Concept Architecture
Concept
Theoretically, Robust Optimisation is a mathematical optimisation framework that identifies solutions which remain feasible and near-optimal under uncertainty in model parameters. Rather than optimising for a single assumed scenario, robust optimisation explicitly incorporates uncertainty sets into the optimisation problem, producing decisions that are resilient to variation in costs, resource availability, demand, epidemiological parameters, or other uncertain inputs. In health economics, it is used where parameter uncertainty may substantially affect resource allocation or policy decisions.
Mathematically, robust optimisation extends a conventional optimisation problem by requiring constraints and objective functions to remain valid for all parameter values contained within predefined uncertainty sets. The optimisation therefore seeks the decision that performs best under the worst-case realisation of uncertainty, while maintaining feasibility across all allowable parameter values. Depending on the uncertainty structure, the resulting problem may remain linear or become nonlinear or mixed-integer.
In practice, robust optimisation is implemented by specifying uncertain parameters, defining appropriate uncertainty sets, formulating the robust optimisation model, and solving it using recognised optimisation algorithms. In health economics it is applied to healthcare resource allocation, workforce planning, hospital capacity management, vaccination strategies, supply chain optimisation, and treatment planning where uncertainty must be explicitly incorporated into decision-making.
Purpose
Used to determine healthcare decisions that remain feasible and perform well despite uncertainty in model parameters, thereby improving the resilience and reliability of resource allocation decisions.
Mathematical Formulae
Primary Formula
The robust optimisation problem is
min?x ? X? max?u ? ??? f(x, u)
where:
- x = decision variable vector
- u = uncertain parameter vector
- ?? = uncertainty set
- f(x, u) = objective function
Supporting Formulae
Robust linear constraint:
a(u)?x � b, ?u ? ??
Worst-case objective:
max?u ? ??? f(x, u)
Related Mathematical Methods
- Linear programming
- Nonlinear programming
- Convex optimisation
- Minimax optimisation
- Scenario optimisation
- Stochastic optimisation
- Sensitivity analysis
- Numerical optimisation
Example
A regional health authority allocates vaccine doses while future demand is uncertain.
The objective is to minimise total distribution cost while satisfying demand under all plausible scenarios.
Assume demand is expected to be 100,000 doses with an uncertainty interval of �10%.
The optimisation is formulated over the uncertainty set
90,000 � D � 110,000.
The robust solution recommends stocking 108,000 doses. Although this is not the least-cost solution for the expected demand alone, it remains feasible throughout the uncertainty interval and avoids shortages under higher-demand scenarios.
Excel Implementation
| Function | Example Formula | Health Economics Application |
|---|---|---|
| Solver Add-in | Minimise objective subject to robust constraints | Solves robust resource allocation and planning problems. |
| MAX | =MAX(B2:B11) | Identifies worst-case parameter values across uncertainty scenarios. |
| MIN | =MIN(C2:C11) | Evaluates minimum performance across scenarios. |
| SUMPRODUCT | =SUMPRODUCT(B2:B11,C2:C11) | Calculates objective functions under alternative uncertainty scenarios. |
VBA (Optional)
A VBA procedure can automate robust optimisation by generating uncertainty scenarios, solving each optimisation problem, and identifying decisions that satisfy all specified uncertainty constraints.
Sources
- Ben-Tal A, El Ghaoui L, Nemirovski A. Robust Optimization. Princeton University Press.
- Bertsimas D, Sim M. The Price of Robustness. Operations Research. 2004;52(1):35?53.
- Briggs A, Claxton K, Sculpher M. Decision Modelling for Health Economic Evaluation. Oxford University Press.
- Drummond MF, Sculpher MJ, Claxton K, Stoddart GL, Torrance GW. Methods for the Economic Evaluation of Health Care Programmes. Oxford University Press.
- NICE. Health Technology Evaluation Manual.
- ISPOR Good Practice Reports on decision-analytic modelling.
Related Concepts (4)
Library
Publications
1
A Taxonomy of Model Structures for Economic Evaluation of Health Technologies — Brennan, Chick & Davies, Vol. 15, No. 12 ed., 2006 (Health Economics)
An influential paper classifying decision-analytic model structures along axes of expected value vs randomness, entity heterogeneity, and Markovian vs non-Markovian structure — providing a framework for choosing between decision trees, Markov cohort models, microsimulation, discrete event simulation and system dynamics.
Journal ArticleView source →
Frequently Asked Questions (6)
What is robust optimisation?
An optimisation approach seeking a solution that performs well across a range of possible uncertain parameter values, rather than one best-guess scenario.
Source: Ben-Tal & Nemirovski 1998
How does robust optimisation differ from optimising a single best guess?
Optimising to a single best-guess set of inputs can give a solution that is excellent if those inputs are right but poor if they are even slightly wrong. Robust optimisation instead seeks a solution that performs acceptably across the whole range of plausible inputs, deliberately sacrificing peak performance in the expected case for protection against bad ones. It answers to uncertainty in the inputs rather than assuming them known. The result is steadier but rarely the best in any single scenario. Ben-Tal and colleagues (2009) set out the approach.
Source: Ben-Tal et al. 2009
Why is robust optimisation used?
Robust optimisation is used because a solution optimised for a single estimate of uncertain parameters can perform badly if the true values differ, so where parameters are uncertain, a solution that is good only for the best guess may be fragile. Robust optimisation instead seeks a solution that performs acceptably across the range of possible values, guarding against the harm that uncertainty could cause. It is valuable when the consequences of a poor outcome under adverse conditions are serious and worth protecting against.
Source: Ben-Tal & Nemirovski 1998
How does robust optimisation handle uncertainty?
Robust optimisation handles uncertainty by defining an uncertainty set of possible parameter values and seeking a solution that optimises performance under the worst case within that set, or that remains feasible and good across it. Rather than using a single value for each uncertain parameter, it considers the range they might take and protects against the least favourable outcomes. The size and shape of the uncertainty set control how conservative the solution is, balancing protection against uncertainty with performance under typical conditions.
Source: Ben-Tal & Nemirovski 1998
What is the trade-off in robust optimisation?
The trade-off in robust optimisation is between robustness and performance: protecting against the worst case within the uncertainty set makes the solution more resilient to adverse conditions but generally less good under typical or favourable conditions, since it sacrifices some optimality for safety. A larger uncertainty set gives more protection but a more conservative, and potentially costlier, solution. The decision maker chooses how much robustness to seek, balancing guarding against uncertainty against giving up performance when conditions are benign.
Source: Ben-Tal & Nemirovski 1998
How does robust optimisation apply to health decisions?
Robust optimisation applies to health decisions made under parameter uncertainty, such as planning resources or designing interventions where costs, demand, or effects are uncertain, and where a plan optimised for a single estimate might fail if reality differs. Seeking a solution that performs well across the plausible range protects against poor outcomes under adverse conditions. This suits decisions where resilience matters, complementing approaches that handle uncertainty probabilistically, by focusing on performing acceptably across the range rather than on the expected case alone.
Source: Ben-Tal & Nemirovski 1998
Trust Record
Verified by Dr Darrin Baines
British health economist
Professional identity: darrinbaines.org
Verification date: 3 Oct 2025
Content version: 1.0.0
Canonical Identity
- Persistent URI
- https://healtheconomics.wiki/concept/robust-optimisation
- Term code
- HE-EM-DM-086
Stable URI · Machine-readable · Resolvable · CC BY 4.0