WHITE PAPER Turning Data into Reality: Seizing the Opportunity for Transformation Using Big Data Analytics in Human Services May 2013 1 cgi.com 2013 CGI GROUP INC.
Introduction The chatter around big data and data analytics suggests the potential for a revolution in organizational decision making. According to the hype, by combining large volumes of data, we can find the answers to almost any question. This is true to an extent: through emerging technologies, we are able to combine data sets much more easily, allowing us to answer new questions or those we previously thought were unanswerable. However, the challenge in realizing this technology s full potential is in deciding which questions need answering. This is not only a technology challenge but a human one as well. Some industries are seeing concrete benefits of compiling and analyzing big data. For example, the retail industry is utilizing consumer purchasing data to model behaviors and target specific demographics in marketing campaigns. Retailers are harnessing this data to learn which buyers visit each store location, how often to re-stock certain products, and where to strategically place products to increase sales. The financial industry is analyzing large quantities of data related to credit card usages. They use this information to limit fraud by alerting customers when unusual transactions occur. In the public sector, Medicare and Medicaid programs use big data attributes to root out improper payments and waste, fraud and abuse. In fact, these efforts led to the collection of $2.4 billion worth of payments in Fiscal Year 2012 alone. What Does Big Data Mean for Health and Human Services? The Policy Council of the American Public Human Services Association (APHSA) launched Pathways: The Opportunities Ahead for Human Services in 2012 that describes an integrated health and human service system as one that produces value for the people and communities served from the perspective of: a) gainful employment and independence; b) stronger families, adults, and communities; c) healthier families, adults, and communities; and d) sustained well-being of children and youth. To create, deliver, and realize that value, APHSA developed the following vision statement: A Vision of the Future of Health and Human Services A fully integrated health and human services system that operates a seamless, streamlined information exchange, shared services and coordinated care delivery system that is a consumer-focused, modern marketplace experience designed to improve consumer outcomes, improve population health over time, and bend the health and human services cost curve by 2025. Many states are transforming their human services delivery system by moving to implement this vision - but from different perspectives and at different rates. These efforts range from developing on-ramps to the Federal Health Insurance Exchanges/Marketplaces to creating their own free-standing versions of the FHE, as well as re-building or modifying their eligibility and enrollment systems and business processes. Regardless of the approach, states are compelled to redefine their strategic objectives in line with this vision of the 21st Century business model for health and human services; i.e., to create operating systems 2
that are client-focused, interoperable and focused on improved outcomes. Key to this transformation is a state s ability to leverage advancements in technology and data analysis to support each organization s unique business objectives. Regardless of the different perspectives or rate of change, understanding how data can be shared and better utilized within the overall context of the organization s future direction, allows considerable steps toward seeing health and human services from a larger perspective. APHSA developed a 21 st Century Business Model that outlines the components of the vision identified through Pathways. To operationalize this vision, APHSA developed a Health and Human Services Integration Maturity Model and subsequent self-assessment tool to assist states identify where they lie along the integration continuum. The model describes current operations, or the as-is state, and a vision of transformation, or the to-be state. Key to this transformation is their ability to leverage advancements in technology and data analysis to support each organization s unique business objectives. Regardless of the specific steps followed, however, by understanding how data can be shared and better utilized within the overall context of the organization s future direction, we can take considerable steps toward seeing health and human services from a larger perspective. Advancement in data analytics becomes possible as different data sets become more readily available under this model. When the topic of big data arises, big hype and big cost typically follow in the discussion. The good news is that these traps can be avoided. The technologies that provide the powerful capabilities to analyze the structured data of today s systems, as well as the fast-moving unstructured data of sources such as social media, are readily available and affordable. When combined with cloud computing, these technologies allow states the potential to analyze their data in new ways. More importantly, there are now mechanisms to truly demonstrate outcomes and apply scarce resources to the best effect. This will be important not only as budget pressures increase, but to best provide health and human services to our increasing population of aging citizens. Traditionally, big data is described in terms of variety, volume and velocity. Big Data Term Volume Variety Velocity Definition Amount of available data Different sources and types of data, both structured and unstructured Speed of data being produced, processed, and made available for access and delivery Though the Three V s may be the traditional definition, this does not mean that a big data project must be far-reaching. In fact, any data analysis project, whether using traditional structured data or unstructured data, should start small with a known information problem that needs to be addressed. Big data programs are also maturing. Today, most big data projects are focused on driving process efficiencies, generating revenue, and improving strategic performance. Strategic performance projects are typically much more complicated and cut across departments focusing on multi-departmental or state-wide goals. 3
Figure 2: Types of Big Data Projects State Health and Human Services Seeing Early Success States are beginning to use data analytics to improve services and help reduce costs. Budget pressures push the importance of recovering revenue. For example, the State of Illinois Department of Health and Human Service (HHS) used data analytics to identify over $27M in overpayments to health care providers and an additional $14M in improper payments. States like Louisiana are using this model of detecting fraud in Medicaid to benefit programs related to worker s compensation, unemployment and taxes. Data analytics around fraud, waste, and abuse of state benefits will continue to mature. But the opportunities don t end there. Some states are beginning to use data analytics to help with more mature performance goals both administrative and programmatic. Traditional program goals revolve around the number of cases closed, or number of days to process a claim. This is one reason that information technology systems were designed to manage a single process. A manager can query the single system to determine those simple metrics. As processes develop and time-to-complete is reduced, there is still the acknowledgment that the number of claims is increasing. Therefore, the program may need to be looked at differently. Perhaps the goal is not to quickly process what comes in, but rather, to find out why the number of claims continues to 4
increase. We are beginning to see how states are using data differently, and how to analyze it against other sets of data that are free from the current siloed system. For example, the state of Michigan set a goal to enhance child protection services through improving case management around child abuse, neglect, foster care, adoption and legal guardianship among 14,500 children enrolled in child welfare programs. By partnering with State Administration Court, the Department of Health and Human Services shared performance measurements in the areas of safety, pregnancy, timeliness and well-being. By sharing data and performing analytics on related information, they were able to increase family reunifications by more than 30% among temporary court wards in just one year. To date, most of the data analytics projects look at what occurred after the process in order to make improvements to a program. However, more mature data analytics use data to predict future outcomes to shift resources before the problem occurs which often saves money. A great example is the U.S. federal government s attempt to predict homelessness rates and the potential to predict where the rates might increase. The U.S. Department of Housing and Urban Development (HUD) is partnering with the Veterans Affairs Department to identify locations with high concentrations of military personal that are retiring or not reenlisting. While it is common for young people to sign up for only 2 or 4 years of military service, those areas with large numbers of these individuals are often in need of housing assistance and other programs. Rather than waiting for the problem to occur, the government can proactively support these areas. Examples of Big Data Analytics Opportunities in Health and Human Services Process Efficiency: Family Case management improvements Adoption Case management Recovering revenue: Fraud related to child care benefits Fraud related to workers compensation Fraud related to Medicaid Strategic Direction: Reduce the amount of low birth weight babies born to mother on welfare Reduce the number of food stamp recipients Lower the percentage of citizens living in poverty 5
Getting Started It is important to recognize that you cannot simply go full throttle when it comes to big data. Think big, but start slow -- it is best to cultivate expertise and build on demonstrated successes. Answers to the following questions can be a useful guide. ` Questions 1) What kind of transformation is the most important to the department? What problems need solving? 2) What kind of data can help me solve the problem? How do I need to organize to gain access? 3) Do you have analysts and subject matter experts available and skilled to support? 4) Do you have the technical tools to bring the data together and support analysis? Figure 2: Getting Started with Data Analytics Big data projects are initiated to solve a business problem. Transformational projects can include major business process improvements, revenue generation, or new performance goals. In some cases, transformation may involve multiple goals and objectives. Once the problem is identified, enlist subject matter experts to help identify data that can be combined to solve the issue. It may make sense to bring in outsiders who have relevant experience in either the same industry, or a similar agency, to bring a fresh perspective. 6
Data Inventory is Key Take a data inventory of all existing systems that serve the state health and human service enterprise, including judgments about data quality and compatibility. Take a data inventory of all existing external systems that send/receive data to/from the health and human service enterprise, including judgments about data quality and compatibility. Make an inventory of external data sources of health care data from the federal government and other relevant sources. Engage with outside data experts for further insights into possible data sources. External publicly-available information is often overlooked, yet valuable. Once you identify the data sources, assess the sensitivity of each data set. Are there security or privacy concerns associated with the data set? If so, make sure to take every precaution to ensure the protection of this data. Know which data is sensitive and understand the level of risk associated with a security breach. Often data can be protected within the construct and design of your data system, but knowing what needs protecting is important to that design. Finally, involve a systems and data architect and technical staff to design how the data will be brought together. The architects build the application layers and identify the proper tools and data storage requirements based on the business needs. Technical and program staff work together to develop rule engines and models used in analytical tools. They also work together to design the visual display of the data. This may seem complicated, but today there are many different analytical tools some of which can be stood up in a cloud environment to be shared across agencies. Once a data project has been executed, the emerging data culture should be institutionalized. This includes developing a data governance model, extending policy and practices for data sharing, and ensuring that privacy and cyber security are adequately addressed. In the current budget environment, it will serve you well to proceed slowly but deliberately with specific goals and objectives in mind. By creating a data culture, while establishing and institutionalizing good data management practices, you will position yourself well for future needs and requirements. not e deleted. 7
References: McKinsey Global Institute, Big Data: The Next Frontier for Innovation, Competition, and Productivity, May 2011 TechAmerica, Demystifying Big Data: Practical Guide to Transforming the Business of Government Cari DeSantis for the American Public Human Services Association, Business Model for Horizontal Integration of Health and Human Services, 2012 CMS, Medicare National Recovery Audit Quarterly Newsletter, September 30, 2012. Optum, White Paper - Embracing a New Data Culture, 2012 cgi.com 2013 CGI GROUP INC. 8