Analysis of Mammography Screening Policies under Resource Constraints

Abstract

Breast cancer, the leading cause of cancer death for women, can be detected at earlier stages through mammography screening. Therefore, most developed countries implemented population‐based mammography screening programs. However, cost of mammography and limited resources in terms of number of trained personnel and diagnostic machines prevent mammography screening to be adopted by many other countries. In fact, even in resource‐rich countries, there is a growing concern about cost of mammography screening. In this study, we investigate the optimal allocation of limited mammography resources to screen a population. We propose a constrained partially observable Markov decision process (CPOMDP) model that maximizes total expected quality‐adjusted life years of the patients when they are allowed only a limited number of mammography screenings. We use a variable resolution grid‐based approximation scheme to convert the CPOMDP model into a mixed‐integer linear program and conduct several numerical experiments using breast cancer epidemiology data. We observe that as mammography screening capacity decreases, patients in the 40–49 age group should be given the least priority with respect to screening. We further find that efficient allocation of available resources between patients with different risk levels leads to significant quality‐adjusted life year gains, especially for the patients with higher breast cancer risk.

Preference‐Sensitive Management of Post‐Mammography Decisions in Breast Cancer Diagnosis

“…First, we introduced a modeling scheme for tailoring sequential medical interventions toward risk-sensitive care. Extant OM literature on disease management addresses conditions including chronic diseases (Ibrahim et al 2016, Skandari et al 2015), transplantation (Alagoz et al 2004, Su and Zenios 2004, 2006), cancer (Cevik et al 2018, Erenay et al 2014, Güneş et al 2015, Zhang et al 2012), and HIV (Zaric et al 2000) where decision makers are risk-neutral. Although there exist few recent studies that investigated risk-sensitive MDP modeling for various medical decisions, our approach in modeling sequential decisions is unique because our model accounts for the expected future longevity as the reference point at which the risk-sensitive decision is made and also treats the expected intermediate longevity as the pay-off of the lottery.…”

Scalable grid‐based approximation algorithms for partially observable Markov decision processes

“…For instance, the LB algorithm was previously used to solve a POMDP model for prostate cancer screening 6 . Moreover, other than providing a solution quality check for the LB heuristic, the UB heuristic is useful for handling POMDPs with mixed‐observability (i.e., the state space includes both fully observable and partially observable states), 7 and provides a building block for constrained POMDP algorithms 8 . There exist many other approximation algorithms for POMDPs such as Monte Carlo (MC) based approaches 9 and Point Based Value Iteration (PBVI) based algorithms, 10 which provide basis for various other approximation methods.…”

“…It is important to note that while these POMDP approximation methods may perform better for certain practical instances (especially for those with large state spaces), the main advantage of Lovejoy's LB and UB methods 5 over those is the bounding mechanism that enable assessing the quality of solution. Accordingly, also considering their relative simplicity for implementation and customization purposes, Lovejoy's methods 5 and the variants have been used in various recent studies 7,8,11 …”

Investigating the Effectiveness of Breast Cancer Supplemental Screening Considering Radiologists’ Bias

“…Molani et al [34] developed POMC models to quantify the age and stage-specific overdiagnosis risks while considering the uncertainty in a patient’s adherence behavior. Cevik et al [14] proposed a POMDP model to maximize the total expected QALYs of a patient when there is a constraint on the number of mammograms the patient can undergo. Sandikci et al [44] formulated a POMDP model to determine the optimal breast cancer screening policies considering patients’ breast density.…”