Advancing Healthcare through the Application of Predictive Knowledge Management

 Perot Systems' Perspective on PKM

In order to understand how Perot Systems’ perspective on PKM is unique, we must first understand what elements are necessary for true PKM capabilities, which the first part of this section addresses.  PKM requires effective data warehousing, mining, and analytic support.  In designing a data warehouse, there are clear advantages in building an approach that can be deployed across multiple organizations so that different questions can be answered utilizing the same resource.

Data mining requires the use of software tools to support the type of analysis required by various components of the healthcare industry.  While, historically, this required customized solutions, there are now available options that provide “off-the-shelf” data analytic tools, which can be used by healthcare organizations.

Finally, data warehousing and mining are insufficient without effective data analytics, which requires the skills of highly talented experts (e.g., biostatisticians, clinical informaticists, bioeconomists, and others).  Warehousing is a commodity service.  Mining is an off‑the‑shelf capability that can be purchased or leased.  Analytics is the essential tool that provides the value-add service for healthcare organizations seeking to develop their PKM capability.

Data Warehousing. Perot Systems believes a leveraged model provides a unique approach/method for collecting data from provider point-of-care clinical and administrative information systems and aggregating it into longitudinal, comparable, patient-centric records.  With a primary data source established, healthcare providers would be able to extract valuable information related to clinical quality, safety, and cost-effectiveness.  At the same time, the data can—at the option of the client—be de-identified and re-purposed for use by multiple other agencies (e.g., payers, pharmaceutical companies, device manufacturers, public health entities, government agencies, and other healthcare participants).  Such an approach would allow the system to extract valuable information at a reasonable cost from many diverse sources.

Through PKM initiatives, the Perot Systems model assists clients in collecting information from provider operational and clinical systems, such as the administrative, pharmacy, radiology, laboratory, materials management, and financial systems.  The information would then be aggregated at the patient level so that all aspects of a patient’s longitudinal experience are chronologically linked together.  After the data is cleansed and consolidated, a common vocabulary and cross-institutional patient index would be applied to enhance the value of this traditionally unorganized, noncomparable raw data.

Data transformed in this manner is of critical importance for healthcare providers focused on improving the effectiveness of their services.  This same data—de-identified and combined with similar data from other organizations—can form the foundation of a new and highly valuable source of applied health research services across multiple sectors of the healthcare market.

As an example, for the pharmaceutical industry, data warehousing and data analysis techniques can be applied to solve critical business issues, such as:

• Detecting safety signals
• Measuring outcomes
• Sizing a market
• Defending a product
• Monitoring off-label product usage
• Reacting on a real-time basis to the effectiveness of a product launch or patient recruitment

The Perot Systems PKM model is designed to be a foundational platform that in the right environment would enable numerous business and scientific applications.  Characteristics of clinical data used include the following:

• Clinically Rich—Clinical data collected from multiple clinical systems, including admissions, patient management, laboratory, radiology, and pharmacy
• Longitudinal—Data chronologically linked within and across multiple inpatient and outpatient encounters
• Real-Time—Data continuously collected from healthcare systems with an availability lag measured in hours
• Fidelity— Data captured at the point of care
• Transferable Meaning—Universal comparability, regardless of system, facility, provider, format, coding scheme, or vocabulary
• Representative—Multiple, U.S.-wide clinical sources with broad demographics for representative sampling
• De-identified—Patient identifiers excluded, consistent with HIPAA guidelines, patient privacy, and protection of data suppliers
• Secure—Encryption and security measures in place to restrict access to authorized users

The Perot Systems PKM approach could also be provided as a leveraged model across the more than 620 hospitals and 130 health plans (2006) that use its services.  The intent is to grow a large data warehouse operated, managed, and marketed by a trusted entity, but where the data is owned and controlled by the providers of the data [NOTE: This represents a critical difference from other PKM models.  The philosophy of Perot Systems is that the providers of data own the data].  Hospitals, hospital systems, and other healthcare provider organizations benefit in two ways when they provide data to the PKM model.  First, they are able to proactively respond to the increasing pressure from public and private organizations to validate improvements in quality of care and the need to be more disciplined in the way they capture and analyze operational data.  Second, healthcare providers can be compensated for the data they provide to other segments of the healthcare industry as a new source of income, if desired.

Data Mining.  The process of data mining employs the “law of large numbers.”  This data can then be used to address issues that surround decision-making where uncertainty exists.  Data mining requires the autonomous extraction of information from large amounts of data with the end result being the identification of patterns and/or relationships in the data that may be beneficial to a particular segment of the industry.  In the case of healthcare, these identified patterns and relationships could conceivably change the way in which healthcare is delivered.  Data mining is also referred to as a process for knowledge discovery. 

Data Analytics. Data analytics represents the true value-add and leveragable service in a PKM initiative.  Analytics capabilities require highly skilled workers with unusual and highly sought-after skill sets.  In general, these individuals are highly specialized and their knowledge can be leveraged across multiple clients.  In addition, the analytics knowledge, once developed, can be used across organizations for driving deeper analysis of issues and/or problems.  It is anticipated that the application of virtual approaches to staffing analytic services will also evolve over time.

Continued >