Decision Scientist III - Quality and Patient Safety initiative

About the role

The Decision Scientist III supports the Quality and Patient Safety initiative (QPSi) within the UF College of Medicine by applying decision-science approaches to improve patient safety, care quality, and healthcare operations across UF Health and partner institutions. The role focuses on helping teams evaluate clinical and operational decisions by modeling options, constraints, tradeoffs, and expected outcomes.

Responsibilities

• Develop and evaluate multi-objective optimization approaches (e.g., weighted objectives, constrained formulations, and Pareto-frontier analyses) that balance patient outcomes, quality/value of care metrics, clinician workload, and workflow feasibility (e.g. alert fatigue/sensitivity).
• Use offline reinforcement learning or related policy-evaluation methods when historical data are strong enough to compare decision strategies.
• Build models that represent how patients, care pathways, resources, or bottlenecks change over time.
• Test how recommendations change under different assumptions, data limitations, and implementation constraints.
• Produce analyses, documentation, and code that can be reviewed, reused, and extended by the broader QPSi team.
• Identify the data needed for decision modeling, including state definitions, actions, outcomes, constraints, and exclusion criteria.
• Work with data engineering, modeling, causal inference, and process-improvement staff to prepare data for optimization and policy evaluation.
• Convert predictions, risk scores, causal estimates, process measures, clinical rules, and operational data into usable model inputs.
• Evaluate whether available data are appropriate for policy comparison, especially when data are missing, actions are sparse, or timing is inconsistent.
• Prototype analyses using Python, SQL, and distributed computing resources when large healthcare datasets require them.
• Work with clinicians, operational leaders, and project teams to ensure recommendations are realistic for the care setting.
• Compare policy options and identify where additional evidence, workflow redesign, or prospective evaluation may be needed.
• Coordinate with staff leading predictive modeling, causal inference, process mapping, data engineering, and MLOps so optimization work fits into the broader implementation lifecycle.
• Prepare reports, presentations, documentation, and visuals that make optimization results and policy comparisons understandable.
• Explain methods, uncertainty, limitations, and recommended next steps to technical and non-technical audiences.
• Document reusable methods, assumptions, and implementation lessons for QPSi knowledge-sharing.

Requirements

• A Bachelor’s Degree in data science, statistics, bioinformatics, analytics, or similar field and five years of experience;
• A Master’s Degree in data science, statistics, bioinformatics, analytics, or similar field and three years of experience;
• A Doctoral Degree in data science, statistics, bioinformatics, analytics, or similar field and one year of experience.

Preferred

• Strong Python skills for data analysis, optimization, simulation, statistical modeling, and reproducible workflows.
• Experience with SQL and relational data sources, especially longitudinal or time-stamped real-world data.
• Demonstrated experience with multi-objective optimization, constrained optimization, operations research, dynamic programming, Bayesian optimization, or closely related methods.
• Experience constructing and interpreting Pareto fronts or Pareto regions, including approaches for robustness, uncertainty, outlier sensitivity, and clinically realistic comparator selection.
• Experience using reinforcement learning or policy-evaluation methods with historical or observational data. Relevant methods may include offline reinforcement learning, off-policy evaluation, contextual bandits, or dynamic treatment regimes.
• Experience modeling systems that evolve over time, such as patient trajectories, workflows, resource use, or sequential care decisions.
• Ability to define objective functions, reward functions, constraints, action spaces, state representations, and evaluation metrics with input from stakeholders.
• Practical judgment about when optimization or reinforcement learning methods are appropriate, and how to evaluate safety, uncertainty, bias, and implementation risk.
• Experience with healthcare data, EHR-derived data, or clinical quality improvement is strongly preferred.
• Experience with large-scale computing environments such as SLURM, Dask, GPU computing, or high-performance computing systems is preferred.
• Familiarity with version control, code review, testing, documentation, and other software engineering practices.
• Able to explain tradeoffs, assumptions, limitations, and uncertainty to both technical and non-technical audiences.
• Resourceful, collaborative, and comfortable working independently in ambiguous interdisciplinary project settings.