Drive-Time Access Modeling (Joinery)
OpenRouteService · GeoPandas · Streamlit · CMS ASPIRE · NC InCK
Drive-time access model routing from Census block group centroids to the nearest pediatric provider across 8 specialties and 18 Western NC counties. Every county and specialty pair is classified into four tiers, from Accessible at 30 minutes or less to Effectively Unreachable past 90 minutes, alongside the share of each county that falls out of reach. Built into a nine-view care coordination platform covering risk stratification, quality measures, referral and prior authorization tracking, and social needs workflows, designed against CMS ASPIRE and NC InCK specifications.
Geospatial layer uses live NPPES and Census data. Patient-level views run on simulated records pending production claims integration. Deployed under the working name Time to Care.
Geospatial Epidemiology (Parkinson's Risk Factors)
CDC PLACES · EPA AQI · Census ACS · IHME GBD · USGS Pesticide · USDA NASS · GeoPandas · Statsmodels
County-level analysis of Parkinson's across 3,143 U.S. counties, regressing nine modifiable risk factors and an agricultural pesticide layer against state PD prevalence (GBD 2023). Modifiable factors explained only ~11% of the geographic variation. A promising per-capita pesticide signal reversed under a per-acre-of-cropland denominator, a self-correction that prevented a false environmental conclusion.
Healthcare Access Analytics (NC Pediatric Specialists)
NPPES · Census ACS · GeoPandas · Folium · GIS
Built a reproducible geospatial pipeline mapping pediatric specialist supply vs. child population across North Carolina's 100 counties. Identified 67% of counties as severe access deserts and 8 counties with zero pediatric specialists.
Business Intelligence (NC Pediatric Access Dashboard)
Power BI · DAX · Star Schema · Azure Maps · Healthcare BI
Four-page interactive Power BI report rebuilding the NC pediatric access analysis as a stakeholder-facing BI deliverable. Five-table star schema, custom DAX measures with region-aware filtering, Azure Maps choropleths, and a fully documented methodology page covering 6,821 providers across 100 NC counties.
Healthcare Access Analytics (NC Prenatal Care)
HRSA AHRF · Census ACS · HPSA · CDC WONDER · GeoPandas · Statsmodels
Five-source geospatial analysis of prenatal care access across NC's 100 counties; identified 27 counties with zero OB/GYN providers, 96/100 counties with active maternity shortage designations, and a federal surveillance gap that renders all 23 Western NC counties invisible in CDC natality data.
Civic Analytics (MS Legislative Scorecard)
Streamlit · LegiScan API · SQLite · Weighted Scoring · Live App
Production-deployed accountability platform grading Mississippi state legislators on first responder policy votes. Four-category weighted scoring formula, sponsorship and absence tracking, and live LegiScan API data on active bills in the current session. Built in collaboration with TAC Solutions.
Process Improvement (DMAIC)
MBC-638 · Statistics · Six Sigma
Six Sigma DMAIC project applying regression (R²=0.77), multilinear modeling (R²=0.95), and hypothesis testing to personal fitness data. Identified early morning combo workouts as optimal process; validated +200 cal/day improvement via two-sample t-test (p=0.0006).
Membership Database (SQL Server)
IST-659 · ER/3NF · Triggers · Governance
Normalized 3NF schema (8 entities) with T-SQL triggers and views for LARC (150+ members); migrated from Access → SQL Server for scalability and integrity.
Machine Learning (Preventive Pulse)
CDC BRFSS · scikit-learn · SMOTE · XGBoost · Matplotlib
Heart-disease prediction on 319K+ CDC survey records; logistic regression with SMOTE achieved 83.5% ROC-AUC and 75.7% F1-score across 6 classifiers. Identified age, stroke history, and self-reported health as the dominant risk factors. Applied the trained pipeline to 445K unseen 2022 records.
Deep Learning (Melanoma Detection)
HAM10000 · TensorFlow/Keras · EfficientNet-B0 · Grad-CAM · Transfer Learning
Image classifier detecting melanoma from 10,015 dermoscopy images using two-stage transfer learning with EfficientNet-B0. Achieved Macro ROC-AUC of 0.925 across 7 lesion types. Grad-CAM heatmaps visualize model focus regions for clinical interpretability.
