News and Events

---

 

Chair’s Message

We are moving toward our vision with a number of activities across our various programs. We have updated our strategic plan in response to the 10-year academic program review that we recently completed. For our research-oriented MS and PhD programs, we have recently added a specialization in Data Science. We are completing a curriculum revision for our on line applied clinical informatics MS which will be effective Fall 2020. The work of our fellows in the clinical informatics fellowship program has received plaudits from clinical administrators and faculty, and we are currently recruiting a new faculty member in our department to assist with this program (view position description).  We are also recruiting a faculty member in medical education to start Summer 2020 (view position description). This is the beginning of a new cycle of admissions to our graduate programs, and we look forward to another productive year, and new growth in our department.

Cordially,

Peter Tarczy-Hornoch, MD
Chair and Professor, Department of Biomedical Informatics and Medical Education

The University of Washington Welcomes New Chief Research Information Officer 

The University of Washington is delighted to announce Dr. Shawn Murphy, MD, PhD will be joining the University of Washington as a core faculty in Biomedical Informatics and Medical Education (BIME) with a joint appointment in Neurology. In addition to being a core BIME faculty member and attending in an outpatient neurology clinic, he will be serving as UW Medicine Chief Research Information Officer in IT Services, informatics lead of the Institute for Translational Health Sciences (ITHS) Data Science Core, and the Director of the Institute for Medical Data Sciences (IMDS).

“I am honored and excited to join the Data Science workforce at the University of Washington.  I have always admired the closeness UW students have to the top teachers in the tech industry.  I am hoping to enable Clinical and Informatics Research to use this and take Data Science and AI to a new level at UW,” said Dr. Murphy.

Dr. Murphy is currently the Chief Research Information Officer at Mass General Brigham and a Professor of Neurology. Over the past 30 years, he has been an integral leader and innovator in supporting research data and technology there.  He is the creator of the Research Patient Data Registry (RPDR) and co-founded i2b2 (used in > 300 sites globally) enabling scalable, audit-ready clinical data access for the research community nationally and internationally.  Shawn has also been a leading scientific collaborator and principal investigator on numerous multi-institutional NIH grants, including RECOVER, eMerge, ACT, AoU, PCORI, SHRINE and other major award programs that helped elevate MGB into a leader in translational informatics. His work has shaped standards for privacy-preserving data sharing, cohort discovery tools, and clinical data harmonization, influencing research data operations across Mass General Brigham hospitals and enabling thousands of studies per year.

“We are delighted to have Dr. Murphy join UW, bringing decades of expertise in informatics and data science research and operational work around instrumenting the health care enterprise for discovery in the course of clinical care,” said Dr. Peter Tarczy-Hornoch, Chair of BIME and Interim Director for IMDS.

“ITHS welcomes Dr. Murphy to this crucial role.   He is exactly the right person to take on the new informatics challenges of clinical and translational research,” said Dr. John Amory, Principal Investigator of ITHS and Associate Dean of Translational Sciences. 

Dr. Murphy will be bringing this rich and deep expertise to his UW roles as BIME faculty, Neurology faculty, UW Medicine CRIO, ITHS informatics lead and Institute for Medical Data Science Director.

“I am excited to welcome Dr. Murphy to UW Medicine and confident that his decades of leadership in research informatics at Mass General Brigham will significantly advance our research, data science, and innovation efforts,” said Eric Neil, UW Medicine Chief Information Officer.

Dr. Murphy’s position begins effective February 1, 2026.

 

Biomedical Informatics and Medical Education Newsletter

 

May 11, 2026 – May 15, 2026

UPCOMING LECTURES AND SEMINARS
BIME 590
Presenter: Maria Adela Grando, PhD, FAMIA, FACMI
Thursday, May 21^st – 11-11:50 am
850 Republican Street, Building C, Room 123 A/B
Zoom Information: https://washington.zoom.us/my/bime590
Speaker will present via Zoom

Title: Building Trustworthy AI for Mental Health: Ethical Design, Risks, and Real-World Evaluation.

Abstract: This talk examines the rapid growth of AI-driven mental health applications from a biomedical informatics perspective, with a focus on data governance, model robustness, and real-world evaluation. Consumer-facing tools—particularly LLM-based chatbots—raise concerns about transparency, safety, and the handling of highly sensitive mental health data.

The presentation introduces the FUTURE-AI framework as a structured approach to operationalizing ethical principles into measurable system requirements, including robustness, fairness, and human-centered design. Through case examples, it highlights how informatics methods can support the development and evaluation of trustworthy AI systems in mental health.

Speaker Bio: Dr. Grando is a Professor of Biomedical Informatics at Arizona State University and Director of Research Education. Her work focuses on consumer health informatics, human-centered AI, and patient-centered decision support, with an emphasis on data privacy, consent, and ethical AI in healthcare. She has led multiple federally funded projects, including NIH-supported research on sensitive health data sharing and AI-driven health technologies, and serves as Co-PI on large-scale initiatives addressing substance use disorder and health equity. Dr. Grando is also Co-founder and Chief Scientific Officer of ComplyLight, a company focused on data privacy and regulation-based medical record sharing.

UPCOMING EXAMS

General Exam
Title: Modeling Linguistic Signals of Behavioral Change: From Online Forums to Mental Health Dialogues
Student: Oliver Li
Date/Time: Monday, May 18, 8-10am PT
Zoom: https://washington.zoom.us/my/cohenta

Abstract: Modifiable behavioral factors, such as tobacco use and delays in seeking or adhering to appropriate care, contribute substantially to preventable morbidity and mortality from chronic disease. Effective interventions promoting healthier behaviors that align with individuals’ readiness to change have the potential to improve chronic disease management. However, scalable and interpretable computational frameworks for modeling behavioral stages and detecting help-seeking advancement from naturalistic online discourse remain limited.

To address these limitations, in the proposed research I will develop and apply language-based methods for characterizing behavioral change in two domains: smoking/vaping cessation and mental health help-seeking among youth at clinical high risk for psychosis. Because the study of stage-relevant behavioral signals first requires reliable inference of behavioral stages, this work begins by addressing a key methodological challenge: supervised learning models for stage classification are often limited by class imbalance. To address this data imbalance challenge, I propose to develop a concept-guided data augmentation framework that uses large language models to induce stage-relevant concepts and guide synthetic data generation. This approach is designed to improve transparency, conceptual diversity, and downstream behavioral stage classification in class-imbalanced settings.

Using models trained with the proposed augmented data, I will apply stage classification methods to vaping cessation forum posts and integrate predicted stages with communication themes and Behavioral Change Techniques to characterize stage-specific linguistic profiles and identify signals of progression, stagnation, and relapse. Finally, I will examine text-based dialogues between Mental Health America platform navigators and youth at risk for psychosis to identify linguistic, behavioral, and interactional markers of help-seeking advancement or disengagement.

 

Final Exam
Title: Towards Trustworthy Modeling of Patient Trajectory with Longitudinal Electronic Health Records
Student: Sihang Zeng
Date/Time: Friday, May 22, 2-4pm PT
In-person location: SLU C122
Zoom: https://washington.zoom.us/my/melihay

 

Abstract: Patient trajectory modeling, which predicts future clinical events using data from longitudinal electronic health records (EHRs), is expected to be of value for personalizing disease management. Yet the adoption of powerful deep learning (DL) models is often hindered by their “black-box” nature, creating a barrier to clinical trust. This challenge is compounded when modeling complex temporal dependencies between lab test results, treatments, and clinical events in the EHR, and is further exacerbated by the persistent difficulty these models face in generalizing across diverse patient populations, varying data elements, and different disease states.

This dissertation develops novel interpretable and generalizable frameworks for patient trajectory modeling, motivated by the hypothesis that models that account for the full dynamics of a patient’s history will produce more reliable predictions than simpler models, and that these predictions can also be made transparent and robust across diverse clinical contexts. The work is structured around four complementary aims that collectively affirm this hypothesis while innovating in terms of both deep learning methods and interpretable learning tools. It begins by developing an interpretable and generalizable deep learning framework for predicting survival in metastatic prostate cancer from pre-metastasis serial PSA values and treatments, establishing the value of trajectory-based modeling in a focused clinical setting. Building on this foundation, the work then advances from discrete-time modeling to a more precise continuous-time framework by introducing a model that learns continuous latent trajectories and uses a divide-and-conquer interpretation to explain how clinical changes drive outcomes. To broaden generalizability beyond training task-specific models, the dissertation next develops a multi-agent system that leverages a chain of large language model (LLM) agents with a long-term memory to reason over long, noisy, and heterogeneous EHR data for zero-shot cancer early detection. Finally, the work enables the self-evolving capability of this multi-agent system through an evolving experience pool and multi-agent reinforcement learning for lung cancer early detection, allowing the system to continuously adapt to new patient cohorts.

Through these complementary aims, this research traces an arc from task-specific prediction to generalizable and self-evolving reasoning, powered by DL-based sequential models and LLM-based systems. It shows that faithfully modeling the temporal information in patient histories can make the predictions accurate, robust, and interpretable, with interpretability and generalizability advancing together rather than in tension. In doing so, this dissertation seeks to contribute to the development of more trustworthy AI tools that can support personalized clinical decision-making across a spectrum of complex medical domains.

 

May 1, 2026 – May 8, 2026

UPCOMING LECTURES AND SEMINARS
BIME 590
Presenter: Eric Horvitz, MD, PhD
Thursday, May 14^th – 11-11:50 am
850 Republican Street, Building C, Room 123 A/B
Zoom Information: https://washington.zoom.us/my/bime590
Speaker will present via Zoom

 

Title: AI and Medical Reasoning: Advances, Limitations, and Frontiers

Abstract: Medical reasoning has long rested on statistical and decision-theoretic foundations, including probabilistic inference under uncertainty, utility-sensitive treatment choice, and the value of information in deciding what to observe next. Recent advances in AI are creating new opportunities for clinical decision support, while also raising challenges in probabilistic calibration, transparency, and alignment with clinical utilities and patient preferences. In this talk, I will review the evolving role of AI in medical decision support, moving from classical approaches to current frontiers including multimodal models, real-time reasoning, and sequential diagnostic strategies that account for the costs and benefits of additional testing. I will highlight recent studies demonstrating both the promise and limitations of these methods, and describe emerging work on coupling the capabilities of frontier AI with formal probabilistic inference and decision analysis. I will conclude with the importance of thoughtful translation of core AI capabilities into clinical workflows, and the opportunity for innovation through designs that promote effective clinician–AI collaboration.

 

Speaker Bio: Dr. Eric Horvitz is Microsoft’s Chief Scientific Officer. His career has focused on advancing artificial intelligence and its applications in health care and the biosciences. He is known for pioneering work in AI for diagnostic reasoning, predictive modeling, and decision-making under uncertainty. He has been a leading innovator in building and deploying predictive models for outcomes such as readmissions, hospital-associated infections, and clinical deterioration. Beyond core advances in AI, he has pursued methods and studies on clinician-AI collaboration, with a focus on using AI to augment clinical expertise. Dr. Horvitz has received the Allen Newell Award and the Feigenbaum Prize for advances in AI and has been elected to multiple honorific societies. He has served on advisory boards for the NIH and NSF, and on the President’s Council of Advisors on Science and Technology, where he co-chaired the influential national report on Transforming Patient Safety. More recently, he co-authored the National Academy of Medicine’s AI Code of Conduct. He earned his MD and PhD from Stanford University.

 

ANNOUNCEMENTS
Please join us in congratulating Aparajita Saha who successfully passed her General Exam!

Title: A unified framework for early detection, disease trajectory analysis and equitable prediction of autoimmune disease using longitudinal clinical text and patient records

Abstract: Autoimmune diseases are often difficult to diagnose early because symptoms emerge gradually and vary widely across patients. Delayed diagnosis can lead to prolonged patient suffering and irreversible organ damage. Early manifestations are frequently documented in longitudinal clinical notes and electronic health records long before formal diagnosis, yet these signals remain underutilized by current diagnostic or predictive models. This general exam proposes a longitudinal framework for identifying early disease signals from electronic health records using large language models and time-aware machine learning approaches. The work focuses on detecting pre-diagnostic patterns from clinical notes, characterizing disease and comorbidity trajectories over time, examining outcome-wide associations across these trajectories and developing robust evaluation methods for longitudinal prediction. While centered on autoimmune diseases, this framework may also inform earlier detection and trajectory modeling for other chronic systemic conditions.

 

UPCOMING EXAMS

Final Exam
Title: Towards Trustworthy Modeling of Patient Trajectory with Longitudinal Electronic Health Records
Student: Sihang Zeng
Date/Time: Friday, May 22, 2-4pm PT
In-person location: SLU C122
Zoom: https://washington.zoom.us/my/melihay

 

Abstract: Patient trajectory modeling, which predicts future clinical events using data from longitudinal electronic health records (EHRs), is expected to be of value for personalizing disease management. Yet the adoption of powerful deep learning (DL) models is often hindered by their “black-box” nature, creating a barrier to clinical trust. This challenge is compounded when modeling complex temporal dependencies between lab test results, treatments, and clinical events in the EHR, and is further exacerbated by the persistent difficulty these models face in generalizing across diverse patient populations, varying data elements, and different disease states.

This dissertation develops novel interpretable and generalizable frameworks for patient trajectory modeling, motivated by the hypothesis that models that account for the full dynamics of a patient’s history will produce more reliable predictions than simpler models, and that these predictions can also be made transparent and robust across diverse clinical contexts. The work is structured around four complementary aims that collectively affirm this hypothesis while innovating in terms of both deep learning methods and interpretable learning tools. It begins by developing an interpretable and generalizable deep learning framework for predicting survival in metastatic prostate cancer from pre-metastasis serial PSA values and treatments, establishing the value of trajectory-based modeling in a focused clinical setting. Building on this foundation, the work then advances from discrete-time modeling to a more precise continuous-time framework by introducing a model that learns continuous latent trajectories and uses a divide-and-conquer interpretation to explain how clinical changes drive outcomes. To broaden generalizability beyond training task-specific models, the dissertation next develops a multi-agent system that leverages a chain of large language model (LLM) agents with a long-term memory to reason over long, noisy, and heterogeneous EHR data for zero-shot cancer early detection. Finally, the work enables the self-evolving capability of this multi-agent system through an evolving experience pool and multi-agent reinforcement learning for lung cancer early detection, allowing the system to continuously adapt to new patient cohorts.

Through these complementary aims, this research traces an arc from task-specific prediction to generalizable and self-evolving reasoning, powered by DL-based sequential models and LLM-based systems. It shows that faithfully modeling the temporal information in patient histories can make the predictions accurate, robust, and interpretable, with interpretability and generalizability advancing together rather than in tension. In doing so, this dissertation seeks to contribute to the development of more trustworthy AI tools that can support personalized clinical decision-making across a spectrum of complex medical domains.

 

April 27, 2026 – May 7, 2026

UPCOMING LECTURES AND SEMINARS
BIME 590
Presenters for NLM Panel: Lauren Fanning, M.P.H., Ph.C., Nick Jackson, PhD Student, Emma Croxford, Ph.C.
Thursday, May 7^th – 11-11:50 am
Zoom Only, No In-Person option: https://washington.zoom.us/my/bime590

Speaker Lauren Fanning
Title: The impact of cigarette smoking on gene expression in preinvasive colorectal lesions and effect modification by alcohol consumption and lesion histology

Abstract: Colorectal cancer (CRC) is one of the most common and deadly cancers in the U.S. Cigarette smoking is a risk factor for CRC and preinvasive colorectal lesions, though the mechanisms are not completely understood. Smoking is implicated with alterations in the host immune response and chronic inflammation, which is a known instigator of carcinogenesis. We evaluated the difference in gene expression in preinvasive colorectal lesions by smoking status, duration and intensity. Expression of innate, adaptive, and other gene groups in 438 lesions were contrasted by smoking status, duration and intensity, adjusting for age, sex, alcohol consumption and histologic type. Global gene expression was assessed using omnibus distance-based multivariate tests for among-group () and between-group () differences by smoke exposures. Variations in individual genes were assessed using limma DE analysis from the NanoTube R package. We found global differences in innate gene expression by smoking status, duration and intensity. DE analysis showed overexpression of complement genes specific to the lectin pathway in low duration smokers relative to never smokers. Effect modification by alcohol consumption and histologic type was observed, where differences in innate gene expression were only present in nondrinkers and in tubular adenomas.

 

Speaker Bio: Lauren Fanning is a Ph.D. candidate in the Epidemiology program at the Medical University of South Carolina, working under Dr. Alexander Alekseyenko and Dr. Kristin Wallace. Her research focuses on the impact of cigarette smoking on the tumor immune microenvironment in preinvasive colorectal lesions, specifically evaluating T-cell densities and gene expression. Lauren’s presentation will feature her dissertation work on evaluating gene expression by different measures of smoke exposure using omnibus distance-based methods and differential expression analysis. Lauren received a Bachelor of Science degree in biology from the College of Charleston and a Master of Public Health in Epidemiology degree at the Medical University of South Carolina. She is a pre-doctoral trainee in the Biomedical Informatics & Data Science for Health Research (SC BIDS4Health) T15 Training Program.

 

Speaker Nick Jackson
Title: Improving the Scientific Validity of Synthetic Data

Abstract: Biomedical research is constrained by barriers to patient-level data sharing imposed by privacy regulations, limiting large-scale collaboration and reproducibility. Synthetic data generation has emerged as a practical solution, enabling researchers to share datasets that emulate real statistical patterns without exposing individual records. The value of synthetic data depends entirely on whether it preserves the specific statistical relationships that downstream analyses require. Yet, existing methods optimize general measures of statistical similarity and provide no mechanism for researchers to specify or enforce particular properties in the synthetic data. To address this, we introduce a reinforcement learning-based generative framework that allows researchers to directly specify and enforce statistical patterns of interest in synthetic data. Evaluated across several biomedical benchmarks, our approach substantially improves the preservation of scientifically meaningful relationships with minimal cost to data fidelity and no measurable impact on privacy.

 

Speaker Bio: Nick received his bachelor’s degree in computer science from the University of Florida. He is currently a 3rd-year PhD student at Vanderbilt University under the mentorship of Dr. Bradley Malin. His research focuses on developing methods to share biomedical data while preserving participant privacy. He has recently been awarded a F31 dissertation award where he will study methods to improve synthetic data generation, with a focus on improving AI-driven science for underrepresented populations.

 

Speaker Emma Croxford
Title: LLM-as-a-Judge: Automating and Scaling Generative AI evaluations in Medicine

Abstract: Electronic Health Records (EHRs) contain vast amounts of clinical data, yet providers often struggle to synthesize this information into clear and actionable insights. Large language models (LLMs) offer a promising solution through automated summarization, with the potential to reduce cognitive burden and improve clinical efficiency. However, the safe deployment of these systems requires rigorous methods to evaluate the accuracy, completeness, and clinical usefulness of generated summaries. In collaboration with Epic, our team developed and validated the Provider Documentation Summarization Quality Instrument (PDSQI-9) – a structured rubric for expert medical evaluation of LLM-generated summaries.

While expert review remains the gold standard, it is resource-intensive and difficult to scale across real-world clinical environments. To overcome this limitation, we then introduce an LLM-as-a-Judge framework that automates evaluation by aligning model-based assessments with PDSQI-9 criteria. This approach enables rapid, standardized scoring of summaries while maintaining fidelity to expert-defined quality dimensions. Our results demonstrate that LLM-based evaluators achieve high agreement with human raters across multiple domains, while reducing evaluation time from minutes to seconds.

Together, these contributions provide a scalable and practical framework for assessing LLM-generated clinical documentation. This work supports the development of trustworthy AI systems and offers a pathway for continuous monitoring and quality assurance as LLMs are integrated into routine clinical workflows.

 

Speaker Bio: Emma Croxford is a Ph.D. student in the Biomedical Data Science program at the University of Wisconsin–Madison, advised by Dr. Majid Afshar. Her work focuses on the evaluation of large language models in clinical settings, with an emphasis on developing and validating frameworks for assessing the quality, accuracy, and reliability of AI-generated medical summaries.

Emma received her Bachelor of Science in Mathematics and Data Science from the University of Indianapolis. She is actively involved in interdisciplinary efforts at the intersection of data science, healthcare, and responsible AI, with a particular interest in improving clinical workflows through trustworthy AI systems.

 

PAPERS, PUBLICATIONS & PRESENTATIONS

• A new Sabin publication. This paper is especially interesting- COVID 19 frontline providers’ perspectives on stigmatization during Covid.Sabin JA, Kett PM, Mohammed SA, Frogner BK, Lee D. Frontline clinicians’ perceptions of the stigmatization of covid-19 and suggestions to prevent future health-related stigma: a qualitative study. BMC Public Health. 2026/04/23 2026;doi:10.1186/s12889-026-27463-5

 

• Dror Ben-Zeev NIH Profile: https://www.fic.nih.gov/News/GlobalHealthMatters/march-april-2026/Pages/improving-treatment-reducing-symptom-severity-for-people-with-serious-mental-illness.aspx

 

• Feng Chen, Weizhe Xu, Changye Li, Serguei Pakhomov, Alex Cohen, Simran Bhola, Sandy Yin, Sunny X. Tang, Michael Mackinley, Lena Palaniyappan, Dror Ben-Zeev, Trevor Cohen,
  Reading Between the Lines: Combining Pause Dynamics and Semantic Coherence for Automated Assessment of Thought Disorder. Accepted by Neuropsychologia, special issue: Language in psychosis: a multidisciplinary approach. DOI: 10.1016/j.neuropsychologia.2026.109473.

 

• Ze Cai, Hanzhe Liang, Sihang Zeng, Binbin Zhou, Jun Wen. CONTEXTOR: Contextualized High-order Contrastive Learning. ICML 2026.

 

UPCOMING EXAMS
General Exam
Title: A unified framework for early detection, disease trajectory analysis and equitable prediction of autoimmune disease using longitudinal clinical text and patient records
Student: Aparajita Saha
Date/Time: Monday, May 4, 11-1pm PT
In-person location: Institute for Systems Biology, Room- 432
Zoom: https://isbscience.zoom.us/j/88432669777?pwd=aEQBOhkzKiLqbbXcNKhIxld7kWT2Hq.1
Meeting ID: 884 3266 9777
Passcode: 507282

Abstract: Autoimmune diseases are often difficult to diagnose early because symptoms emerge gradually and vary widely across patients. Delayed diagnosis can lead to prolonged patient suffering and irreversible organ damage. Early manifestations are frequently documented in longitudinal clinical notes and electronic health records long before formal diagnosis, yet these signals remain underutilized by current diagnostic or predictive models. This general exam proposes a longitudinal framework for identifying early disease signals from electronic health records using large language models and time-aware machine learning approaches. The work focuses on detecting pre-diagnostic patterns from clinical notes, characterizing disease and comorbidity trajectories over time, examining outcome-wide associations across these trajectories and developing robust evaluation methods for longitudinal prediction. While centered on autoimmune diseases, this framework may also inform earlier detection and trajectory modeling for other chronic systemic conditions.

 

April 20, 2026 – April 24, 2026

UPCOMING LECTURES AND SEMINARS
BIME 590
Presenter:  Shawn Murphy, MD, PhD,
Thursday, April 30^th – 11-11:50 am
850 Republican Street, Building C, Room 123 A/B
Zoom Information: https://washington.zoom.us/my/bime590
Speaker will present In-Person

Title: Enhancing the Digital Representation of the Patient using Computational Phenotypes and Generative AI

Abstract: The use of real-world electronic health record data for research has enabled massive advances in how we understand health and disease in our patients.   However, use of the data should always be done with the understanding that it is often not collected for research and therefore may be fraught with inconsistencies.   Enabling the “digital twin” of the patient often involves applying complex algorithms and AI paradigms to result in a highly accurate digital representation.

 

Speaker Bio: Shawn Murphy serves as the Chief Research Information Officer at UW Medicine and as Professor of Biomedical Informatics and Medical Education and Professor of Neurology.  Over his prior 30 years he has been an integral leader and innovator at Massachusetts General Hospital.  His work has shaped standards for privacy-preserving data sharing, cohort discovery tools, and clinical data harmonization, influencing research data operations across hospitals and enabling thousands of studies per year to use real world data for science.

 

PAPERS, PUBLICATIONS & PRESENTATIONS
On Tuesday, April 21^st, Oliver Bear Don’t Walk participated in a panel organized by Sámi University of Applied Sciences, Sámi Parliament in Norway, Norwegian Ministry of Local Government and Regional Development!

The panel was during a Side Event at the 25th Session of the United Nations Permanent Forum on Indigenous Issues. His talk was titled “Artificial Intelligence and Indigenous Health Knowledge”.

Event Details:
AI Opportunities for Indigenous Health: Realizing benefits and addressing risks

New health AI applications for diagnosis, treatment, and care will reshape healthcare systems, and they offer major opportunities to reduce inequities faced by Indigenous Peoples by enabling new services that better support Indigenous languages and cultures. However, if Indigenous Peoples are absent from training and evaluation data, AI tools can underperform, contributing to poorer decision support, misprioritization, and unequal access to timely care. Therefore, urgent actions are needed to remove barriers to inclusion, especially access to relevant data, robust evaluation for Indigenous Peoples’ populations, and accountability in how health AI is designed, developed, and used for Indigenous Peoples.

Speakers
Prof. Lars Ailo Bongo, Sámi AI Lab, Sámi University of Applied Sciences
Maren B. N. Storslett, Member of Governing Council, Sámi Parliament in Norway
Sigrid Ina Simonsen, State Secretary, Norwegian Ministry of Local Government and Regional Development
Dr. Oliver J. Bear Don’t Walk IV, University of Washington
TBA, Member of UNPFII
Risten Rajala, Sámi Path Finders, Sámi University of Applied Sciences.

 

UPCOMING EXAMS

General Exam
Title: A unified framework for early detection, disease trajectory analysis and equitable prediction of autoimmune disease using longitudinal clinical text and patient records
Student: Aparajita Saha
Date/Time: Monday, May 4, 11-1pm PT
In-person location: Institute for Systems Biology, Room- 432
Zoom: https://isbscience.zoom.us/j/88432669777?pwd=aEQBOhkzKiLqbbXcNKhIxld7kWT2Hq.1
Meeting ID: 884 3266 9777
Passcode: 507282

Abstract: Autoimmune diseases are often difficult to diagnose early because symptoms emerge gradually and vary widely across patients. Delayed diagnosis can lead to prolonged patient suffering and irreversible organ damage. Early manifestations are frequently documented in longitudinal clinical notes and electronic health records long before formal diagnosis, yet these signals remain underutilized by current diagnostic or predictive models. This general exam proposes a longitudinal framework for identifying early disease signals from electronic health records using large language models and time-aware machine learning approaches. The work focuses on detecting pre-diagnostic patterns from clinical notes, characterizing disease and comorbidity trajectories over time, examining outcome-wide associations across these trajectories and developing robust evaluation methods for longitudinal prediction. While centered on autoimmune diseases, this framework may also inform earlier detection and trajectory modeling for other chronic systemic conditions.

 

April 13, 2026 – April 17, 2026

UPCOMING LECTURES AND SEMINARS
BIME 590
Presenter:  Uba Backonja, PhD, MS, RN
Thursday, April 23^rd – 11-11:50 am
850 Republican Street, Building C, Room 123 A/B
Zoom Information: https://washington.zoom.us/my/bime590

Speaker will present Via Zoom

Title: Applying user-centered design principles to develop HL7® FHIR® resources for post-acute care data interoperability

Abstract: Data silos are a major issue across healthcare, especially in post-acute care (PAC) like skilled nursing facilities and home health. These silos can negatively impact patient outcomes due to lack of complete healthcare data available to healthcare providers. To address this silo issue, the Centers for Medicare & Medicaid Services (CMS) and MITRE have partnered to create HL7® Fast Healthcare Interoperability Resources (FHIR®) for health IT developers so there can be a standardized way to collect and transfer data across settings (https://pacioproject.org). To create these resources, MITRE applies user-centered design (UCD) principles to collaborate with providers, vendors, and other PAC partners to understand and address interoperability gaps that can be addressed via FHIR. This talk will describe FHIR basics, unique data needs for PAC settings, and how MITRE uses UCD to collaborate with partners throughout the FHIR resource development lifecycle.

 

Speaker Bio: Dr. Uba Backonja (pronounced “OO-ba BAHch-ko-nya”) is a researcher with expertise in health informatics, UCD, community and public health nursing, anthropology, data visualization, and other topics. She completed her training at the University of Wisconsin-Madison, was a Predoctoral Fellow @the NIH/NICHD, and a Postdoctoral Fellow @BIME. Before MITRE, Dr. Backonja was faculty @UW Tacoma School of Nursing and Healthcare Leadership and conducted research with BIME faculty and the UW’s Northwest Center for Public HeDalth Practice (e.g., https://sharenw.nwcphp.org). You can find her on LinkedIn and PubMed.

PAPERS, PUBLICATIONS & PRESENTATIONS

• DeconDTN-Toolkit: A Library for Evaluation and Enhancement of Robustness to Provenance Shift. Yongsen Tan, Zhecheng Sheng, Xiruo Ding, Serguei V. S. Pakhomov, Trevor Cohen. Accepted: 7th Annual Conference on Health, Inference, and Learning (CHIL). Seattle, WA. 2026.

 

• SocialLM: Using LLMs and contextual aggregation to track patient-provider communication. Manas Bedmutha, Feng Chen, Andrea L Hartzler, Trevor Cohen, Nadir Weibel.  Accepted: 7th Annual Conference on Health, Inference, and Learning (CHIL). Seattle, WA. 2026.

 

• Wen, J., Zeng, S., Bonzel, CL. et al. Phenotypic prediction of missense variants via deep contrastive learning. Nat. Biomed. Eng (2026). https://doi.org/10.1038/s41551-026-01636-4

 

ANNOUNCEMENTS
Please join us in congratulating Danner Peter who successfully passed his General Exam!

Title: Responsible Predictive Modeling for Tribal Health: Integrating Fairness Evaluation and Indigenous Data Governance

Abstract: Predictive modeling has the potential to support health equity in Tribal communities, yet most models are developed using data that underrepresent American Indian and Alaska Native (AIAN) populations and are rarely evaluated for fairness or validity in Indigenous health contexts. This dissertation addresses a critical gap at the intersection of Tribal health priorities, algorithmic fairness, and Indigenous Data Governance (IDGov) by asking: under what conditions can predictive modeling be responsibly and beneficially used to support Tribe-identified health priorities? This dissertation pursues three aims: (1) in collaboration with the Northwest Tribal Epidemiology Center (NWTEC), identify community-defined health priorities suitable for predictive analytic approaches; (2) develop and evaluate predictive models with explicit AIAN fairness assessment using the NIH All of Us dataset; and (3) assess governance-informed readiness for predictive modeling within a Tribal Epidemiology Center. Together, these aims integrate technical evaluation with IDGov principles to determine not only how models perform, but when their use is appropriate or inadvisable. This work contributes structured, governance-aligned criteria that Tribes and Tribe-serving organizations could use to evaluate the feasibility and community impacts of predictive modeling.

______________________________________________

Are you interested in research on digital health, social media data, patient experiences, and the criminal legal system? I am looking for a graduate student collaborator to work together on qualitative coding of Reddit posts examining substance use and health behaviors among individuals under community supervision (e.g., probation, parole).

Objective: The goal of this project is to better understand how individuals describe their experiences with community supervision and how these experiences shape substance use, recovery, and broader health-related behaviors. Findings from this work aim to improve understanding of patient experiences and inform public health research and policy for criminal legal–involved populations.

The graduate student collaborator will:

• Conduct qualitative coding of Reddit posts using a semi-structured coding framework
• Participate in regular meetings to discuss coding decisions and refine code definitions
• Contribute to interpretation of findings and potential manuscript development

This is a collaborative opportunity with potential for authorship, continued involvement, and participation in future related projects. Our team includes myself, Dr. Annie Chen, and Dr. Mandy Owens.

Preferred Background:

• Graduate student
• Experience with qualitative coding or qualitative research methods
• Interest in digital health, substance use, or criminal legal system research

Time Commitment:

• Approximately 5–7 hours per week
• Minimum duration of ~10 weeks for coding (please note the possibility of an extended timeline if necessary)
• Includes a weekly meeting plus independent coding time

If you’re interested or would like more information, please send an email to mamagid@uw.edu.

 

UPCOMING EXAMS

General Exam
Title: A unified framework for early detection, disease trajectory analysis and equitable prediction of autoimmune disease using longitudinal clinical text and patient records
Student: Aparajita Saha
Date/Time: Monday, May 4, 11-1pm PT
In-person location: Institute for Systems Biology, Room- 432
Zoom: https://isbscience.zoom.us/j/88432669777?pwd=aEQBOhkzKiLqbbXcNKhIxld7kWT2Hq.1
Meeting ID: 884 3266 9777
Passcode: 507282

Abstract: Autoimmune diseases are often difficult to diagnose early because symptoms emerge gradually and vary widely across patients. Delayed diagnosis can lead to prolonged patient suffering and irreversible organ damage. Early manifestations are frequently documented in longitudinal clinical notes and electronic health records long before formal diagnosis, yet these signals remain underutilized by current diagnostic or predictive models. This general exam proposes a longitudinal framework for identifying early disease signals from electronic health records using large language models and time-aware machine learning approaches. The work focuses on detecting pre-diagnostic patterns from clinical notes, characterizing disease and comorbidity trajectories over time, examining outcome-wide associations across these trajectories and developing robust evaluation methods for longitudinal prediction. While centered on autoimmune diseases, this framework may also inform earlier detection and trajectory modeling for other chronic systemic conditions.

 

April 6, 2026 – April 10, 2026

UPCOMING LECTURES AND SEMINARS
BIME 590
Presenter:  Christopher Longhurst, MD, MS
Thursday, April 16^th – 11-11:50 am
850 Republican Street, Building C, Room 123 A/B
Zoom Information: https://washington.zoom.us/my/bime590
Speaker will present In-Person

Title: Fostering Patient-Centered Communication in Clinical Note Writing

Abstract:
Details to come next week!

Speaker Bio:
As CEO of Seattle Children’s, Dr. Christopher Longhurst, MD, MS leads a world-class team of healthcare providers, researchers, and staff who are united by Children’s mission to provide hope, care and cures to help every child live the healthiest and most fulfilling life possible.

A practicing pediatrician for the past 25 years, Dr. Longhurst joined Seattle Children’s in 2026 after serving in a dual role as chief medical officer (CMO) and chief digital officer (CDO) at UC San Diego Health, as well as Professor of Biomedical Informatics and Pediatrics at UC San Diego School of Medicine. Combining a passion for innovation with the drive to improve quality, safety, equity and patient experience, Longhurst was instrumental in securing a philanthropic gift to establish the Joan & Irwin Jacobs Center for Health Innovation where he served as the center’s founding executive director leading the AI portfolio across the system.

Dr. Longhurst now serves as an affiliate Professor of Pediatrics and Biomedical Informatics at the University of Washington and continues to contribute thought leadership and scholarship in care quality, patient safety and health informatics. He has published over 150 peer-reviewed articles in journals like the New England Journal of Medicine, JAMA and Pediatrics, and serves on the National Academy of Medicine steering committee for Patient Safety in the AI Era.

Before joining UC San Diego Health, Dr. Longhurst spent 15 years at Stanford University, serving as chief medical information officer for Stanford Children’s Health, where he led efforts to improve children’s health and provider workflow using information technology. He also founded and led the nation’s first accredited clinical informatics fellowship at Stanford, where he was a clinical professor of pediatrics and biomedical informatics.

Dr. Longhurst is an elected fellow of the prestigious American College of Medical Informatics. He earned his medical degree and Master of Science in Medical Informatics from UC Davis, completed his residency at Stanford University, and holds a Bachelor of Science from UC San Diego.

 

PAPERS, PUBLICATIONS & PRESENTATIONS

• Amanda Hall: Exploring the Future of AI in Clinical Collaboration: A Study on Tumor Board Case Preparation – Microsoft Research
• Amanda Hall: Not Another EHR: Reimagining Physician Information Needs with Generative AI Technology – Microsoft Research
• Ruyuan Wan, Changye Li, Ting-Hao Kenneth Huang. “Newspaper Eat” Means “Not Tasty”: A Taxonomy and Benchmark for Coded Languages in Real-World Chinese Online Reviews. To appear in ACL 2026. A preprint can be found here: https://arxiv.org/abs/2601.19932
• SocialLM: Using LLMs and contextual aggregation to track patient-provider communication” Manas Bedmutha, Feng Chen, Andrea L Hartzler, Trevor Cohen, Nadir Weibel accepted by CHIL 2026 in Seattle.

 

UPCOMING EXAMS
General Exam
Title: Responsible Predictive Modeling for Tribal Health: Integrating Fairness Evaluation and Indigenous Data Governance
Student: Danner Peter
Date/Time: Thursday April 16, 8-10 am PT
In-person location: 850 Republican St., Building C, Room C122
Zoom: https://washington.zoom.us/my/peter.th

Abstract: Predictive modeling has the potential to support health equity in Tribal communities, yet most models are developed using data that underrepresent American Indian and Alaska Native (AIAN) populations and are rarely evaluated for fairness or validity in Indigenous health contexts. This dissertation addresses a critical gap at the intersection of Tribal health priorities, algorithmic fairness, and Indigenous Data Governance (IDGov) by asking: under what conditions can predictive modeling be responsibly and beneficially used to support Tribe-identified health priorities? This dissertation pursues three aims: (1) in collaboration with the Northwest Tribal Epidemiology Center (NWTEC), identify community-defined health priorities suitable for predictive analytic approaches; (2) develop and evaluate predictive models with explicit AIAN fairness assessment using the NIH All of Us dataset; and (3) assess governance-informed readiness for predictive modeling within a Tribal Epidemiology Center. Together, these aims integrate technical evaluation with IDGov principles to determine not only how models perform, but when their use is appropriate or inadvisable. This work contributes structured, governance-aligned criteria that Tribes and Tribe-serving organizations could use to evaluate the feasibility and community impacts of predictive modeling.