Upload your resume. PhdFit's six research agents compare your background with faculty profiles, recent publications, lab focus, and outreach opportunities, then rank professors with evidence you can review.
Resume-aware matchingEvidence from recent researchOutreach drafts included
Advisor Match Brief Resume uploaded
Resume signals5 found
NLPHCIFairnessGraph MLPyTorch
Six-agent analysisRunning
Axel · Candidate AnalystResume parsed
Nova · Professor ResearcherRecent papers
Echo · Match ExplainerEvidence linked
Atlas · Strategy AdvisorFit ranked
Lyra · Outreach WriterDraft queued
Juno · Meeting Prep CompilerNext step ready
SC
Dr. Sarah Chen
Stanford · NLP & interpretability
91
fit
Fairness + NLP projects overlap with her recent work.
Strong methods fit: model evaluation and interpretability.
Clear outreach angle for current trustworthy AI research.
Outreach angle
Ask how her lab is extending interpretability methods into fairness audits for real-world AI systems.
Karen Chan
· OceanographyVerified
University of Washington · Program on the Environment
Karen Chan is an Assistant Professor at the College of the Environment at the University of Washington. Her work focuses on ecology and marine science, contributing to the understanding of ocean health and marine ecosystems. She is involved in research related to oceanography and marine biology, working within the Oceanography Program at the university. Her academic and research activities are centered on advancing knowledge in marine sciences and addressing environmental challenges related to oceanic and marine environments.
Accurate classifiers that use novel biomarkers and readily available predictors significantly enhance decision-making in various clinical scenarios, such as assessing the need for biopsies in cancer diagnosis. When classification performance is limited, a decision framework can be applied to rule in or rule out invasive diagnostic procedures while incorporating a neutral zone for indeterminate classifications. Building on this framework, we propose a new family of two-step classifiers that selectively use costly biomarker testing for a targeted subset of individuals undergoing multiple evaluations. The optimal solution expands upon the Neyman-Pearson Lemma, highlighting a vital trade-off between the costs of expensive biomarker measurements and improving classification performance while minimizing uncertainty in the decision process. We demonstrate the practical utility of our approach through a biomarker study focused on prostate cancer diagnosis.
Data harmonization is the process of developing an equivalence between two measurements of a common domain. Our problem is motivated by dementia research in which multiple neuropsychological tests have been used in practice to measure the same underlying cognitive ability, such as memory or attention. We connect this statistical problem to mixing distribution estimation common in empirical Bayes approaches. We introduce and study a nonparametric latent trait model, develop a method that enforces the uniqueness of the regularized maximum likelihood estimator, show how a nonparametric EM algorithm will converge weakly to its maximizer, and illustrate its superior computational efficiency to off-the-shelf solvers. Furthermore, we develop methods for model selection and assessing the goodness-of-fit for the measurement model, an area neglected in most mixing distribution estimation problems. We develop methods for score conversion with uncertainty quantification in order to draw inferences on a whole population with multiple score scales. We apply our method to the National Alzheimer’s Coordination Center Uniform Dataset and show that we can use our method to convert between score measurements and account for the measurement error. We show that this method outperforms standard techniques commonly used in dementia research.
Understanding how biomarkers change in relation to disease pathogenesis is a key area in biomedical research. We propose a semiparametric joint model to analyze the temporal evolution of biomarkers prior to the onset of disease. The model allows for a flexible biomarker trajectory that depends on two time scales: a natural time scale such as age and time to disease onset. In practice, the natural time scale often differs from time-on-study, leading to analytical challenges such as left-truncation bias. We introduce a profile kernel estimating equation approach to estimate regression coefficients and unspecified baseline mean trajectory functions. We establish the large-sample properties of the proposed estimators and conduct simulation studies to evaluate their finite-sample performance. Our method is applied to investigate brain biomarker trajectories before the onset of preclinical Alzheimer's disease. We observed a decline in cortical thickness prior to disease onset across brain regions, with APOE4 carriers showing lower levels compared to non-carriers.
INTRODUCTION: Alzheimer's Disease Research Centers have used the Uniform Data Set Version 3 (UDS3) neuropsychological battery since 2015, but whether it exhibits differential sensitivity to change across race is unknown. We examined whether the UDS3 cognitive battery was comparably sensitive to longitudinal change between African American and White participants. METHODS: Data were obtained from the National Alzheimer's Coordinating Center (NACC). Linear mixed-effects models examined racial differences in baseline and longitudinal change in standardized test scores, controlling for age, sex, education, recruitment source, health factors, family history of dementia, and diagnostic groups defined by baseline and longitudinal changes in Clinical Dementia Rating (CDR) scores. RESULTS: Compared to White participants, African American participants had significantly lower baseline Z-scores on all tests (difference: -0.097 to -0.592). Nevertheless, differences in longitudinal decline were non-significant (annualized difference: -0.018 to 0.031). DISCUSSION: Despite baseline score differences, longitudinal change relative to clinical ratings appears comparable across racial groups. HIGHLIGHTS: Version 3 of the NACC Uniform Data Set neuropsychological battery (UDS3) has been implemented since 2015 and administered to over 14,000 participants. We observed small differences in longitudinal decline across all test scores when comparing White and African American participants. The findings support the continued use of the UDS3 battery in ADRCs and its potential applicability in other studies of cognitive aging and decline.
Abstract Analytical methods to incorporate potential concurrent analgesic use into primary statistical summaries are underutilized in pain randomized controlled trials (RCTs). Without valid inclusion of analgesic use, the primary analyses of pain may produce diminished estimated treatment effects. We used contemporary causal inference methods that can account for concurrent treatment to reanalyze RCT data examining the effect of epidural steroid injection (ESI). Specifically, we define an “attributable to ESI estimand”, which is the controlled direct effect of ESI. We used a simple composite pain intensity outcome, the QPAC 1.5, and structural mean models (SMM) to estimate the target estimand. Compared to traditional methods such as strict intention to treat analysis (strict ITT), SMMs can account for analgesic use without assuming sequential ignorability. We estimated treatment effects of ESI on leg pain intensity (LPI) using the numeric rating scale (NRS) with strict ITT, 3 SMM estimating methods (estimating equations [EE], g-estimation, and generalized method of moments [GMM]), and the QPAC 1.5 . The treatment effect of ESI on LPI using strict ITT was -0.751 NRS points (95% confidence interval [CI]: [-1.287, -0.214]). Estimates for the attributable to ESI estimand were -0.864 (95% CI: [-3.207, 1.478]) for EE, -0.935 (95% CI: [-1.779, 0.090]) for g-estimation, -0.653 (95% CI: [-1.218, -0.089]]) for GMM, and -0.930 (95% CI: [-1.508, - 0.352]) for the QPAC1.5. We illustrate how contemporary causal inference methods and alternative estimands can be used to account for concurrent analgesic use in pain RCTs, in a manner that may result in larger treatment effects.
OBJECTIVE: We aimed to evaluate the impact of an electronic health record (EHR)-based default "nudge" intervention on opioid prescribing after common surgical operations. BACKGROUND: Given ongoing national challenges in opioid use and opioid-related adverse events, there is a need to optimize opioid prescribing after surgery. Behavioral nudges built into the EHR may be effective and useful, but they have not been widely tested among surgeons. METHODS: This is a randomized clinical trial at a large academic medical center involving adult surgical patients. System-wide, 118 surgical clinicians were randomized to control or intervention versions of an electronic discharge order set; the intervention arm had suggested pre-populated opioid and adjunct prescriptions. The outcome was morphine milligram equivalents (MME) prescribed at discharge, analyzed at the surgical encounter level. RESULTS: 663 surgical encounters (377 intervention and 286 control) were analyzed. 57% of the patients were female, and the median age was 49 years. The most common operations were laparoscopic cholecystectomy (39%), laparoscopic appendectomy (25%), and laparoscopic/robotic colectomy (12%). The median opioids prescribed at discharge was 75 MME (~ 10 oxycodone-5 mg tablets). Overall, in the intention-to-treat analysis, there were similar MME prescribed in the control and intervention groups (adjusted difference: 2.4 MME, 95% CI: -14.7 to 19.4, P=0.79). Only 21% in the intervention group received prescriptions from pre-populated defaults, while 79% received free form prescriptions. In as-treated analysis, the pre-populated prescriptions were substantially smaller than those written free form (adjusted difference: -22.5 MME, 95% CI: -34.7 to -10.2). CONCLUSIONS: The offering of pre-populated opioid prescriptions to surgeons did not reduce postoperative opioid prescribing overall. However, the subset of surgeons who accepted the defaults prescribed much less opioids. Default nudge interventions may be useful in optimizing opioid prescribing, but stronger versions of defaults and co-interventions are likely needed.
