Overview
What this challenge is about.
You receive a simplified pathway: 5 possible underlying conditions, 8 possible diagnostic tests each with documented sensitivity and specificity, and an outcome payoff matrix from clinical guidelines. Formulate a POMDP with belief states over conditions, test actions, and terminal treat-or-refer actions. Solve with QMDP, PBVI, or a small-scale SARSOP variant. Compare on expected cost-per-patient and expected diagnostic delay versus a hand-coded decision tree on a 1,000-patient simulated cohort. The deliverable is the formulation, the solver, and a clinical-leadership memo.
The Brief
What you'll do, and what you'll demonstrate.
Show whether a POMDP-based pathway beats the current decision tree on expected cost and diagnostic delay in simulation.
Earning criteria — what you'll demonstrate
- Formulate a clinically-grounded POMDP with realistic observation models
- Implement and reason about an approximate POMDP solver
- Compare policies on operational and clinical metrics jointly
- Communicate model limits to a regulated-product audience
Program Fit
Where this fits in your program.
Sharpens the same skills your degree expects you to demonstrate.
Skills
Skills you'll demonstrate.
Each one shows up on your verified credential.
Careers
Roles this prepares you for.
Real titles. Real skill bridges. Pick the one closest to your trajectory.
Career paths this builds toward
Canonical rolesML Researcher
POMDP work on real clinical problems is a strong portfolio piece for any RL or sequential-decision ML research role.
This challenge sharpens
- pomdp-modeling
- belief-states
- approximate-solvers
Research Scientist
Building a benchmark-grade comparison against the incumbent decision tree is exactly the rigor expected of a junior research scientist.
This challenge sharpens
- pomdp-modeling
- simulation
- decision-theory
Applied AI Scientist
Translating a POMDP into an FDA-readable memo is the daily work of applied AI scientists in regulated industries.
This challenge sharpens
- pomdp-modeling
- simulation
- python