Comprehensive collection of 8 clinical data science and health analytics projects focusing on disease prediction, risk stratification, and treatment pattern analysis using advanced machine learning algorithms and statistical modeling. Portfolio: https://nana-safo-duker.github.io/

Interactive comorbidity analysis on the web

Explainable ML model for 1-year mortality prediction in hemodialysis patients, focused on vascular access, inflammation, and clinical risk stratification.

a project for peer assignment in Predictive Modelling course of Clinical Data Science Specialization on Coursera.

Real-world hemodialysis data analysis focused on intradialytic hypotension, clinical risk patterns, and patient-level hemodynamic phenotyping.

Python R Hybrid clinical trial machine learning project for clinical development using CDISC-aligned data, SDTM/ADaM workflows, safety analytics, endpoint modeling, and survival analysis

ACO revenue leakage quantification via Medicare specialty referral network analysis (NetworkX, Python)

AI/ML-driven analysis of clinical trial oversight, reporting timeliness, and quality using AACT and ClinicalTrials.gov data.

A high-fidelity triage system using Isotonic-calibrated ensembles and human-in-the-loop deferral.

Unsupervised learning for patient phenotype discovery in critical care hypotension cohort. Applied clustering algorithms on 5,000+ ICU patients to identify distinct subgroups with significantly different mortality rates and length of stay outcomes.

Medical knowledge graph from 47K PubMed articles with RAG-powered clinical Q&A (Neo4j, Python)

Clinical data science pipeline predicting 30-day hospital readmission using MIMIC-III ICU data — featuring ICD-9 comorbidity engineering, lab biomarker extraction, SHAP explainability, fairness audit, and Streamlit dashboard

Portfolio repository for the CariSurg Healthcare AI Training Programme, focusing on emergency department triage, clinical data analysis and AI-assisted decision support.

Clinical data science investigation of environmental covariates in ABA outcomes — tiered prediction model, OSF-pre-registered, with interactive prototypes.

EHR-based observational analysis of adult non-ICU hospital admissions using MIMIC-IV. Evaluates early RAAS inhibitor exposure and in-hospital mortality with multivariable logistic regression, absolute risk estimation, and SAS-Python validation.

EHR-based observational survival analysis of ICU patients with COPD using MIMIC-IV. Evaluates the association between pre-ICU RAAS inhibitor exposure and in-hospital mortality using time-to-event methods.