EHR encounters screened
HEALTHCARE DATA SCIENTIST
SAMANTHA
WENJIN ZHOU
CLINICAL RESEARCHER / EHR ANALYTICS
I translate clinician-led questions into reproducible EHR and clinical research workflows.
00
IMPACT AT A GLANCE
Patients in analytic cohort
Reduction in manual analysis time
Reproducible evidence, built at the intersection of data systems, study design, and scientific communication.
01 / SELECTED WORK
EVIDENCE
CASE 01
Epic Cosmos / Gynecologic Surgery Outcomes
EHR COHORT DESIGN + OUTCOMES
From a clinician-led question to an auditable cohort and decision-ready analysis.
SCREENED1,259,836MIH encounters
ANALYTIC COHORT236,799encounters
236,242patients
19,776providers
RESEARCH QUESTION
How do surgeon specialty and annual volume relate to admission and postoperative complications after minimally invasive hysterectomy?
MY ROLE
Data scientist and Co-Investigator collaborator: cohort strategy, reproducible SQL/R pipeline, modeling, feasibility tables, and clinician translation.
METHODS
Diagnosis and procedure definitions, denominator audits, age/BMI validation, 730-day Charlson lookback, missing-data checks, multivariable logistic regression, and sensitivity analyses.
RESULT
A 236K-patient analytic cohort and reproducible evidence workflow that shortened feasibility timelines by approximately 50%.
02 / SELECTED WORK
AUTOMATION
CASE 02
DeepLabCut Behavioral Analysis
COMPUTER VISION + PHENOTYPING
BEFORE8 HOURS
AFTER10 MINUTES
~98% less manual analysis time
01VIDEOstructured input
02POSEkeypoint estimation
03METRICSobjective behaviors
04ANALYSISreproducible output
PROBLEM
Manual behavioral scoring limited throughput and made consistent phenotyping difficult to scale.
MY ROLE
Led an NIH R01 Aim 3a dose-finding study and built the Python/DeepLabCut pipeline from pose-estimation outputs to analysis-ready data.
APPROACH
Defined objective behavioral metrics, automated scoring, structured outputs, and documented an end-to-end workflow for repeatable use.
RESULT
Reduced analysis from about eight hours to ten minutes and created onboarding manuals that shortened independent training to approximately one month.
03 / SELECTED WORK
SYSTEMS
CASE 03
REDCap / Clinical Research Operations
DATA SYSTEMS + IRB OPERATIONS
Designing the infrastructure that keeps protocol, specimens, and analysis-ready data aligned.
PROTOCOLSURGICAL FINDINGSPATIENT-REPORTED OUTCOMESBIOSPECIMEN METADATA
REDCAP LOGICELIGIBILITYSPECIMEN QCIRB ALIGNMENT
STRUCTUREDAUDITABLEANALYSIS-READY
PROBLEM
Clinical research operations needed one consistent system for eligibility, study events, patient-reported outcomes, and specimen-linked quality control.
MY ROLE
Functional Clinical Research Coordinator and data-system architect across REDCap configuration, IRB submissions, and protocol-data alignment.
APPROACH
Automated eligibility logic, specimen-linked QC workflows, harmonized clinical fields, and structured, IRB-compliant datasets.
RESULT
A research workflow designed to catch quality issues early and keep operational decisions traceable from protocol to dataset.
Illustrative workflow only. No participant records or internal screenshots are shown.
04 / RESEARCH + SERVICE
RIGOR IN THE
WORK. CLARITY
IN THE TELLING.
HONORS THESIS / 2025
Secularization in Pastoral Elegy: From Sacred Nature to Liberal Selves
An 80-page interdisciplinary thesis integrating literary analysis, sociology, and philosophy. Presented and defended before faculty committees; awarded Highest Honors.
VIEW EMORY REPOSITORYPEER REVIEW / 2026
PCORI Proposal Review
Reviewed patient-centered comparative effectiveness proposals involving AI/ML and large-scale EHR data networks, with attention to scientific rigor, methodological validity, and translational impact.
GRANT REVIEW / 2026
EU-PDA Travel Awards
Reviewed competitive early-career research applications for scientific merit, innovation, and relevance.
EDUCATION / EMORY UNIVERSITY
Biology + English
Bachelor's degree, double major. Highest Honors and Dean's List.
05 / ABOUT
Biology taught me how to ask testable questions. English taught me how to make the answers matter.
My path into healthcare data science began across two disciplines: one grounded in experimental systems, the other in close reading and precise communication. Today, I bring both to complex clinical questions.
I am interested in reproducible EHR evidence, outcomes research, clinical data infrastructure, patient-centered analytics, and collaborations where rigorous methods can improve real decisions.
06 / CONTACT
LET'S BUILD
BETTER EVIDENCE.
Open to healthcare data science, clinical research, and interdisciplinary collaboration.