Cardiovascular Predictive Modeling Specialist

My immediate goal is to develop robust predictive models that can meaningfully inform cardiovascular research and clinical decision-making. I have aggregated multiple datasets—ranging from structured EHR extracts to imaging-derived variables and device telemetry—and now need a data scientist who can turn these raw inputs into clinically relevant risk scores and outcome forecasts.

Scope of work
• Clean, integrate, and document the disparate datasets I will share (CSV, SQL dump, and optional imaging features in HDF5).
• Engineer features, test several algorithms (e.g., gradient boosting, random forests, neural nets), and iterate toward an interpretable solution.
• Provide model performance metrics—AUROC, calibration plots, and decision-curve analysis—so clinicians can easily gauge utility.
• Package the final model as a reproducible Python notebook / script with clear inline comments, environment file, and concise README.

Acceptance criteria
1. AUROC ≥0.80 on the held-out test set.
2. Code executes end-to-end with `conda env create -f environment.yml`.
3. All steps, from preprocessing through validation, are traceable in a single notebook or Markdown report.

When you respond, attach a detailed project proposal outlining: your planned workflow, preferred libraries (scikit-learn, XGBoost, PyTorch, etc.), anticipated timeline with milestones, and any relevant prior cardiovascular or biomedical work you can publicly reference.

I am fully open to alternative techniques or supplementary data sources you may recommend, provided they enhance predictive power and remain explainable to a clinical audience.