Ongoing Research

The following are some of the ongoing research projects being conducted by CHIDS Faculty and Fellows.

Public Health and Primary Care Integration through Enhanced Public Health Information Technology (PHIT) Maturity: A Case for Behavioral Health

The quality, cost and effectiveness of population health management in communities nationwide is critically dependent on effective coordination across primary care and public health organizations. Public health information technology (PHIT) provides unique opportunities for improved integration and coordination across public health systems and primary care providers. Limited evidence and understanding currently exists to aid communities in determining how effective their PHIT systems are for achieving care coordination. A primary objective of this project is to develop and validate an evidence-based PHIT Maturity Index to guide improvements in the usefulness, efficiency and outcomes of technology-mediated strategies for public health and clinical care program integration. The project will focus on behavioral health across a racially diverse underserved population and investigate integration across multiple PHIT systems including electronic health records (EHRs), public health reporting (PHR) data systems (such as surveillance systems); and ancillary systems (such as health information exchanges). The project is led by senior researchers and administrators from the University of Maryland, Montgomery County Department of Health and Human Services (MCDHHS), and the Primary Care Coalition of Montgomery County (PCC), and in coordination with the Chesapeake Regional Information System for Our Patients (CRISP) and Maryland’s State Health Improvement Process (SHIP). Funding provided by the Robert Wood Johnson Foundation.

Uma Ahluwali, Ritu Agarwal, Kenyon Crowley, Steve Galen, Bob Gold, Dushanka Kleinman, Tom Lewis, Stephen Manti, Karoline Mortensen, Colleen Ryan-Smith, Yang Yu

Evaluation and Innovation of Predictive Modeling for Healthcare Fraud, Waste and Abuse Prevention

CHIDS and CNSI of Gaithersburg, Md., have formed a partnership that marries innovation from industry with scientific research to reduce health care billing fraud, waste and abuse. This rampant problem creates excessive costs in the health care system and negatively impacts patient care, prompting the need for effective IT solutions to detect improper billing. CNSI’s ClaimsSure® solution can analyze sets of big data to detect fraud, waste and abuse and assess a health care claim’s risk, driving efficiency and cost savings through predictive probability analysis. Researchers at CHIDS will collaborate with CNSI engineers to conduct a series of technical, operational and economic assessments to explore ClaimsSure’s® modeling approach and effectiveness. The partnership aims to advance performance through scientific research for model technical excellence, system-wide impact assessment, and for the behavioral aspects of using the ClaimsSure® product. Through this partnership, CNSI and CHIDS hope to better understand the ability of analytical software to enable improved health care quality outcomes through IT innovation across the public sector.

Ritu Agarwal, Sean Barnes, Margret Bjarnadottir, Kenyon Crowley, Jennifer Jin, CK Kumar, Cecil Murthy, Kislaya Prasad, Nilakantan Rajaraman, Yang We

A Motivation Psychology-Based Smart Engagement System for Improved Older Adult Chronic Disease Management

Mobile technologies provide unprecedented opportunities for patients to engage with physicians and other peer patients and self-manage for disease management. However, there is no guarantee that the social and self-management produces desirable outcomes for all patients. Despite their increasing popularity, most behavioral intervention programs use a “one size fits all” approach, and individual patient’s psychological characteristics are seldom taken into account. As a result, social engagement and health self-management support, if not carefully designed, can be ineffective or even backfire for certain patients. This highlights the urgent need for personalized and adaptable smart social engagement and self-management support technologies for disease that are guided by a deep understanding of motivation psychology.

The objective of this project is to design, implement, and optimize a motivation psychology-based smart engagement system (MOSES) for older adults with diabetes. The MOSES tool, delivered via tablets, empowers the patients with two forms of social engagement, with healthcare providers, and with peer patients, as well as personalized self-engagement support. Our goal is to use MOSES as a research platform to advance knowledge on how the interventions should be optimized at the individual patient level by integrating motivational psychology theories. The project is being conducted in collaboration with the University of Maryland Medical School and the Fraunhofer Center for Experimental Software Engineering.

Ritu Agarwal, Kenyon Crowley, Michelle Dugas, Gordon Gao, Arie Kruglanski, Catherine Plaisant, Arnab Ray, Charles Song, Nanette Steinle, Bo Xie

Predicting Prostate Cancer Risk Using Magnetic Resonance Imaging Data

Prostate cancer is widely prevalent and hard to diagnose. The National Cancer Institute estimates that 16% of men born today will be diagnosed with prostate cancer in their lifetime (Howlader et al. 2012). In 2013, an estimated 238,590 new prostate cancer cases will be diagnosed and an estimated 29,720 men will die because of prostate cancer in United States (Siegel et al. 2013). Currently, the two main methods for diagnosing prostate cancer are a prostate specific antigen (PSA) test and a trans-rectal extended biopsy. Both of these methods have serious drawbacks. Although it has some predictive power, a PSA test can be unreliable with a high error rate (Hoffman et al. 2002). While a biopsy is more accurate, it is expensive and highly invasive with negative side effects (Cooper et al. 2004). Since a biopsy is conducted randomly within the prostate gland, it can result in a significant number of misses in cancer diagnosis, as well. Even when biopsy does detect cancer, the location can be uncertain and the localization of tumor within the prostate remains imprecise (Salomon et al 1999). We propose a non-invasive method for classifying patient risk using Magnetic Resonance Imaging (MRI) data. Logistic regression and nearest neighbor classification are combined to identify the risk of cancer. Our method performs well, having 79% predictive accuracy, and an area under the ROC curve of 0.85. It identifies the most aggressive cancers with 82% accuracy.

David Anderson, Bruce Golden, Edward Wasil, and Hao Howard Zhang

Understanding Sticker Prices: An Analysis of CMS Provider Charge Data

This year, the Centers for Medicare and Medicaid Services (CMS) released data containing information on provider charges (i.e., sticker prices) and CMS payments for the 100 most common inpatient services and 30 common outpatient services for the fiscal year 2011. Our objective is to evaluate how CMS payment models explain payments and sticker prices for both inpatient and outpatient services. We have found that payments for inpatient and outpatient services are explained extremely well by their respective models, but these payment models do not explain sticker prices well. We are currently applying advanced machine learning methods to an augmented data set to improve our understanding of sticker prices. Our analysis demonstrates that providers are using additional (and potentially diverse) information to determine their prices. Ongoing research investigates whether higher sticker prices are associated with higher quality providers, higher profits, and other potential provider-based outcomes.

Margret Bjarnadottir, Sean Barnes

Data Driven Detection of Episodes-of-Care

Healthcare reimbursement is at the forefront of healthcare reform debates in the United States. Bundled payment systems have been proposed as a practical and promising reimbursement alternative to induce incentives that lower healthcare expenditures. Bundled payments reimburse a single amount for an episode of care in contrast to separate payments for every single medical service rendered in the current fee-for-service model. Significant evidence suggests that bundled payment systems achieve improved quality outcomes, and better coordinated care at a lower cost. The present process of defining an episode of care relies on the consensus of expert panels on which services constitute an episode. This manual approach is inconsistent and not systematic. The approach is labor-intensive and inflexible and is inefficient for dynamically incorporating heterogeneity in patients’ health conditions, comorbidities, and medical practices across communities. Additionally, these manual approaches have focused on individually definitions of episode of care, which does not allow communities to analyze the collective impact of moving to a bundled payment system on their overall expenses and healthcare outcomes. Consequently, it is currently impossible to analyze and design the necessary conditions that will facilitate the adoption of bundled payments in a community, and across the country. Although the expert panel approach is clinically meaningful, the process of defining an episode of care could be enhanced and accelerated by employing sophisticated data mining and optimization models.

This study is developing data mining algorithms that deal with large unstructured and heterogeneous data to characterize episodes of care for different groups of patients. This study uses a large repository of claims data, over 110 million claim records of 660,000 insured customers, to derive service distributions for different episodes of care, using advanced data mining methods. The study will analyze ten episodes that are chosen to form a representative mix of conditions which considers the following: (1) particular episodes exhibit fairly standard utilization of services while others are highly variable; (2) certain episodes affect a wide range of patient demographics, and (3) some episodes tend to be associated with co-morbidities, while others are treated in isolation. To analyze the effects of moving to a bundled payment system, optimization models will be built that aim to minimize the overall costs of the system while restricting for example the financial risks of providers, taking into account the variability in services and provider volumes. This project aims to (1) accelerate knowledge acquisition of bundled payments systems through a data-driven approach for episode of care definition, and (2) analyze how episode of care definitions impact overall costs of payers and the financial risks of providers. This project will result in new knowledge about the set of services that should constitute an episode of care. Combined with system pricing optimization this can facilitate the analysis and potential large-scale implementation of bundled reimbursement systems. The implementation of a bundled payment system holds promise for both healthcare cost reduction and quality improvement. Understanding the extent to which the system can be adopted and the effects of its implementation on healthcare providers and other stakeholders are instrumental to the general adoption of bundled payment systems. Finally, the project will not only advance the understanding of episodes of care and bundled payment systems, it will also contribute to the science of data mining and data driven algorithms, as in order to address the research questions, new methods and extensions to previous methodologies will be developed. In addition, the data analysis methods developed in this proposal can be applied to various analyses of the health care system including investigation of differences in services across providers and knowledge discovery of best practices (successful treatment patterns).

Cost Analysis of Reducing Unnecessary Cranial Computed Tomography Utilization among Children with Minor Blunt Head Trauma

The project objective is to quantify the financial impact of electronic health record (EHR) integrated decision support aimed at reducing unnecessary emergency department (ED) computed tomography (CT) for head trauma. Assessment of the financial impact of prediction rules as decision support, is important for widespread adoption. We performed a retrospective study of subjects 0 to 18 years of age, with head trauma presenting to the ED from January 2010 to March 2012. Two validated clinical prediction rules were implemented as decision support in the EHR of the ED of a children’s hospital to reduce unnecessary computed tomography scan use for head trauma.  We performed a detailed analysis of the clinical impact, costs, charges, and revenue for patients before and after implementation. Costs were determined on a relative value units basis.  The full results/data of this study will be published soon, but briefly, implementation of EHR integrated computerized decision support was associated with a reduction in rates of CT, total costs, total charges, and total revenues. Since the reduction in revenue is greater than the cost savings, decreasing computed tomography utilization does not necessarily translate to direct financial benefits for healthcare providers.  Careful consideration will be needed to align financial incentives when reducing unnecessary diagnostic testing.