly performing. We present reasons, by building on previously published data, to support the reasoning that GPPEs should ideally contain as much information as the original data contain, and highlight the perils of trying to remove sensitive attributes from a GPPE. We also emphasise that downstream prediction models trained for specific tasks and settings, whether developed using GPPEs or not, should be carefully designed and evaluated to avoid bias that makes models vulnerable to issues such as distributional shift. These evaluations should be done by a diverse team, including social scientists, on a diverse cohort representing the full breadth of the patient population for which the final model is intended.
View details
A Toolbox for Surfacing Health Equity Harms and Biases in Large Language Models
Heather Cole-Lewis
Awa Dieng
Nenad Tomašev
Liam McCoy
Leo Anthony Celi
Alanna Walton
Chirag Nagpal
Akeiylah DeWitt
Philip Mansfield
Sushant Prakash
Joelle Barral
Ivor Horn
Karan Singhal
Nature Medicine (2024)
Preview abstract
Large language models (LLMs) hold promise to serve complex health information needs but also have the potential to introduce harm and exacerbate health disparities. Reliably evaluating equity-related model failures is a critical step toward developing systems that promote health equity. We present resources and methodologies for surfacing biases with potential to precipitate equity-related harms in long-form, LLM-generated answers to medical questions and conduct a large-scale empirical case study with the Med-PaLM 2 LLM. Our contributions include a multifactorial framework for human assessment of LLM-generated answers for biases and EquityMedQA, a collection of seven datasets enriched for adversarial queries. Both our human assessment framework and our dataset design process are grounded in an iterative participatory approach and review of Med-PaLM 2 answers. Through our empirical study, we find that our approach surfaces biases that may be missed by narrower evaluation approaches. Our experience underscores the importance of using diverse assessment methodologies and involving raters of varying backgrounds and expertise. While our approach is not sufficient to holistically assess whether the deployment of an artificial intelligence (AI) system promotes equitable health outcomes, we hope that it can be leveraged and built upon toward a shared goal of LLMs that promote accessible and equitable healthcare.
View details
Health equity assessment of machine learning performance (HEAL): a framework and dermatology AI model case study
Terry Spitz
Malcolm Chelliah
Heather Cole-Lewis
Donald Martin
Tiam Jaroensri
Geoff Keeling
Stephanie Farquhar
Qinghan Xue
Jenna Lester
Cían Hughes
Patricia Strachan
Fraser Tan
Peggy Bui
Craig Mermel
Lily Peng
Sunny Virmani
Ivor Horn
The Lancet eClinicalMedicine (2024)
Preview abstract
Background
Artificial intelligence (AI) has repeatedly been shown to encode historical inequities in healthcare. We aimed to develop a framework to quantitatively assess the performance equity of health AI technologies and to illustrate its utility via a case study.
Methods
Here, we propose a methodology to assess whether health AI technologies prioritise performance for patient populations experiencing worse outcomes, that is complementary to existing fairness metrics. We developed the Health Equity Assessment of machine Learning performance (HEAL) framework designed to quantitatively assess the performance equity of health AI technologies via a four-step interdisciplinary process to understand and quantify domain-specific criteria, and the resulting HEAL metric. As an illustrative case study (analysis conducted between October 2022 and January 2023), we applied the HEAL framework to a dermatology AI model. A set of 5420 teledermatology cases (store-and-forward cases from patients of 20 years or older, submitted from primary care providers in the USA and skin cancer clinics in Australia), enriched for diversity in age, sex and race/ethnicity, was used to retrospectively evaluate the AI model's HEAL metric, defined as the likelihood that the AI model performs better for subpopulations with worse average health outcomes as compared to others. The likelihood that AI performance was anticorrelated to pre-existing health outcomes was estimated using bootstrap methods as the probability that the negated Spearman's rank correlation coefficient (i.e., “R”) was greater than zero. Positive values of R suggest that subpopulations with poorer health outcomes have better AI model performance. Thus, the HEAL metric, defined as p (R >0), measures how likely the AI technology is to prioritise performance for subpopulations with worse average health outcomes as compared to others (presented as a percentage below). Health outcomes were quantified as disability-adjusted life years (DALYs) when grouping by sex and age, and years of life lost (YLLs) when grouping by race/ethnicity. AI performance was measured as top-3 agreement with the reference diagnosis from a panel of 3 dermatologists per case.
Findings
Across all dermatologic conditions, the HEAL metric was 80.5% for prioritizing AI performance of racial/ethnic subpopulations based on YLLs, and 92.1% and 0.0% respectively for prioritizing AI performance of sex and age subpopulations based on DALYs. Certain dermatologic conditions were significantly associated with greater AI model performance compared to a reference category of less common conditions. For skin cancer conditions, the HEAL metric was 73.8% for prioritizing AI performance of age subpopulations based on DALYs.
Interpretation
Analysis using the proposed HEAL framework showed that the dermatology AI model prioritised performance for race/ethnicity, sex (all conditions) and age (cancer conditions) subpopulations with respect to pre-existing health disparities. More work is needed to investigate ways of promoting equitable AI performance across age for non-cancer conditions and to better understand how AI models can contribute towards improving equity in health outcomes.
View details
The Case for Globalizing Fairness: A Mixed Methods Study on the Perceptions of Colonialism, AI and Health in Africa
Awa Dieng
Iskandar Haykel
Chirag Nagpal
Aisha Walcott-Bryant
Sanmi Koyejo
2024
Preview abstract
With growing machine learning (ML) and large language model applications in healthcare, there have been calls for fairness in ML to understand and mitigate ethical concerns these systems may pose. Fairness has implications for health in Africa, which already has inequitable power imbalances between the Global North and South. This paper seeks to explore fairness for global health, with Africa as a case study.
We conduct a scoping review to propose fairness attributes for consideration in the African context and delineate where they may come into play in different ML-enabled medical modalities. We then conduct qualitative research studies with 625 general population study participants in 5 countries in Africa and 28 experts in ML, Health, and/or policy focussed on Africa to obtain feedback on the proposed attributes. We delve specifically into understanding the interplay between AI, health and colonialism.
Our findings demonstrate that among experts there is a general mistrust that technologies that are solely developed by former colonizers can benefit Africans, and that associated resource constraints due to pre-existing economic and infrastructure inequities can be linked to colonialism. General population survey responses found about an average of 40% of people associate an undercurrent of colonialism to AI and this was most dominant amongst participants from South Africa. However the majority of the general population participants surveyed did not think there was a direct link between AI and colonialism.Colonial history, country of origin, National income level were specific axes of disparities that participants felt would cause an AI tool to be biased
This work serves as a basis for policy development around Artificial Intelligence for health in Africa and can be expanded to other regions.
View details
Towards Physician-Level Medical Question Answering with Large Language Models
Karan Singhal
Juro Gottweis
Le Hou
Kevin Clark
Heather Cole-Lewis
Amy Wang
Sami Lachgar
Philip Mansfield
Sushant Prakash
Bradley Green
Ewa Dominowska
Nenad Tomašev
Renee Wong
Sara Mahdavi
Joelle Barral
Arxiv (2023) (to appear)
Preview abstract
Recent artificial intelligence (AI) systems have reached milestones in "grand challenges" ranging from Go to protein-folding. The capability to retrieve medical knowledge, reason over it, and answer medical questions comparably to physicians has long been viewed as one such grand challenge.
Large language models (LLMs) have catalyzed significant progress in medical question answering; Med-PaLM was the first model to exceed a "passing" score in US Medical Licensing Examination (USMLE) style questions with a score of 67.2% on the MedQA dataset. However, this and other prior work suggested significant room for improvement, especially when models' answers were compared to clinicians' answers. Here we present Med-PaLM 2, which bridges these gaps by leveraging a combination of base LLM improvements (PaLM 2), medical domain finetuning, and prompting strategies including a novel ensemble refinement approach.
Med-PaLM 2 scored up to 86.5% on the MedQA dataset, improving upon Med-PaLM by over 19% and setting a new state-of-the-art. We also observed performance approaching or exceeding state-of-the-art across MedMCQA, PubMedQA, and MMLU clinical topics datasets.
We performed detailed human evaluations on long-form questions along multiple axes relevant to clinical applications. In pairwise comparative ranking of 1066 consumer medical questions, physicians preferred Med-PaLM 2 answers to those produced by physicians on eight of nine axes pertaining to clinical utility (p < 0.001). We also observed significant improvements compared to Med-PaLM on every evaluation axis (p < 0.001) on newly introduced datasets of 240 long-form "adversarial" questions to probe LLM limitations.
While further studies are necessary to validate the efficacy of these models in real-world settings, these results highlight rapid progress towards physician-level performance in medical question answering.
View details
Large Language Models Encode Clinical Knowledge
Karan Singhal
Sara Mahdavi
Jason Wei
Hyung Won Chung
Nathan Scales
Ajay Tanwani
Heather Cole-Lewis
Perry Payne
Martin Seneviratne
Paul Gamble
Christopher Kelly
Abubakr Abdelrazig Hassan Babiker
Nathanael Schaerli
Aakanksha Chowdhery
Philip Mansfield
Dina Demner-Fushman
Katherine Chou
Juraj Gottweis
Nenad Tomašev
Alvin Rajkomar
Joelle Barral
Nature (2023)
Preview abstract
Large language models (LLMs) have demonstrated impressive capabilities, but the bar for clinical applications is high. Attempts to assess the clinical knowledge of models typically rely on automated evaluations based on limited benchmarks. Here, to address these limitations, we present MultiMedQA, a benchmark combining six existing medical question answering datasets spanning professional medicine, research and consumer queries and a new dataset of medical questions searched online, HealthSearchQA. We propose a human evaluation framework for model answers along multiple axes including factuality, comprehension, reasoning, possible harm and bias. In addition, we evaluate Pathways Language Model (PaLM, a 540-billion parameter LLM) and its instruction-tuned variant, Flan-PaLM on MultiMedQA. Using a combination of prompting strategies, Flan-PaLM achieves state-of-the-art accuracy on every MultiMedQA multiple-choice dataset (MedQA, MedMCQA, PubMedQA and Measuring Massive Multitask Language Understanding (MMLU) clinical topics), including 67.6% accuracy on MedQA (US Medical Licensing Exam-style questions), surpassing the prior state of the art by more than 17%. However, human evaluation reveals key gaps. To resolve this, we introduce instruction prompt tuning, a parameter-efficient approach for aligning LLMs to new domains using a few exemplars. The resulting model, Med-PaLM, performs encouragingly, but remains inferior to clinicians. We show that comprehension, knowledge recall and reasoning improve with model scale and instruction prompt tuning, suggesting the potential utility of LLMs in medicine. Our human evaluations reveal limitations of today’s models, reinforcing the importance of both evaluation frameworks and method development in creating safe, helpful LLMs for clinical applications.
View details
