Biomedical Informatics and Medical Education > News and Events

News and Events

Chair’s Message

pth-use-this-one 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

Biomedical Informatics and Medical Education Newsletter

October 27, 2025 – October 31, 2025

UPCOMING LECTURES AND SEMINARS
BIME 590
Presenter: Changye Li, PhD
Thursday, November 6^th – 11-11:50 am
850 Republican Street, Building C, Orin Smith Auditorium (Not C123)
Zoom Information: https://washington.zoom.us/my/bime590
Speakers will present In-Person

Title: Informative Errors: Automatic Speech Recognition Application in Speech-based Behavioral Tasks

Abstract:
Neural language models (NLMs) have demonstrated impressive performance in distinguishing spontaneous speech samples produced by cognitively healthy individuals from those with Alzheimer’s disease dementia, particularly in responses to cognitive tasks. However, to fully realize the potential of NLMs for mental status assessment, these models typically require verbatim transcripts of participant speech, which creates a significant bottleneck for both research and clinical applications. While automatic speech recognition (ASR) systems have improved significantly in recent years, they still exhibit high error rates when processing spontaneous, disfluent, or impaired speech. In this talk, I will present findings from my previous work exploring how these ASR errors can serve as informative features in the downstream classification task. This line of research enables the automated analysis of speech collected from participants at least in the dementia screening settings, and it has the potential to expand to a variety of other clinical applications as well in which both language and speech characteristics are affected.

Speaker Bio:
Changye Li, Ph.D., is a Postdoctoral Scholar in the Department of Biomedical Informatics and Medical Education and Data Science Postdoctoral Fellow in eScience Institute at the University of Washington. Her research interests include better adapting and developing explainable natural language processing and speech processing models with biomedical/clinical and behavioral data. Prior to joining UW, she obtained her Ph.D. in Health Informatics at the University of Minnesota, where her dissertation focused on detecting cognitive impairment from language and speech for early screening of Alzheimer’s disease dementia using interpretable transformer-based language models.

ANNOUNCEMENTS
Dear BIME faculty, staff, students, and trainees,

The first frost of autumn glazed my backyard yesterday morning. In the hollows, thin ice formed over the fallen leaves beneath the dogwood tree. As reluctant as I am to let go of summer, there is an unmistakable excitement that accompanies the chilly air: the end of one thing begins something new.

Please celebrate the new season by joining me and the BIME staff on Thursday, November 6 at 4pm in the SLU Lounge (just off to the side of the main lobby near the front desk in Building C) for a lively social hour of fall crafts, autumn-infused mocktails, and nibbles!

_______________________
Please join us in congratulating the following students who were awarded a Graduate Student Conference Presentation Award from UW Graduation School for their upcoming presentations at AMIA:

Kaylin Ji

Cat Kim

Kevin Li

Ojas Ramwala

Aparajita Saha

Michael (Tianmai) Zhang

_______________________
Avery Yu – My previous colleagues is hiring a postdoc immediately:

JOB DESCRIPTION (SHORT):

Postdoctoral Researcher to advance research at the intersection of artificial intelligence for healthcare, multimodal data analysis (EHRs, medical imaging, omics, physiological signals, clinical notes), and causal AI (causal inference, discovery, counterfactual reasoning). The successful candidate will collaborate with an interdisciplinary team of computer scientists, biomedical informaticians, clinicians, and public health researchers to develop deployable, trustworthy methods that improve patient outcomes and health system operations.

JD link: https://indiana.peopleadmin.com/postings/30712

_______________________
In the annual AMIA signature awards, two of our alumni were featured:

Chunhua Weng (2005) was awarded the Don Lindberg award for Innovation in informatics.

Yue Guo (2024) was listed as a Finalist for the Ted Shortliffe Doctoral Dissertation Award.

The awards nicely pairs our very first graduate with one of our most recent graduates. I expect continued accolades for both of them!

–John Gennari

UPCOMING EXAMS

General Exam
Title: Advancing Predictive Modeling in Oncology Through Integrative Machine Learning Approaches
Student: Yein Jeon
Date/Time: Friday, October 31, 2025, 12pm – 2 pm PT
In-person location: 850 Republican Street, Building C, SLU C259
Zoom: https://washington.zoom.us/my/peter.th

Abstract: Targeted early interventions for high-risk patients can improve treatment outcomes, but they require accurate risk stratification. While machine learning (ML)–based predictive modeling has shown strong potential for identifying risk factors and improving prognostic accuracy, its application and robustness in novel immunotherapies such as Chimeric Antigen Receptor T (CAR-T) cell therapy, particularly using multimodal longitudinal data, remain underexplored. To address these gaps, we propose three specific aims: (1) identify the best-performing imputation methods for incomplete multivariate longitudinal blood analyte data from patients treated with cellular therapies ; (2) develop a multimodal model integrating diverse data modalities, such as imaging and longitudinal blood analyte data, to predict CAR-T therapy outcomes; and (3) develop a method for integrating incomplete, high-dimensional multi-omics data to improve predictive performance while maintaining explainability. Collectively, these aims will establish a methodological foundation for predictive modeling in CAR-T therapy, enabling earlier identification of high-risk patients and supporting targeted, data-driven clinical decision-making.

October 20, 2025 – October 24, 2025

UPCOMING LECTURES AND SEMINARS
BIME 590
Presenters:
Jared Slone, PhD Student
Spencer Halberg-Spencer, PhD Student
Aparajita Kashyap, PhD Student
Thursday, October 30^th – 11-11:50 am
850 Republican Street, Building C, Orin Smith Auditorium (Not C123)
Zoom Information: https://washington.zoom.us/my/bime590
Speakers will present Via Zoom

Speaker Jared Slone
Title: Leveraging 3D Protein Structures In Machine Learning to Further Development of Cellular Immunotherapies

Abstract: Strong binding between T cell receptors (TCRs) and peptide–HLA (pHLA) complexes plays a key role in the adaptive immune response. Accurately predicting which TCRs will bind to which pHLAs is central to advancing personalized immunotherapies such as TCR-T. Most existing ML approaches to TCR–pHLA binding prediction rely only on amino acid sequences, limiting their ability to capture the full complexity of binding interactions. Recent advances in protein structure modeling provide a unique opportunity to integrate 3D structural information into ML pipelines. We developed STAG-LLM, a multimodal model that combines protein language models with geometric deep learning to leverage both sequence data and computationally generated 3D structures of TCR–pHLA complexes in making predictions. Our results show that STAG-LLM outperforms existing methods, even when trained on datasets three times smaller. Our findings demonstrate that structure-based approaches can meaningfully improve prediction of TCR–pHLA specificity.

Speaker Bio:
Jared Slone received his bachelor’s degree in applied mathematics from Brigham Young University where he also minored in physics and computer science. He is currently a fifth-year PhD student at Rice University under the mentorship of Dr. Lydia Kavraki. His interests are centered around using mathematical models and machine learning to solve complex problems in healthcare. At Rice, Jared’s research has focused on developing models to predict interactions between proteins involved in the body’s immune response.

Speaker Spencer Halberg-Spencer
Title: SCOTCH: A method for simultaneous identification of cellular states and gene programs with single-cell orthogonal matrix tri-factorization

Abstract: Single-cell genomics enables the profiling of “omic” features such as gene expression and chromatin accessibility at cellular resolution. These data have transformed our ability to study heterogeneous populations of cells from diverse tissue, disease, and developmental contexts. A first step in the analysis of such data is to cluster groups of cells with similar expression profiles and annotate them to determine the sample cell type and state composition. Current approaches use two steps. Cell clusters are first identified then differentially expressed genes are utilized to annotate these clusters. This two-step process may not accurately capture gene expression programs that are important for multiple cellular states and are especially limited for less studied systems.

To address this gap, we have developed single-cell orthogonal tri-factorization for clustering high-dimensional data (SCOTCH) which allows for simultaneous identification of cell clusters and their associated genomic regulatory programs. We applied SCOTCH to a single-cell RNA sequencing (scRNA-seq) and single-cell assay for transposase-accessible chromatin sequencing (scATAC-seq) from labeled peripheral blood mononuclear cells (PBMC), a widely used benchmarking cell type. SCOTCH outperformed NMF and Louvain clustering for identification of hematopoietic cell types and simultaneously captured expression programs which define these cell types. We identify both unique expression programs and programs that are shared between cell types with similar biological function. We also performed well in clustering on scATAC-seq data relative to two state-of-the-art methods, ArchR and cisTopic. SCOTCH accurately identifies cell types from chromatin accessibility data and open regions of chromatin specific to each cell type.

Finally, we applied SCOTCH to a novel multi-sample dataset of aged mouse dermal fibroblasts transient reprogramming to induced pluripotent stem cell. Transient reprogramming is a process where somatic cells are briefly exposed to reprogramming factors, initiating pluripotency-associated pathways without loss of the original cell identity. This approach has been shown to reverse molecular hallmarks of aging, leading to cellular rejuvenation, though the underlying process is poorly understood. SCOTCH resolves the complex cellular landscape and identifies cell clusters correspond to the reprogramming, aging, and rejuvenating trajectories. Simultaneously, SCOTCH identifies distinct and shared gene programs which define these trajectories. Taken together, our results suggest that SCOTCH is a powerful and flexible approach that can be applied to genomic data to identify gene expression programs and accessible chromatin regions which define cellular state.

Speaker Bio:
Spencer Halberg-Spencer is a Ph.D. student in the Biomedical Data Science program at the University of Wisconsin-Madison, working in the laboratory of Dr. Sushmita Roy. His research focuses on gene regulatory network inference and the development of unsupervised learning methods for single cell sequencing data. During his presentation, Spencer will discuss his recent work on applying non-negative matrix tri-factorization to identify cell clusters and gene programs driving dynamics in complex biological systems.

Spencer received Bachelor of sciences degrees in biomedical engineering and mathematics form the University of Iowa, and a Master of Science degree in biomedical data science from UW-Madison. He is a fellow of the Computation and Informatics in Biology and Medicine (CIBM) training program and co-founder and former co-president of the Biomedical Data Science Student Society at UW-Madison.

Speaker Aparajita Kashyap
Title: Leveraging data-driven analyses to characterize subpopulation robustness of a schizophrenia prediction model

Abstract: Machine learning models in healthcare may lack robustness by learning to rely on spurious correlations rather than clinically actionable information. This is particularly problematic with high-dimensional observational data, where health status signals may be interwoven with confounding such as patient care access patterns. We investigate the robustness of a machine learning model designed to predict diagnostic transition from psychosis to schizophrenia. We focus on two questions: (1) How can we identify relevant subpopulations to assess model robustness? (2) How can robustness be eTectively evaluated in these subpopulations?

We develop a recurrent neural network that uses administrative claims data to predict schizophrenia onset. Our initial feature importance analysis found that 9 of the 10 most important features are either psychiatric features or healthcare utilization features. Notably, low- and moderate-complexity outpatient visits negatively correlated with predicted risk, while overall outpatient visit frequency positively correlated. Our feature stability analysis further suggested that the model may lack robustness on the basis of healthcare utilization. These ﬁndings suggest a strong hypothesis to analyze model robustness based on healthcare utilization. We devise a new measure of healthcare utilization with two modes (“psychiatric” and “non-psychiatric”). Psychiatric utilization refers to visits associated psychiatric diagnoses (ICD10 Codes: F00-F99).

When examining the relative levels of healthcare utilization among true positive (TP), false positive (FP), true negative (TN), and false negative (FN) predictions, we ﬁnd that true schizophrenia outcomes (TP and FN) were associated with higher inpatient and outpatient psychiatric utilization. This indicates that the model may focus more on psychiatric utilization as a predictor of schizophrenia and miss patterns related to non-psychiatric utilization.

Through analysis of healthcare utilization, we generated hypotheses about how our model learned about schizophrenia onset; this work emphasizes the importance of evaluating the robustness of models by combining data-driven insights with domain expertise.

Speaker Bio:
Aparajita (Apara) Kashyap is a PhD student at Columbia University in the Department of Biomedical Informatics. Her research focuses on building fair and equitable models for use in healthcare settings, with a focus on mental health disorders and reproductive health disorders. At Columbia, Aparajita is coadvised by Shalmali Joshi and Noémie Elhadad. Prior to her work at Columbia, Aparajita studied Biophysics at Johns Hopkins University.

ANNOUNCEMENTS
Dear BIME faculty, staff, students, and trainees,

UPCOMING EXAMS

PAPERS, PUBLICATIONS & PRESENTATIONS

On October 10th, Ziqing (Kaylin) Ji presented to UW Medicine Primary Care leaders at the monthly PCMD meeting on “EHR-Based Social Needs Screening and Referral in Primary Care: Clinician and Staff Perspectives on Practices, Barriers, and Benefits.” This project was a survey of primary care clinicians and staff about their readiness for standardized screening and referral in the electronic health record. Findings inform improvements to SDoH workflow and tools. This work was accepted as a student paper at AMIA, which Kaylin will present again in November!

Mentors: Andrea Hartzler and Pat Wedgeworth

Clinical Champions: Sara Jackson, Kathy Mertens and Nkem Akinsoto

September 29, 2025 – October 3, 2025

UPCOMING LECTURES AND SEMINARS
BIME 590
Presenter: Jim Phuong, MSPH, PhD
Thursday, October 9^th – 11-11:50 am
850 Republican Street, Building C, Orin Smith Auditorium (Not C123)
Zoom Information: https://washington.zoom.us/my/bime590
Speaker will present in-person

Title:
Secondary-use data and Real-world data Integration in National Research Consortia

Abstract:
Health systems are uniquely positioned to survey the health of their patient population, including the effects of natural hazards, disaster disruptions, and public health emergencies. As an integral part of the biomedical research landscape, Health systems data sharing supports data-driven research and iterative quality improvements to the data from each health system. Apart from clinical outcomes, health systems are gradually increasing their focus upon collecting and addressing gaps in understanding Social Determinants of Health (or Social Drivers of Health, SDoH) and their dynamic role in maintaining health and wellness. This includes integrating patient-level information as well as place-based information integrated from geocoding and secondary use of spatial-temporal datasets. In this talk, I will discuss the nexus of environmental health, disaster management and population health research, and biomedical informatics research. I will also discuss the directions from health system preparedness, the broader implications towards research data sharing and research consortia with a precision medicine focus, and the analytical capacities needed for research with multiple data types in cloud infrastructure.

Presenter Bio:
Jimmy Phuong obtained his Masters of Science in Public Health degree from University of North Carolina at Chapel Hill in 2014 and his Ph.D. degree in Biomedical Informatics and Medical Education from the University of Washington in 2020. Dr. Phuong’s research is at the intersection of secondary use of Electronic Health Records (EHRs), geospatial-temporal data integration, and informatics capacity development and maturity to enhance population health and precision medicine research. Since 2020, Dr. Phuong co-led the National Clinical Cohort Collaborative (N3C) Social Determinants of Health (SDoH) domain team, focusing efforts to enhance clinical research that integrates patient-level and community-level Social Determinants of Health information. He is the current Chair for the American Medical Informatics Association (AMIA) Informatics Maturity Working Group, site Principal Investigator for the NIH All of Us Research Program Center for Linkage and Acquisition of Data (CLAD) in which he leads Geocoding and Spatial-temporal data integration, and serves on the NIH-NCATS Clinical and Translational Science Award (CTSA) Real-World Data Workforce Development Task Force. I am also an active contributor to the CDC Region 10 Public Health Emergency Preparedness and Response (PHEPR) data ecosystem planning and its collaborative deliverables related to Public Health Data Modernization Initiatives. His interests include exploring cloud-based and research platform data modernization on issues like mass casualty emergencies, trauma outcomes research, and State-Tribal-Local-Territory (STLT) regional partnerships to prepare for and mitigate the impact of regional disasters and environmental health effects.

ANNOUNCEMENTS
Join us for the annual Science & Engineering Career Fair, hosted by the Science & Engineering Business Association (SEBA). This event brings together top employers and talented students for an afternoon of networking, recruiting, and career exploration. Employers will showcase job and internship opportunities across diverse industries, while students gain direct access to recruiters and career resources.
Link to student organization: here

📍 Location: Husky Union Building (HUB), North & South Ballrooms
📅 Date: October 15
🕚 Time: 11:30 AM – 3:30 PM
Whether you’re an employer seeking top talent or a student exploring future opportunities, the SEBA Career Fair is the place to connect, learn, and launch the next step in your career.

_______________________

Emerging Topics in Quality Improvement Webinar Series

Always free and open to the public, these webinars highlight trending issues and practical methodologies in quality improvement. Webinars take place on the second Tuesday every other month from 12-1pm PT. Registration is required to attend.

Join us for the next webinar on
Tuesday, October 14, 2025 from 12-1pm

Pathway Assistant: A Use Case in AI for Quality and Safety

Yu-Hsiang “Clara” Lin, MD
Chief Medical Information Officer and VP, Digital Health & Informatics
Seattle Children’s
Clinical Associate Professor
Department of Pediatrics, UW Medicine

Darren S. Migita, MD
Medical Director of Clinical Effectiveness, Center for Quality and Patient Safety, Seattle Children’s Hospital
Clinical Professor
Department of Hospital Medicine, UW Medicine

Register Now for this 10/14/25 Webinar

PAPERS, PUBLICATIONS & PRESENTATIONS

Ojas A. Ramwalawill be giving a talk on Explainable AI for Biomedical Image Processing at PyData Seattle, to be held November 7-9, 2025, at Bellevue College. Additional information can be found here: https://cfp.pydata.org/seattle2025/speaker/BVZDEJ/
Sihang Zeng, Yujuan Fu, Sitong Zhou, Zixuan Yu, Lucas Jing Liu, Jun Wen, Matthew Thompson, Ruth Etzioni, Meliha Yetisgen. Traj-CoA: Patient Trajectory Modeling via Chain-of-Agents for Lung Cancer Risk Prediction. NeurIPS 2025 GenAI4Health Workshop.
Tianmai M Zhang*, Zhaoyi Sun*, Sihang Zeng*, Chenxi Li*, Neil F Abernethy, Barbara D Lam, Fei Xia, Meliha Yetisgen. UW-BioNLP at ChemoTimelines 2025: thinking, fine-tuning, and dictionary-enhanced LLM systems for chemotherapy timeline extraction. Clinical NLP Workshop
Tianmai M Zhang, Neil F Abernethy. Reviewing scientific papers for critical problems with reasoning LLMs: baseline approaches and automatic evaluation. arXiv:2505.23824. AI for Science Workshop at NeurIPS

September 22, 2025 – September 26, 2025

UPCOMING LECTURES AND SEMINARS
BIME 590
Presenter: Gang Luo, PhD
Thursday, October 2^nd – 11-11:50 am
850 Republican Street, Building C, Orin Smith Auditorium (Not C123)
Zoom Information: https://washington.zoom.us/my/bime590
Speaker will present In-Person

Title: Identifying Patients Who are Likely to Receive Most of Their Care from a Specific Healthcare System

Abstract:
Background: In the United States, health care is fragmented in numerous distinct healthcare systems including private, public, and federal organizations like private physician groups and academic medical centers. Many patients have their complete medical data scattered across several healthcare systems, with no particular system having complete data on any of them. Several major data analysis tasks like predictive modeling using historical data are considered impractical on incomplete data.
Objective: Our goal is to find a way to enable these analysis tasks for a healthcare system with incomplete data on many of its patients.
Methods: This study presents, to the best of our knowledge, the first method to use a geographic constraint to identify a reasonably large subset of patients who tend to receive most of their care from a given healthcare system. A data analysis task needing relatively complete data can be conducted on this subset of patients. We demonstrated our method using data from University of Washington Medicine (UWM) and PreManage data covering use of all hospitals in Washington state. We compared ten candidate constraints to optimize the solution.
Results: For the UWM, the best constraint is that the patient has a UWM primary care physician and lives within five miles of at least one UWM hospital. About 16.01% of UWM patients satisfied this constraint. Around 69.38% of their inpatient stays and emergency department visits occurred within the UWM in the following six months, more than double the corresponding percentage for all UWM patients.
Conclusions: Our method can identify a reasonably large subset of patients who tend to receive most of their care from the UWM. This enables several major analysis tasks on incomplete medical data that were previously deemed infeasible.

Presenter Bio: Gang Luo obtained his Ph.D. degree in Computer Science minor in Mathematics at the University of Wisconsin-Madison in 2004. Between 2004 and 2012, he was a Research Staff Member at the IBM T.J. Watson research center. Between 2012 and 2016, he was a faculty member in the Department of Biomedical Informatics at the University of Utah. Gang is currently a Professor in the Department of Biomedical Informatics and Medical Education of the School of Medicine at the University of Washington. His research interests include health/clinical informatics (software system design/development and data analytics), machine learning, database systems, information retrieval, natural language processing, big data, and data mining with a focus on health applications. He invented the first method to automatically provide rule-based explanations for any machine learning model’s predictions with no accuracy loss, the first method to efficiently automate machine learning model selection, the questionnaire-guided intelligent medical search engine iMed, intelligent personal health record, and SQL, machine learning, and compiler progress indicators.

September 15, 2025 – September 19, 2025

UPCOMING LECTURES AND SEMINARS
BIME 590
Presenter: Peter Tarczy-Hornoch, MD, FACMI
Thursday, September 25^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: 2025 BIME Vision, History, Strategic Plan and Praxis

Abstract: The presentation will provide an overview of the Department of Biomedical Informatics and Medical Education through the lens of the current strategic plan. The overview will include vision, history and evolution looking at the synergy between research, education and practice (praxis) as well as the synergy between practice, applied research and foundational research. Within each area (research, praxis, education) current activities and future plans will be reviewed.

Presenter Bio: Peter Tarczy-Hornoch has over 40 years of experience in computer science, over 35 years in biomedical informatics and over 20 years in clinical medicine (pediatrics and neonatology). He has been at the University of Washington since 1992, serving as Head of the Division of Biomedical and Health Informatics since 2001 and serving as Chair of the Department of Biomedical Informatics and Medical Education since 2011. He has served in a variety of operational leadership roles in UW Medicine IT Services since 1992 in the analytics, research, and clinical computing domains, currently (since January 2022) serving as UW Medicine Chief Data Officer. He has played a leadership role in the creation and evolution of the BIME educational programs (undergraduate (joint with iSchool), MS/PhD, postdoctoral, applied clinical informatics MS (CIPCT) joint with Nursing, Clinical Informatics Fellowship joint with Family Medicine). He has led a number of key initiatives in informatics practice (praxis) including the areas of telemedicine, digital library, electronic medical records, data warehousing, analytics, clinical research informatics). His unifying theme of research over the last two decades has been data integration of electronic biomedical data (clinical, genomic and other including data) both for a) knowledge discovery and b) in order to integrate this knowledge with clinical data at the point of care for decision support. His current research focuses on a) secondary use of electronic medical record (EMR) for translational research including outcomes research, learning healthcare systems, patient accrual and biospecimen acquisition based on complex phenotypic eligibility criteria, b) the use of EMR systems for cross institutional comparative effectiveness research, and c) integration of genomic data into the EMR for clinical decision support.

PAPERS, PUBLICATIONS & PRESENTATIONS

A Novel Communication Rating Scale to Mitigate the Effect of Implicit Bias, Jennifer Tjia MD, MSCE; Chengwu Yang, MD, MS, PhD; Julie Flahive, MS; Kelly Harrison, MD; Geraldine Puerto, MPH; Vennesa Duodu, MPH; Lisa A. Cooper, MD, MPH; Olga Valdman, MD; Janice Sabin, PhD, MSW,

JAMA Netw Open. 2025; 8(9):e2532319.10.1001/jamanetworkopen.2025.32319

September 8, 2025 – September 12, 2025

UPCOMING LECTURES AND SEMINARS
BIME 590 – See you in autumn quarter – 9/25/2025!

ANNOUNCEMENTS
Please join us in congratulating Dr. Raghav Madan who successfully passed his PhD Defense!
Title: NPP: tissue-anchored, GIS-inspired spatial modeling of tau neuropathology

Abstract: I developed NeuroPathPredict (NPP), a pipeline that fuses quantitative histopathology (Alzheimer’s disease) with neuroimaging to generate voxel-based tau distribution maps in unsampled regions . Slide-level AT8 measures are mapped to MNI 2009b via ex vivo MRI (QNPtoVox); voxels are enriched with an atlas “brain-GIS” of tracts, functional networks, vascular territories, cortical geometry, and distance transforms (I-BIS). I fit universal kriging with external drift, selecting a tractable set of primary covariates via Elastic Net, estimating variograms, and validating with spatial block cross-validation. Applied to Adult Changes in Thought (ACT) study data, NPP yields anatomically interpretable effects, competitive prediction versus non-spatial models, and reproducible, donor-comparable pathology distribution fields suited for cross-modal evaluation. Looking ahead, NPP can serve as a reference layer for cross-modal validation and hypothesis testing, and be extended to other proteinopathies and cohorts for scalable, tissue-anchored brain mapping.

_______________________

Please join us in congratulating Faisal Yaseen who successfully passed his General Exam!
Title: Treatment response prediction in advanced non-small cell lung cancer: Biomarkers, multiscale, multimodal, and uncertainty quantification

Abstract: With metastatic disease, treatment response can be highly variable, leading to challenging clinical decision making. . Traditional assessment methods, such as computed tomography (CT) lesion size, often fail to capture the spatial heterogeneity of tumor response. Fluorodeoxyglucose (FDG) PET/CT imaging offers the ability to detect early metabolic changes across patient, lesion, and voxel levels. To improve early treatment response prediction and clinical utility, we propose three synergetic yet independent aims: (1) identify FDG PET imaging and blood-based biomarkers to discriminate patient-level treatment response and survival outcomes, (2) develop a multiscale regression framework to predict voxel-level tumor response with uncertainty quantification using conformal prediction, and (3) integrate imaging, blood biomarkers, and clinical data into an multimodal AI framework for early response. Collectively, this project will deliver a clinically relevant decision support prototype to guide personalized therapy in mNSCLC.

PAPERS, PUBLICATIONS & PRESENTATIONS

Feng Chen, Dror Ben-Zeev, Gillian Sparks, Arya Kadakia, Trevor Cohen. Detecting PTSD in Clinical Interviews: A Comparative Analysis of NLP Methods and Large Language Models. Paper accepted by Pacific Symposium on Biocomputing 2026.

September 1, 2025 – September 5, 2025

UPCOMING LECTURES AND SEMINARS
BIME 590 – See you in autumn quarter – 9/25/2025!

PAPERS, PUBLICATIONS & PRESENTATIONS

Mastrianni, A., Twede, H., Sarcevic, A., Wander, J., Austin-Tse, C., Saponas, S., Rehm, H., Conard, A.M. and Hall, A.K., 2025. AI-Enhanced Sensemaking: Exploring the Design of a Generative AI-Based Assistant to Support Genetic Professionals. ACM Transactions on Interactive Intelligent Systems.

UPCOMING EXAMS

Final Dissertation Defense
Title: NPP: tissue-anchored, GIS-inspired spatial modeling of tau neuropathology
Student: Raghav Madan
Date/Time: Wednesday, September 10^th, 2025, 12pm PT
In-person location: SLU, BLDG C, C122
Zoom: https://washington.zoom.us/my/jhgennari?pwd=TUx0clkwKzdnS1ZQV1dXRnZqMWMzZz09

General Exam
Title: Treatment response prediction in advanced non-small cell lung cancer: Biomarkers, multiscale, multimodal, and uncertainty quantification
Student: Faisal Yaseen
Date/Time: Thursday September 11^th, 2025, 9am-11am PT
In-person location: HSEB 421
Zoom: https://washington.zoom.us/my/jhgennari?pwd=TUx0clkwKzdnS1ZQV1dXRnZqMWMzZz09

August 25, 2025 – August 29, 2025

UPCOMING LECTURES AND SEMINARS
BIME 590 – See you in autumn quarter – 9/25/2025!

PAPERS, PUBLICATIONS & PRESENTATIONS

Placental network differences among obstetric syndromes identified with an integrated multiomics approach. Piekos SN, Barak O, Baumgartner A, Chu T, Parks WT, Hadlock J, Hood L, Price ND, Sadovsky Y. Commun Biol8, 1239 (2025). https://www.nature.com/articles/s42003-025-08631-6
Xu W, Pakhomov S, Heagerty P, Horvitz E, Bradley ER, Woolley J, Campbell A, Cohen A, Ben-Zeev D, Cohen T. Perplexity and proximity: Large language model perplexity complements semantic distance metrics for the detection of incoherent speech. Journal of Biomedical Informatics. 2025 Aug 21:104899. https://www.sciencedirect.com/science/article/pii/S1532046425001285
Wang, L. C., Pike, K. C., Conway, M., & Chen, A. T.(accepted). Identifying stigma phenotypes in social media narratives of substance use: An observational study. Journal of Medical Internet Research.
Chen, A. T.,Wang, L. C., Pike, K. C., Conway, M., & Glass, J. E. (accepted). Comparing the use experiences, contextual factors, and recovery strategies associated with different substances: An analysis of social media narratives. Substance Use & Misuse. DOI: 10.1080/10826084.2025.2540938
Liu, J., Bessler, S., Zhang, Y., Komi, M. M., & Chen, A. T.(2025). Changes of information needs and emotions during COVID-19: A longitudinal view. Library & Information Science Research, 47(3), 101367.
Chen, A. T.*, Dunn, L. H.*, Fan, W.,& Agrawal, N. (2025). Audience responses to online public shaming during COVID-19: A mixed-methods study. Journal of Medical Internet Research, 27. DOI: 10.2196/67923.

UPCOMING EXAMS
Title: NPP: tissue-anchored, GIS-inspired spatial modeling of tau neuropathology
Student: Raghav Madan
Date/Time: Wednesday, September 10^th, 2025, 12pm PT
In-person location: SLU, BLDG C, C122
Zoom: https://washington.zoom.us/my/jhgennari?pwd=TUx0clkwKzdnS1ZQV1dXRnZqMWMzZz09

August 18, 2025 – August 22, 2025

UPCOMING LECTURES AND SEMINARS
BIME 590 – See you in autumn quarter – 9/25/2025!

ANNOUNCEMENTS
Please join us in congratulating Serena Jinchen Xie who successfully passed her PhD Defense!
Title: Cultural Adaptation and Evaluation of AI Mental Health Support Conversational Agents

Abstract: Mental health disparities disproportionately affect underserved caregiver populations, in part due to limited availability of culturally responsive interventions. While large language model (LLM)-driven conversational agents offer promise for scalable mental health support, their outputs often suffer from cultural misalignment, limiting engagement among diverse populations. This dissertation develops and evaluates a generalizable, low-resource workflow for dynamic cultural adaptation of LLM-based mental health agents. Grounded in formative qualitative work with Chinese American and Latino American family caregivers and community stakeholders, I developed a cultural context database capturing salient caregiving challenges. Two adaptation strategies were implemented: (1) prompt-based “cultural prompting” and (2) a retrieval-augmented generation (RAG) workflow that dynamically integrates relevant cultural context during real-time interactions. Controlled evaluations showed the RAG-based approach outperformed both prompt-based and non-adapted agents on cultural responsiveness, perceived empathy, and therapeutic alliance. A randomized pilot study with Chinese American caregivers demonstrated improvements in short-term emotional well-being. This work advances scalable, culturally responsive AI agents for equitable mental health support and offers a generalizable workflow for adaptation across diverse populations and care contexts.

August 4, 2025 – August 8, 2025

UPCOMING LECTURES AND SEMINARS
BIME 590 – See you in autumn quarter – 9/25/2025!

UPCOMING EXAMS
Title: Cultural Adaptation and Evaluation of LLM-Driven Mental Health Conversational Agents
Student: Serena Xie
Date/Time: Monday, August 18^th, 2025, 11am PT
In-person location: Health Science Building (HST) T473
Zoom: https://washington.zoom.us/my/cohenta

July 28, 2025 – August 1, 2025

UPCOMING LECTURES AND SEMINARS
BIME 590 – See you in autumn quarter – 9/25/2025!

ANNOUNCEMENTS
Internship – Real World Evidence (RWE) Team
Location: Remote or Hybrid (based on applicant location)
Duration: 3–6 months
Compensation: Paid Internship
Start Date: Flexible (Summer/Fall 2025 preferred)

Position Summary:

The Natera RWE team is seeking a highly motivated and analytical intern to contribute to real-world data initiatives involving large-scale clinical and genomic datasets. The ideal candidate will have hands-on experience with clinical coding systems (ICD-10, CPT, NDC), a keen interest in advanced analytics using AI/LLMs, and exposure to digital pathology and multimodal biomedical data.

This internship provides a unique opportunity to work at the intersection of clinical informatics, genomics, and machine learning, supporting projects that drive impactful healthcare insights.

For Full Details and How to Apply contact rigreen@natera.com

PAPERS, PUBLICATIONS & PRESENTATIONS

S. Zeng, L. Liu, J. Wen, M. Yetisgen, R. Etzioni, and G. Luo. TrajSurv: Learning Continuous Latent Trajectories from Electronic Health Records for Trustworthy Survival Prediction. Proc. 2025 Machine Learning for Healthcare Conference, Rochester, MN, Aug. 2025.

News and Events

Chair’s Message

Biomedical Informatics and Medical Education Newsletter

October 27, 2025 – October 31, 2025

October 20, 2025 – October 24, 2025

September 29, 2025 – October 3, 2025

September 22, 2025 – September 26, 2025

September 15, 2025 – September 19, 2025

September 8, 2025 – September 12, 2025

September 1, 2025 – September 5, 2025

August 25, 2025 – August 29, 2025

August 18, 2025 – August 22, 2025

August 4, 2025 – August 8, 2025

July 28, 2025 – August 1, 2025

Upcoming Events