Jenna C. Lester
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Research Interests
Jenna C. Lester has a total of 103 co-authors. Here are the top co-authors by number of shared publications:
- Roxana Daneshjou
Department of Biomedical Data Science, Stanford University School of Medicine, Stanford, CA, USA
8 shared publications - Eleni Linos
Department of Dermatology, Stanford School of Medicine, Stanford, CA, United States
3 shared publications Justin L. Jia
Department of Dermatology, Stanford University School of Medicine, Stanford, California
3 shared publicationsSusan C. Taylor
Department of Dermatology, University of Pennsylvania, Philadelphia, Pennsylvania
3 shared publications- Vivian Yang
School of Medicine, University of California, San Francisco, San Francisco, CA, United States
3 shared publications Adaure U. Amuzie
Indiana University School of Medicine, Indianapolis, Indiana
2 shared publications
Latest Publications
- Open Access
- 10 October 2025
- Published by Oxford University Press (OUP) in British Journal of Dermatology
- Vol. 194 (1) , 135-145
Comparison of methods for characterizing skin pigment diversity in research cohorts
Background: Some pulse oximeters perform worse in people with darker skin, and this may be due to inadequate diversity of skin pigment in device development study cohorts. Guidance is needed to accurately and equitably characterize skin pigment to ensure diversity in research cohorts. We tested multiple methods for characterizing skin pigment to assess comparability and impact on cohort diversity. Objectives: The objectives of this study were to assess reliability and comparability of common skin pigment measurement methods, compare findings from different anatomical sites and demonstrate that pigment cannot be assumed from US National Institutes for Health (NIH) race categories. Methods: We used three subjective methods [perceived Fitzpatrick (pFP) scale, Monk Skin Tone (MST) scale and Von Luschan (VL) scale] and two objective methods [Konica Minolta CM-700d spectrophotometer and Delfin Skin Color Catch (DSCC) colorimeter] for individual typology angle (ITA) across multiple measurement sites in adults. We calculated ΔE to estimate operator perceptibility thresholds for subjective methods and to determine reproducibility for objective methods. We used each method to categorize participants as ‘light, medium or dark’ and compared the impact of method selection on cohort diversity. Results: We studied 789 participants, with 33 856 assessments. The MST had the widest luminosity range, and the VL scale had the least discernible adjacent categories. With ‘dark’ defined as ITA < −30°, 14% of participants were categorized ‘dark’ as compared with 26% by pFP or 16% by MST. Approximately half of the ‘dark’ cohort had an ITA < −50°. With an ITA threshold < −50°, only 7% of the cohort was categorized as ‘dark'. When ‘Black or African American’ self-identification was used to define ‘dark’, 23% of the cohort was categorized as such. Each self-assigned NIH race category included a wide range of ITA and subjective scale categories. Both ITA and L* from the KM-700d and DSCC demonstrated strong correlation (ρ > 0.7). Conclusions: Common methods for skin pigment characterization, especially the use of race or subjective scales, have significant limitations. When applied to the same cohort, different methods yield significantly different results and some may overestimate diversity. Previously published ITA thresholds for defining ‘dark’ skin are too light and lead to under-representation of people with darker skin.
- 10 October 2025
- Published by Oxford University Press (OUP) in British Journal of Dermatology
- Vol. 194 (1) , 135-145
Open AccessComparison of methods for characterizing skin pigment diversity in research cohorts
Background: Some pulse oximeters perform worse in people with darker skin, and this may be due to inadequate diversity of skin pigment in device development study cohorts. Guidance is needed to accurately and equitably characterize skin pigment to ensure diversity in research cohorts. We tested multiple methods for characterizing skin pigment to assess comparability and impact on cohort diversity. Objectives: The objectives of this study were to assess reliability and comparability of common skin pigment measurement methods, compare findings from different anatomical sites and demonstrate that pigment cannot be assumed from US National Institutes for Health (NIH) race categories. Methods: We used three subjective methods [perceived Fitzpatrick (pFP) scale, Monk Skin Tone (MST) scale and Von Luschan (VL) scale] and two objective methods [Konica Minolta CM-700d spectrophotometer and Delfin Skin Color Catch (DSCC) colorimeter] for individual typology angle (ITA) across multiple measurement sites in adults. We calculated ΔE to estimate operator perceptibility thresholds for subjective methods and to determine reproducibility for objective methods. We used each method to categorize participants as ‘light, medium or dark’ and compared the impact of method selection on cohort diversity. Results: We studied 789 participants, with 33 856 assessments. The MST had the widest luminosity range, and the VL scale had the least discernible adjacent categories. With ‘dark’ defined as ITA < −30°, 14% of participants were categorized ‘dark’ as compared with 26% by pFP or 16% by MST. Approximately half of the ‘dark’ cohort had an ITA < −50°. With an ITA threshold < −50°, only 7% of the cohort was categorized as ‘dark'. When ‘Black or African American’ self-identification was used to define ‘dark’, 23% of the cohort was categorized as such. Each self-assigned NIH race category included a wide range of ITA and subjective scale categories. Both ITA and L* from the KM-700d and DSCC demonstrated strong correlation (ρ > 0.7). Conclusions: Common methods for skin pigment characterization, especially the use of race or subjective scales, have significant limitations. When applied to the same cohort, different methods yield significantly different results and some may overestimate diversity. Previously published ITA thresholds for defining ‘dark’ skin are too light and lead to under-representation of people with darker skin.
- Open Access
- 13 July 2025
- Published by JMIR Publications Inc. in JMIR Preprints
Medical Students' Perspectives on and Usage of Large Language Models in Medical Education: An Exploratory Survey (Preprint)
BACKGROUND Large Language Models (LLMs) such as ChatGPT are increasingly used by medical students for studying, research, and clinical applications. Despite their growing presence within medical education, concerns persist regarding accuracy, bias, and overreliance. OBJECTIVE To assess how medical students worldwide are using LLMs and to understand their perceptions of and concerns around these tools. METHODS A cross-sectional survey was administered using Qualtrics™ with broad advertising using social media platforms like LinkedIn, Twitter, and Facebook. Responses were summarized using descriptive statistics. RESULTS Among 120 respondents from 13 countries, 88% reported using LLMs, primarily for studying. Common applications included answering questions (67%), summarizing text (51%), and generating practice questions (37%). ChatGPT was the most used LLM (90%). While 84% of medical students reported a positive educational impact, concerns included accuracy (81%), bias (62%), and ethical issues. CONCLUSIONS LLMs are widely used and generally positively perceived by medical students. However, ethical concerns and the need for training in responsible use highlight the importance of educator guidance and further research into educational and clinical outcomes.
- 13 July 2025
- Published by JMIR Publications Inc. in JMIR Preprints
Open AccessMedical Students' Perspectives on and Usage of Large Language Models in Medical Education: An Exploratory Survey (Preprint)
BACKGROUND Large Language Models (LLMs) such as ChatGPT are increasingly used by medical students for studying, research, and clinical applications. Despite their growing presence within medical education, concerns persist regarding accuracy, bias, and overreliance. OBJECTIVE To assess how medical students worldwide are using LLMs and to understand their perceptions of and concerns around these tools. METHODS A cross-sectional survey was administered using Qualtrics™ with broad advertising using social media platforms like LinkedIn, Twitter, and Facebook. Responses were summarized using descriptive statistics. RESULTS Among 120 respondents from 13 countries, 88% reported using LLMs, primarily for studying. Common applications included answering questions (67%), summarizing text (51%), and generating practice questions (37%). ChatGPT was the most used LLM (90%). While 84% of medical students reported a positive educational impact, concerns included accuracy (81%), bias (62%), and ethical issues. CONCLUSIONS LLMs are widely used and generally positively perceived by medical students. However, ethical concerns and the need for training in responsible use highlight the importance of educator guidance and further research into educational and clinical outcomes.
- Open Access
- 11 April 2025
- Published by JMIR Publications Inc. in JMIR Medical Education
- Vol. 11 (1) , e71721
Authors’ Reply: Enhancing AI-Driven Medical Translations: Considerations for Language Concordance
- 11 April 2025
- Published by JMIR Publications Inc. in JMIR Medical Education
- Vol. 11 (1) , e71721
Open AccessAuthors’ Reply: Enhancing AI-Driven Medical Translations: Considerations for Language Concordance
- 29 March 2025
- Published by Elsevier in Journal of Investigative Dermatology
- Vol. 145 (10) , 2598-2600
Uncovering Disparities in Skin Tone Representation among Artificial Intelligence Vision-Language Models
- 29 March 2025
- Published by Elsevier in Journal of Investigative Dermatology
- Vol. 145 (10) , 2598-2600
Uncovering Disparities in Skin Tone Representation among Artificial Intelligence Vision-Language Models
- Open Access
- 21 February 2025
- Published by JMIR Publications Inc. in Journal of Medical Internet Research
- Vol. 27, e50708
Perspectives of Black, Latinx, Indigenous, and Asian Communities on Health Data Use and AI: Cross-Sectional Survey Study
Despite excitement around artificial intelligence (AI)–based tools in health care, there is work to be done before they can be equitably deployed. The absence of diverse patient voices in discussions on AI is a pressing matter, and current studies have been limited in diversity. Our study inquired about the perspectives of racial and ethnic minority patients on the use of their health data in AI, by conducting a cross-sectional survey among 230 participants who were at least 18 years of age and identified as Black, Latinx, Indigenous, or Asian. While familiarity with AI was high, a smaller proportion of participants understood how AI can be used in health care (152/199, 76.4%), and an even smaller proportion understood how AI can be applied to dermatology (133/199, 66.8%). Overall, 69.8% (139/199) of participants agreed that they trusted the health care system to treat their medical information with respect; however, this varied significantly by income (P=.045). Only 64.3% (128/199) of participants felt comfortable with their medical data being used to build AI tools, and 83.4% (166/199) believed they should be compensated if their data are used to develop AI. To our knowledge, this is the first study focused on understanding opinions about health data use for AI among racial and ethnic minority individuals, as similar studies have had limited diversity. It is important to capture the opinions of diverse groups because the inclusion of their data is essential for building equitable AI tools; however, historical harms have made inclusion challenging.
- 21 February 2025
- Published by JMIR Publications Inc. in Journal of Medical Internet Research
- Vol. 27, e50708
Open AccessPerspectives of Black, Latinx, Indigenous, and Asian Communities on Health Data Use and AI: Cross-Sectional Survey Study
Despite excitement around artificial intelligence (AI)–based tools in health care, there is work to be done before they can be equitably deployed. The absence of diverse patient voices in discussions on AI is a pressing matter, and current studies have been limited in diversity. Our study inquired about the perspectives of racial and ethnic minority patients on the use of their health data in AI, by conducting a cross-sectional survey among 230 participants who were at least 18 years of age and identified as Black, Latinx, Indigenous, or Asian. While familiarity with AI was high, a smaller proportion of participants understood how AI can be used in health care (152/199, 76.4%), and an even smaller proportion understood how AI can be applied to dermatology (133/199, 66.8%). Overall, 69.8% (139/199) of participants agreed that they trusted the health care system to treat their medical information with respect; however, this varied significantly by income (P=.045). Only 64.3% (128/199) of participants felt comfortable with their medical data being used to build AI tools, and 83.4% (166/199) believed they should be compensated if their data are used to develop AI. To our knowledge, this is the first study focused on understanding opinions about health data use for AI among racial and ethnic minority individuals, as similar studies have had limited diversity. It is important to capture the opinions of diverse groups because the inclusion of their data is essential for building equitable AI tools; however, historical harms have made inclusion challenging.
- Open Access
- 10 December 2024
- Published by JMIR Publications Inc. in JMIR Medical Education
- Vol. 10, e51435
ChatGPT May Improve Access to Language-Concordant Care for Patients With Non–English Language Preferences
This study evaluated the accuracy of ChatGPT in translating English patient education materials into Spanish, Mandarin, and Russian. While ChatGPT shows promise for translating Spanish and Russian medical information, Mandarin translations require further refinement, highlighting the need for careful review of AI-generated translations before clinical use.
- 10 December 2024
- Published by JMIR Publications Inc. in JMIR Medical Education
- Vol. 10, e51435
Open AccessChatGPT May Improve Access to Language-Concordant Care for Patients With Non–English Language Preferences
This study evaluated the accuracy of ChatGPT in translating English patient education materials into Spanish, Mandarin, and Russian. While ChatGPT shows promise for translating Spanish and Russian medical information, Mandarin translations require further refinement, highlighting the need for careful review of AI-generated translations before clinical use.
- 22 October 2024
- Published by Springer Nature
- p. 127-141
Dermatology
- Open Access
- 23 July 2024
- Published by Springer Nature in Current Dermatology Reports
- Vol. 13 (3) , 156
Correction: The Impact of Artificial Intelligence on Health Equity in Dermatology
- 23 July 2024
- Published by Springer Nature in Current Dermatology Reports
- Vol. 13 (3) , 156
Open AccessCorrection: The Impact of Artificial Intelligence on Health Equity in Dermatology
- 10 July 2024
- Published by Springer Nature in Current Dermatology Reports
- Vol. 13 (3) , 148-155
The Impact of Artificial Intelligence on Health Equity in Dermatology
- 10 July 2024
- Published by Springer Nature in Current Dermatology Reports
- Vol. 13 (3) , 148-155
The Impact of Artificial Intelligence on Health Equity in Dermatology
- Open Access
- 8 July 2024
- Published by JMIR Publications Inc. in JMIR Preprints
Perspectives of Black, Latinx, Indigenous, and Asian Communities on Health Data Use and AI: Cross-Sectional Survey Study (Preprint)
UNSTRUCTURED Despite excitement around artificial intelligence (AI)–based tools in health care, there is work to be done before they can be equitably deployed. The absence of diverse patient voices in discussions on AI is a pressing matter, and current studies have been limited in diversity. Our study inquired about the perspectives of racial and ethnic minority patients on the use of their health data in AI, by conducting a cross-sectional survey among 230 participants who were at least 18 years of age and identified as Black, Latinx, Indigenous, or Asian. While familiarity with AI was high, a smaller proportion of participants understood how AI can be used in health care (152/199, 76.4%), and an even smaller proportion understood how AI can be applied to dermatology (133/199, 66.8%). Overall, 69.8% (139/199) of participants agreed that they trusted the health care system to treat their medical information with respect; however, this varied significantly by income (P=.045). Only 64.3% (128/199) of participants felt comfortable with their medical data being used to build AI tools, and 83.4% (166/199) believed they should be compensated if their data are used to develop AI. To our knowledge, this is the first study focused on understanding opinions about health data use for AI among racial and ethnic minority individuals, as similar studies have had limited diversity. It is important to capture the opinions of diverse groups because the inclusion of their data is essential for building equitable AI tools; however, historical harms have made inclusion challenging.
- 8 July 2024
- Published by JMIR Publications Inc. in JMIR Preprints
Open AccessPerspectives of Black, Latinx, Indigenous, and Asian Communities on Health Data Use and AI: Cross-Sectional Survey Study (Preprint)
UNSTRUCTURED Despite excitement around artificial intelligence (AI)–based tools in health care, there is work to be done before they can be equitably deployed. The absence of diverse patient voices in discussions on AI is a pressing matter, and current studies have been limited in diversity. Our study inquired about the perspectives of racial and ethnic minority patients on the use of their health data in AI, by conducting a cross-sectional survey among 230 participants who were at least 18 years of age and identified as Black, Latinx, Indigenous, or Asian. While familiarity with AI was high, a smaller proportion of participants understood how AI can be used in health care (152/199, 76.4%), and an even smaller proportion understood how AI can be applied to dermatology (133/199, 66.8%). Overall, 69.8% (139/199) of participants agreed that they trusted the health care system to treat their medical information with respect; however, this varied significantly by income (P=.045). Only 64.3% (128/199) of participants felt comfortable with their medical data being used to build AI tools, and 83.4% (166/199) believed they should be compensated if their data are used to develop AI. To our knowledge, this is the first study focused on understanding opinions about health data use for AI among racial and ethnic minority individuals, as similar studies have had limited diversity. It is important to capture the opinions of diverse groups because the inclusion of their data is essential for building equitable AI tools; however, historical harms have made inclusion challenging.