Latest revision as of 07:44, 30 April 2023

Business intelligence (BI) is an umbrella term to describe concepts and methods to improve business decision-making by using fact-based support.^[1] This page discusses BI primarily from a healthcare perspective.

Introduction

Business intelligence, defined in more detail, comprises the strategies, processes, applications, data, products, technologies and technical architectures used to support the collection, analysis, presentation, and dissemination of business information.^[2] The term has been noted as early as 1865,^[3] its origins are mainly attributed to a 1958 article in IBM's Journal of Research and Development ^[4] where it was called the business intelligence system.^[5][1]

BI is sometimes viewed from two positions: data-centric and process-centric.^[5] The data-centric position sees BI as transforming operational data into information using analytical tools to help drive business decisions. This approach often divorces the data from its context, however, jeopardizing its interpretability. The process-centric approach addresses this problem by emphasizing the processes from which the data originate to avoid losing context -- this is often done with diligent capture of metadata.

Some emphasize the presentation of the data: "the integration of data from disparate sources systems to optimize business usage and understanding through a user-friendly interface."^[6] Others contend that the term BI is being replaced by the term analytics.^[7] Analytics "involves the use of data, analysis, and modeling to arrive at a solution to a problem or to identify new opportunities." This is uncannily similar to BI. In general, however, analytics refers to the methods used to process data and BI refers to the over-arching process of curating information to inform business decisions.

Healthcare Business Intelligence

While business intelligence has long been used in many other industries, healthcare largely did not adopt these techniques until the 21st century.^[8] The reasons for this are varied ^[5]. Most industries have singular management structures while healthcare has both clinical and administrative ones. Most industries have a well-defined group of customers with a relatively constrained product range, whereas healthcare serves the needs of patients, insurance companies, accrediting agencies, governmental authorities, researchers, and clinicians -- while they all might theoretically serve the patient, their paths often cross in antagonistic ways. Most industries also have discrete, reliable metrics, but healthcare metrics are defined differently by the varying customers, often difficult to capture reliably, and must attend to poorly defined patient factors.

Healthcare organizations have historically struggled to find the elusive link between the investment in information technology and improved organizational performance.* Aside from the aforementioned adoption barriers, this gap has also been driven by the use of information technology (IT) to digitize clinical workflows with inadequate attention paid to using the care delivery information to make business decisions.^[9] The strategic value of IT lies in its power to provide clinicians and leadership with direct visibility into the care delivery process.

Nevertheless, the state of modern healthcare is often characterized by its scarcity, poor quality, and financial pressures, all of which are ideal targets for BI.^[5][10] The hope of BI in healthcare is to solve these problems by using data to optimizing resource allocation, power quality improvement processes, and make strategic decisions that reduce costs and increase revenue.

Conceptual Frameworks

Mettler & Vimarlund provide a framework for BI in healthcare.^[5] For a visual aid, see Figure 1 in their paper. The framework is organized in four sections: processes, actors, information, and technology.

1. Processes are defined as sets of partially ordered and coordinated tasks that often cut across organizational units and are the target of data collection and analysis. They separate target processes into three categories: medical (e.g., diagnostics and treatment, research and teaching), business (e.g., finance, risk management), and support (e.g., human resources, logistics and supply).
2. Actors are either internal (e.g., doctor, administrator) or external (e.g., patient, government) and have a role in one or more processes. They are important stakeholders in the BI for those processes.
3. Information produced by the processes is captured as data, which is categorized as clinical, administrative, or external. Data close to the source are stored in operational data stores (ODS), which provide close to real-time monitoring. When data from multiple sources are processed and combined, they form a data warehouse. Portions of a data warehouse then may be curated for specific purposes as "data marts."^[8]
4. The data is then transformed into information using technologies (e.g., reports, expert systems, data mining).

El Morr & Ali-Hassan describes the framework of BI differently using four components: a data warehouse, business analytics, business performance management (BPM), and a user interface.^[7] The purpose of a data warehouse is as described above. The business analytics reflect the types of technologies used to process and analyze the data. The BPM separately describes how data is to be used for decision-making, such as identifying metrics and targets, monitoring frequency, and corrective actions. This framework also emphasizes a user interface as an essential component, pointing out the importance of being able to visualize the data (e.g., interactive dashboard, paper report).

Organizational Factors

Mettler & Vimarlund also define certain prerequisites for effective healthcare BI: collaboration, knowledge, trust, institutions, governance.^[5] BI systems are only as good as the organizational relationships that surround them.

1. Collaboration between different healthcare and social services in an area are important to share BI technology and data to drive decisions for the whole. This is challenging when the services are not all consolidated into one system.
2. Knowledge of the system's processes, decisional authorities, and lines of communication are becoming more crucial for making decisions. Modern management strategies often don't fit into formalized organizational charts, which makes finding this information difficult.
3. Trust, whether through contracts, competence, or goodwill, is important for sharing data and responsibility for addressing problems revealed by BI.
4. Institutional structures formalize rules for BI operations. Clear delineations of processes help make sense of large, complex systems.
5. Governance involves processes that safeguard information integrity (see Data Governance) and ensure appropriate responsive actions are taken.

It is important to note that each of these prerequisites contribute to the other. For example, collaboration builds knowledge and trust, which in turn are necessary to create institution and governance. Having an institutional structure and governance can in turn promote trust and collaboration while clarifying knowledge.

Madsen proposes an alternate set of five 'tenets' of BI, adapted into the following more modern terms by El Morr & Ali-Hassan: data quality, technology, value, change management, and leadership.^[6][7]

Regardless of the conceptual framework used, it is clear that deploying effective healthcare BI involves organizational factors in addition to the technological requirements. A BI system cannot stand on its own without organizational support.

Analytics

Analytics are the techniques for processing, analyzing, and visualizing the data to provide insights.^[7] In essence, analytics tries to take data as high up the wisdom hierarchy as possible, from data to information to knowledge to wisdom.^[11] Healthcare analytics have been described with five basic layers: business context, data, analytics (specifically data analytics), quality and performance management, and presentation.^[7]

1. The business context defines the objective and measurable goal of the analytics. Data should not be collected for its own sake for no particular reason. It should be intentional.
2. The data is defined by identifying the context, the data source and its quality, and the method for collection and storage.
3. Data analytics is defined by choosing the appropriate software and algorithms to process the data. There are four general types: descriptive, diagnostic, predictive, and prescriptive.
4. Quality and performance measurement refer to the defined metrics, their target values, and strategies for evaluation and improvement.
   • Metrics with defined targets are called indicators. They provide simplified conclusions about whether goals are being met. Indicators that serve particularly vital organizational goals and allow comparison against similar organizations or approved standards are often called "key performance indicators" (or KPIs).^[12]
5. Presentation refers to the data visualization techniques used to communicate the information in the data meaningfully to various stakeholders.

Common Technologies

Commonly used technologies in BI include the following:^[1]

• Decision Support Systems (of which clinical decision support is a subset)
• Executive Information Systems
• Online Analytical Processing (OLAP)
• Query and Online Reporting
• Business Process Monitoring
• Performance Scorecards and Dashboards
  • A number of dashboard images and interactive examples are viewable on datapine's website
  • The familiar COVID-19 tracking websites, such as the CDC COVID Data Tracker, are also good examples of dashboards.
• Data Mining
• Data Analytics

Key Performance Indicators for Healthcare

Key performance indicators (KPIs) are indicators that serve particularly vital organizational goals and allow comparison against similar organizations or approved standards. Khalifa & Khalid qualitatively analyzed key performance indicators (KPIs) at a tertiary hospital, then standardized and validated them against published research and internationally recognized benchmarks such as those from the Agency for Healthcare Research and Quality (AHRQ) and the Organization for Economic Cooperation and Development (OECD).^[12] They identified 10 categories of KPIs with a total of 58 individual KPIs, only some of which are shown here:

1. Patient Access
   • Number of Patients Referred
   • Percentage of Patients Accepted
2. Inpatient Utilization
   • Average Length of Stay
   • Average Bed Occupancy Rate
   • Mortality Rate
3. Outpatient Utilization
   • Average First Available Slots > 30 Days for New Patients
   • Percentage of No Show Patients
4. OR Utilization
   • Percentage of OR Cancellation Rate
   • OR Utilization Rate
5. ER Utilization
   • ER Waiting Time (Door to Doctor)
   • ER Treatment Time (Doctor to Disposition)
   • ER Admission Waiting Time (Boarding Time)
   • Percentage of Patients LWBS (left without being seen)
6. Generic Utilization
   • Total radiological procedures
   • Total lab procedures
7. Patient Safety
   • Unplanned Readmission within 30 Days of Discharge
   • Unplanned Transfer to any Critical Unit/OR
   • Cardiac or Respiratory Arrest
8. Infection Control
   • Blood Stream Infection
   • Catheter-related Infection
   • Surgical site infection
9. Documentation Compliance
   • Number of Deficient Records (less than 30 days)
   • Number of Delinquent Records (more than 30 days)
10. Patient Satisfaction
    • Inpatient Satisfaction Rate
    • Outpatient Satisfaction Rate

References

1. ↑ ^{1.0 1.1 1.2} Tutunea, M. F., & Rus, R. V. (2012). Business intelligence solutions for SME's. Procedia Economics and Finance, 3, 865-870.
2. ↑ Dedić, N., & Stanier, C. (2016). Measuring the success of changes to existing business intelligence solutions to improve business intelligence reporting. In Research and Practical Issues of Enterprise Information Systems: 10th IFIP WG 8.9 Working Conference, CONFENIS 2016, Vienna, Austria, December 13–14, 2016, Proceedings 10 (pp. 225-236). Springer International Publishing. https://doi.org/10.1007/978-3-319-49944-4_17
3. ↑ Devens, R. M. (1865). Cyclopaedia of Commercial and Business Anecdotes; Comprising Interesting Reminiscences and Facts, Remarkable Traits and Humors of Merchants, Traders, Bankers Etc. in All Ages and Countries. D. Appleton and Company. p. 210
4. ↑ Luhn, H. P. (1958). A business intelligence system. IBM Journal of Research and Development, 2(4), 314-319.
5. ↑ ^{5.0 5.1 5.2 5.3 5.4 5.5} Mettler, T., & Vimarlund, V. (2009). Understanding business intelligence in the context of healthcare. Health Informatics Journal, 15(3), 254-264. https://doi.org/10.1177/1460458209337446
6. ↑ ^{6.0 6.1} Madsen, L. (2012). Healthcare business intelligence: a guide to empowering successful data reporting and analytics. John Wiley & Sons.
7. ↑ ^{7.0 7.1 7.2 7.3 7.4} El Morr, C., & Ali-Hassan, H. (2019). Analytics in healthcare: a practical introduction. Springer.
8. ↑ ^{8.0 8.1} Gartner (2005). Hype Cycle for Business Intelligence and Data Warehousing.
9. ↑ Gartner (2005). Hype Cycle for Healthcare Provider Applications.
10. ↑ Cutler, D. M., & Morton, F. S. (2013). Hospitals, market share, and consolidation. JAMA, 310(18), 1964-1970. https://doi.org/10.1001/jama.2013.281675
11. ↑ Rowley, J. (2007). The wisdom hierarchy: representations of the DIKW hierarchy. Journal of Information Science, 33(2), 163–180. https://doi.org/10.1177/0165551506070706
12. ↑ ^{12.0 12.1} Khalifa, M., & Khalid, P. (2015). Developing strategic health care key performance indicators: a case study on a tertiary care hospital. Procedia Computer Science, 63, 459-466. https://doi.org/10.1016/j.procs.2015.08.368

Submitted by Amrish T. Pipalia