Biography

Miriam is dedicated to advancing equitable healthcare through AI. As a Computer Science graduate student at Emory University, she applies deep learning to critical clinical challenges with a focus on reducing healthcare disparities. Her primary research integrates thermography with deep learning to enable early detection of pressure injuries, particularly for underrepresented skin tones where disparities are most pronounced yet understudied. Preliminary findings from this work have been published in PLOS ONE. Additionally, her research on longitudinal electronic health records identified how improper data partitioning can introduce leakage that artificially inflates model performance, findings accepted at AMIA with broader implications for any EHR-based prediction task. Beyond research, Miriam actively contributes to the ML4H community as both a presenter and reviewer, and also reviews for Springer Cureus. She participates in initiatives like Emory’s Health AI Bias Datathons and regularly attends symposiums and webinars to stay current in the field. She also mentors Emory undergraduates pursuing STEM careers. Through collaborative, responsible innovation, Miriam aspires to build AI tools that improve patient outcomes and reduce healthcare disparities.

Academic Status

PhD Student - 4th

Research Area/Department

Computer Science; Machine Learning/AI

Major/Specialty

Computer Science and Informatics (Biomedical Informatics)

Degrees Earned or in Progress

- PhD Computer Science and Informatics, In progress - MEng Computer Science, 2019 - BSc Computer Science, 2016

Academic Preparation

I have taken the following courses as part of my graduate program: - Biostatistics for machine learning - Machine Learning - Introduction to Ethical Data Science and Informatics From LinkedIn learning: - PyTorch Essential Training: Deep Learning - Hands-On PyTorch Machine Learning - Deep Learning with Python: Convolutional Neural Networks - Data Visualization with Matplotlib and Seaborn

Research/Publications

First-Author Publications: M. Asare-Baiden, SE Sonenblum, K Jordan, A Chung, et al. "A Feasibility Study of Thermography for Detecting Pressure Injuries Across Diverse Skin Tones." medRxiv preprint, 2024. Available at: [https://pmc.ncbi.nlm.nih.gov/articles/PMC11527050/] M. Asare-Baiden, K Jordan, A Chung, SE Sonenblum, et al. "Is thermography a viable solution for detecting pressure injuries in dark skin patients?" arXiv preprint, 2024. Available at: [https://arxiv.org/pdf/2411.10627] M. Asare-Baiden, Vivian Zhang, Vicki Stover Heetrtzberg, Joyce C. Ho "Beyond Random Splitting: Evaluating the Impact of Data Partitioning Strategies on Ventilator-Associated Pneumonia Prediction Using Electronic Health Records" AMIA Annual Symposium Proceedings, 2025. (Accepted) M. Asare-Baiden,Sharon Eve Sonenblum, Kathleen Jordan, Glory Tomi John et al. "Impact of Imaging Protocols on Convolutional Neural Network-Based Pressure Injury Detection" Scientific Reports (Nature), 2025. (Under review) Co-Authored Publications: Sharon Eve Sonenblum, K Jordan, GT John, A Chung, M. Asare-Baiden, et al. "Impact of skin tone, environmental, and technical factors on thermal imaging." PLOS ONE, 2025. Kathleen Jordan, Glory Tomi John, Andrew Chung, Miriam Asare-Baiden et al. "Impact of Skin Tone and Cupping on Erythema and Thermal Imaging Measurements" Research Venues: All my research has been conducted at Emory University's Department of Computer Science and Informatics, in collaboration with the Emory Nell Hudgson School of Nursing.

Research/Academic Interests

My research interests include developing equitable and trustworthy AI systems for healthcare applications, including diagnostic tools and clinical decision support systems. I am particularly drawn to algorithmic fairness, ensuring AI models perform reliably across diverse patient populations through fairness-aware development, subgroup analysis, and validation strategies. Additionally, I and interested in AI governance, including frameworks for auditing AI systems, establishing accountability mechanisms, and developing policies that promote equitable access and mitigate bias. Finally, I am drawn to implementation science, with a focus on understanding how AI tools are adopted, integrated into clinical workflows, and sustained in real-world healthcare settings, particularly in resource-constrained environments.

Computational and Data Science Areas

Applied Computer Science; Artificial Intelligence and Intelligent Systems; Computer Science; Health Sciences

Motivation

Growing up in Africa, I witnessed firsthand how malnutrition and limited access to healthcare created cascading health problems. Many children suffered from kwashiorkor and rickets, while adults developed diabetes that often went undiagnosed until it led to strokes. I vividly remember my grandfather’s stroke, which ultimately claimed his life, and as a child, I became determined to find ways to prevent such tragedies. Although I could not see myself in a traditional clinical role because the sight of blood made me queasy, I was drawn to computers and their possibilities. After high school, I enrolled in a three-month software course that ignited an unexpected passion for technology. Though my journey through college was challenging, I persevered and went on to complete both my bachelor’s and master’s degrees. At one point, I questioned whether I could fulfill my passion for addressing the health and nutrition concerns I had witnessed in my childhood through computer science. Still, I held on to the hope that I could one day reconnect my technical training with my early interest in health. Fortunately, the rise of AI in healthcare offered a way to unite these two paths, enabling me to tackle health challenges by developing intelligent systems that improve outcomes and promote equity, especially for populations that existing technologies often fail to serve adequately. Pressure injury detection particularly caught my attention because malnutrition is a key risk factor, and early detection remains especially difficult in patients with darker skin tones. Motivated by these challenges, I began investigating the feasibility of using thermography, combined with deep learning, to identify early tissue damage that precedes pressure ulcers. I am deeply inspired by the transformative potential of AI-driven approaches to advance health equity, and I see this fellowship as an ideal opportunity to refine my research through mentorship and collaboration, ultimately contributing to AI tools that ensure equitable healthcare for all populations.

Lightning Talk Title

EQUITY AND TRUSTWORTHINESS: RESPONSIBLE AI IN HEALTHCARE

Keywords (Maximum 20 words)

Equitable AI; Trustworthy AI; Algorithmic Fairness ; AI Governance and Policy ; Auditing AI Systems; Healthcare Applications