Optimizing Diagnostic Decisions Under Resource Constraints in Breast Cancer Screening

Primary tabs

Oguzhan Alagoz, PhD - University of Wisconsin-Madison


Mammography is the most commonly used screening tool for early diagnosis of breast cancer, the most common non-skin cancer affecting women in the U.S.  While mammography is inexpensive, the interventional procedures that result from detected abnormalities (both false and true positives) increase the cost of this population-based screening program significantly.  Based on the mammography findings, radiologists need to choose one of the following three options: 1) take immediate diagnostic actions including prompt biopsy to confirm whether an abnormality is in fact a breast cancer; 2) recommend short-interval follow-up mammography; 3) recommend routine mammography. There are no validated structured guidelines to aid radiologists in making patient management decisions after mammography exams. Surprisingly, 55-85% of the breast biopsies ultimately are found to be benign breast lesions and only less than 1% of short-interval follow-up recommendations are found to be malignant, resulting in additional tests, patient-anxiety and expenditures.

We develop a finite-horizon discrete-time constrained Markov decision process (MDP) to model diagnostic decisions after mammography where we maximize the total expected quality adjusted life years (QALYs) of a patient under resource constraints. We prove that the optimal value function is concave in the allocated budget. We use clinical data to estimate the parameters of the MDP model and solve it as a mixed-integer program. We find that as women get older they should be biopsied less aggressively. We further find that short-term follow-ups are the immediate target for elimination when budget becomes a concern. We compare our model to clinical practice and find that our model has a potential to improve breast cancer diagnosis and reduce associated costs significantly.


Speaker Bio:

Oguzhan Alagoz is currently an Associate Professor of Industrial and Systems Engineering at the University of Wisconsin-Madison. In addition, he is an associate professor at the Department of Population Health Sciences and serves as the director of National Institute of Health (NIH)-funded Institute for Clinical and Translational Research-Simulation Center at UW-Madison School of Medicine and Public Health. He received his BS from Bilkent University in 1997, MS from Middle East Technical University in 2000, and PhD in industrial engineering from the University of Pittsburgh in 2004. His research interests include stochastic optimization, medical decision making, completely and partially observable Markov decision processes, simulation, risk-prediction modeling, health technology assessment, and scheduling. He is on the editorial boards of Medical Decision Making and IIE Transactions on Healthcare Engineering. He serves as the president of INFORMS Health Applications Section.  He has received various awards including a CAREER award from National Science Foundation (NSF), outstanding young industrial engineer in education award from IIE, Dantzig Dissertation Honorable Mention Award from INFORMS, 2nd place award from INFORMS Junior Faculty Interest Group best paper competition, best paper award from INFORMS Service Science Section, and best poster award from UW Carbone Comprehensive Cancer Center. He has been the principal investigator and co-investigator on grants more than $2.2 million funded by NSF and the National Cancer Institute of NIH. He is a member of INFORMS, IIE, SMDM, and CISNET.




  • Workflow Status:
  • Created By:
    Colly Mitchell
  • Created:
  • Modified By:
    Fletcher Moore
  • Modified:

Target Audience

    No target audience selected.