A Study of Rural Health Care for Chronic Care Management Using Machine Learning to Determine Costs

Aug 5, 2026

Study compares rural Louisiana Medicaid recipients with comorbidities to national averages and evaluates how machine learning can support Chronic Care Management cost and utilization decisions. The work applies NCQA Quality Measures to blinded Louisiana Department of Health claims, then models Medicaid churn with logistic regression and random forest classifiers. It forecasts future rural medical costs using ARIMA. Results indicate lower rural adherence than the national average, with improvements over the study period.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages193

Reading Time8 h 6 min

AccessFree to download

Abstract

Approval for Scholarly Dissemination

Dedication

Table of Contents

Chapter 1: Introduction

Chapter 2: Background

2.1 Chronic Care Management

2.2 Motivation of Work

2.3 Problem Statement

2.4 Research Objectives

Louisiana Tech University Louisiana Tech Digital Commons | Doctoral Dissertations | Graduate School | | --- | --- | Summer 8-23-2025 A Study of Rural Health Care for Chronic Care Management Using Machine Learning to Determine Costs Joanna Maria Keeling Follow this and additional works at: [https://digitalcommons.latech.edu/dissertations](https://digitalcommons.latech.edu/dissertations) Part of the Computer Sciences Commons, and the Social Statistics Commons A STUDY OF RURAL HEALTH CARE FOR CHRONIC CARE MANAGEMENT USING MACHINE LEARNING TO DETERMINE COSTS by Joanna Keeling, BS, MS A Dissertation Presented in Partial Fulfillment of the Requirements of the Degree Doctor of Philosophy in Computational Analysis & Modeling COLLEGE OF ENGINEERING AND SCIENCE LOUISIANA TECH UNIVERSITY June 2025 ABSTRACT Management of diabetes or heart disease may be uniquely challenging for older individuals with multiple chronic conditions[1] . Chronic Care Management would target patients living with comorbidities in Louisiana, who are not receiving sufficient healthcare services and help them receive the care they deserve[2] . This study is aimed at comparing the results of rural Louisiana Medicaid recipients with comorbidities to the National Averages. We will also use Machine Learning to predict Churn and Medical Costs. To help us identify these patients with comorbidities, we decided to use NCQA Quality Measures. We identified several measures from NCQA that we wanted to use, but we narrowed it down to five[3],[4],[5],[6],[7] . We acquired the blinded data from the Louisiana Department of Health (LDH) Medicaid data. This data is in the form of claims received from hospitals and doctors’ offices for patients in the Medicaid program[8] . After calculating our measures, we used the data we have created to predict the churn of patients in and out of Medicaid using Logistical Regression and Random Forest Classifier models. Finally, we used the initial data and machine learning to predict future costs ofthe rural patients using the ARIMA model. This study shows that rural communities in Louisiana have much lower rates of adherence to doctor visits and medication therapy than the national average. However, these rates are improving and have improved significantly over the six-year period of which this study investigated. We had positive results for both machine learning models in churn, but the best results came from not only the Random Forest Classifier but also from the inclusion of the data from the measures. The cost of medical care is increasing gradually on a monthly basis, but it still remains on a slow steady pace. Chronic Care Management systems would help improve patient outcomes, by helping patients manage treatment and doctor/hospital visits. By getting the patients the care they need more promptly, we can stave off worsening conditions and lower the overall costs of patient care. APPROVAL FOR SCHOLARLY DISSEMINATION The author grants to the Prescott Memorial Library of Louisiana Tech University the right to reproduce, by appropriate methods, upon request, any or all portions ofthis Dissertation. It is understood that “proper request” consists of the agreement, on the part of the requesting party, that said reproduction is for his personal use and that subsequent reproduction will not occur without written approval of the author of this Dissertation. Further, any portions ofthe Dissertation used in books, papers, and other works must be appropriately referenced to this Dissertation. Finally, the author of this Dissertation reserves the right to publish freely, in the literature, at any time, any or all portions ofthis Dissertation. Author Date GS Form 14 (5/03) DEDICATION I dedicate this Dissertation to my mum, Josephine Phares. She was my greatest cheerleader and closest friend during this process. Mum, I miss you and your wisdom. Thank you for believing in me and loving me

FAQs

What is the primary goal of the dissertation?

The dissertation compares rural Louisiana Medicaid recipients with comorbidities to national averages and uses machine learning to predict churn and medical costs.

How are patient comorbidities and quality outcomes identified?

It uses NCQA Quality Measures and applies them to blinded Louisiana Department of Health Medicaid claims data.

Which machine learning and forecasting methods are used in the study?

Medicaid churn is predicted using logistic regression and random forest classifiers, while future rural medical costs are forecast using the ARIMA model.

Aug 5, 2026

Upload by Seraphina