Abstract

In recent years, there has been a tremendous increase in health care expenditures in the United States. The most prevalent reimbursement system for health care expenses, Fee-for-service (FFS), has been deemed as one of the main reasons behind the high health care cost. Medicaid and Medicare have been exploring ways to transition from fee-for-service (FFS) to value-based-payment care plans, and Bundle Payments (BP) in particular. Adopting BPs can potentially improve the quality of care and efficiency by encouraging better coordination among the care providers.

We propose a two-step methodology with clustering and classification to characterize episodes of care by fusing a process in which we first apply spectral clustering to the procedural and revenue codes associated with an encounter of interest, and to those codes associated with the encounters most likely to proceed and to follow such an encounter. Secondly, to enhance cluster homogeneity, we apply a set of supervised learning algorithms to the resulting clusters after fusing their non-procedural information with the cluster characterization.

We compare the performance of the proposed methodology with a benchmark methodology over three encounters of interest: congestive heart failure (CHF), total knee replacement (TKR) and urinary tract infection (UTI) conditions. Our approach significantly reduces the variance of overpayment and underpayment associated with the variation resulting from the FFS payments per encounter and the reimbursement received as a consequence of a single payment per encounter in a cluster.

Library of Congress Subject Headings

Medical fees--Data processing; Medical fees--Classification; Medical care, Cost of--United States

Publication Date

12-19-2017

Document Type

Thesis

Student Type

Graduate

Degree Name

Industrial and Systems Engineering (MS)

Department, Program, or Center

Industrial and Systems Engineering (KGCOE)

Advisor

Ruben Proano

Advisor/Committee Member

Katie McConky

Campus

RIT – Main Campus

Plan Codes

ISEE-MS

Share

COinS