Publications and outputs
Each record is structured for scholarly review: status, venue, abstract, methods, contribution, role, impact, and evidence file.
Maternal Safety & Rescue Systems · Contributed Conference Paper
This study operationalized a scalable Obstetric Rescue Clock in U.S. delivery hospitalizations using the 2021 HCUP National Inpatient Sample. Among high-risk delivery hospitalizations receiving qualifying rescue procedures, early rescue on hospital day 0–1 was compared with delayed rescue on day 2 or later. The targ…
Learning Health Systems & Clinical AI · Peer-reviewed full conference paper
This study evaluated whether concurrent sepsis identifies a high-risk subgroup among gynecologic malignancy admissions and whether EHR-derived early warning can support learning-health-system rescue surveillance. Sepsis-positive admissions had substantially elevated mortality. A temporally validated early-warning mode…
Maternal AI & Decision Support · Peer-reviewed full conference paper
This work tested whether maternal deterioration prediction remains useful when restricted to signals plausibly available in real time and when evaluated under explicit alert-burden constraints. The model achieved a high-specificity, low-burden escalation signal but low sensitivity, showing that available early signals…
Perinatal AI Validation · Peer-reviewed full conference paper
This paper formally quantified evaluation leakage in electrohysterogram-based preterm birth prediction. Under leakage-proof patient-independent validation, apparent discrimination collapsed to near chance, indicating that many published performance claims are likely leakage artifacts. Class-conditional conformal selec…
Global Reproductive Health Equity · Accepted peer-reviewed conference abstract/paper
This multicountry DHS study developed the Digital Reproductive Access Cascade (DRA-C), a reproductive epidemiology instrument linking mobile ownership, smartphone/internet access, active digital engagement, and direct family-planning messaging to contraceptive outcomes. The analysis found plausible information-to-acti…
Global Maternal Health Systems · Peer-reviewed poster/paper
This multicountry DHS analysis created a high Antenatal Continuity Readiness Index using first-trimester initiation, four or more visits, skilled provider ANC, and facility-based ANC. Across 13 sub-Saharan African countries, higher readiness was associated with lower risk of non-facility delivery in most countries, de…
Maternal Morbidity & Equity · Peer-reviewed conference paper
This national survey-weighted analysis of delivery hospitalizations in HCUP NIS 2021 showed that refined nontransfusion severe maternal morbidity was strongly patterned by maternal age, race, payer, and neighborhood income. Weighted descriptive, relative-effect, and absolute-risk results converged, indicating a robust…
Trial Generalizability & Reproductive Justice · Roundtable abstract
This project built an auditable atlas converting oncology-trial eligibility text into reproductive restriction phenotypes and generalizability-loss metrics. Pregnancy and lactation exclusions were common and patterned by intervention class, era, and geography, enabling minimum-necessary restriction redesign and policy…
Women’s Health AI & Explainability · Refereed conference presentation
This analysis used national reproductive, chronic-disease, socioeconomic, and mental-health data to derive latent phenotypes and predict early multimorbidity among U.S. women. SHAP explanations and LLM-generated narratives translated risk predictions into patient-facing language without inventing conditions, demonstra…
Life-course Reproductive Epidemiology · Refereed conference presentation
This nationally representative analysis examined whether adverse reproductive histories mark early multisystem cardiometabolic-renal multimorbidity and depression among young U.S. women. The work positions reproductive history as a life-course signal for preventive primary-care and maternal medicine strategies.
Measurement Error & Maternal Morbidity · Conference submission / proof summary
This project developed a Hospital-Calibrated Latent Evidence Score to reduce obesity misclassification in administrative obstetric data. The model integrates ICD-10 obesity code families, severity-weighted evidence, and hospital-level shrinkage to identify latent adiposity phenotypes and quantify bias in severe matern…
Global Maternal Mortality Policy · Manuscript / preprint with DOI
This longitudinal ecological panel analysis estimated within-country associations between institutional delivery coverage and maternal mortality over time. The primary fixed-effects model found that higher institutional delivery coverage was associated with lower subsequent maternal mortality, but sensitivity analyses…
Intrapartum Care & Newborn Survival · Manuscript / preprint; accepted note in file
This study examined whether facility birth and skilled attendance translate into effective intrapartum capability in Nigeria. Despite high skilled attendance among facility births, early neonatal survival varied substantially by wealth and region, suggesting that remaining deaths reflect timely escalation, functional …
Causal Perinatal Epidemiology · Accepted manuscript / preprint noted in file
This study estimated the marginal effect of unintended pregnancy on preterm birth and tested whether pregnancy intention improves risk stratification. Survey-weighted TMLE estimated a modest positive effect but with overlap limitations and sensitivity to residual confounding. Super Learner improved ranking and net ben…
Maternal-Fetal Imaging & AI · Electronic poster abstract
This specialist maternal-fetal medicine output examined multimodal AI-augmented radiomics and spectral Doppler ultrasound for early pre-eclampsia prediction, connecting imaging biomarkers to interpretable obstetric risk stratification.
Statistical Education & Reproducible Research · Curriculum and in-class teaching workshop paper
This teaching paper provides an end-to-end reproducible workflow for longitudinal linear mixed models, moving from data reshaping and visualization to model selection, diagnostics, and interpretable contrasts. It is designed for biostatistics, epidemiology, and applied statistics teaching.