Abstract:In many applied optimization settings, parameters that define the constraints may not guarantee the best possible solution, and superior solutions might exist that are infeasible for the given parameter values. Removing such constraints, re-optimizing, and evaluating the new solution may be insufficient, as the optimizer's preferences in selecting the existing solutions might be lost. To address this issue, we present an inverse optimization-based model that takes an observed solution as input and aims to improve upon it by projecting onto desired hyperplanes or expanding the feasible set while balancing the distance to the observed decision to preserve the optimizer's preferences. We demonstrate the applicability of the model in the context of radiation therapy treatment planning, an essential component of cancer treatment. Radiation therapy treatment planning is typically guided by expert-driven guidelines that define the optimization problem but remain mostly general. Our model provides an automated framework that learns new plans from available plans based on given clinical criteria, optimizing the desired effect without compromising the remaining constraints. The proposed approach is applied to a cohort of four prostate cancer patients, and the results demonstrate improvements in dose-volume histograms while maintaining comparable target coverage to clinically acceptable plans. By optimizing the parameters of the treatment planning problem and exploring the Pareto frontier, our methodology uncovers previously unattainable solutions that enhance organ-at-risk sparing without sacrificing target coverage. The framework's ability to handle multiple organs-at-risk and various dose-volume constraints highlights its flexibility and potential for application to diverse radiation therapy treatment planning scenarios.

Expert-Guided Inverse Optimization for Convex Constraint Inference

Inferring Linear Feasible Regions using Inverse Optimization

Conformal Inverse Optimization

Inverse Learning: Solving Partially Known Models Using Inverse Optimization

Improving Observed Decisions Quality using Inverse Optimization: A Radiation Therapy Treatment Planning Application

Inverse Optimization of Convex Risk Functions

Inverse optimization in semi-definite programs to impute unknown constraint matrices

A Surrogate Model Based Nested Optimization Framework for Inverse Problem Considering Interval Uncertainty

Constraint Inference in Control Tasks from Expert Demonstrations via Inverse Optimization

Optimization Methods for Inverse Problems

A Direct Coupled Solution Methodology for Efficient Robust Optimizations of Inverse Problems under Uncertainty

Ill-posed linear inverse problems with box constraints: A new convex optimization approach

Deep Inverse Optimization

You Are What You Eat: A Preference-Aware Inverse Optimization Approach

ON THE INVERSE CONTINUOUS OPTIMIZATION AND ITS SMOOTHING FISCHER-BURMEISTER FUNCTION APPROACH

Learning to Solve Optimization Problems with Hard Linear Constraints

Estimation theory and model parameter selection for therapeutic treatment plan optimization

From Inverse Optimization to Feasibility to ERM

Convex Optimisation for Inverse Kinematics

Towards a variational principle for motivated vehicle motion

Derivative-free stochastic bilevel optimization for inverse problems