Gary Palgon, Liaison Technologies

The term “Big Data” has exploded within the healthcare industry, and it’s unlikely that it will go away anytime soon. In fact, these large sets of information have the potential to impact almost every facet of the life sciences value chain, including research and development, finance and administration, manufacturing, supply chain, marketing and sales. There is, however, a disconnect between availability of data and the ability to translate the information into actionable insight.

		Historically, business intelligence has dealt with the analysis of transactional data—order volumes, production costs, sales&mdash;but Big Data’s true value lies in the ability to evaluate interactions among data, not simply the individual transactions. By connecting information across the enterprise and throughout the healthcare ecosystem&mdash;such as sales and manufacturing&mdash;pharma and medical device organizations can support business decisions with real-time, granular-level data that not only streamlines workflows but also improves development and delivery of products.

		Because pharma operations cover all facets of the supply chain up to the actual retail point of sale, pharmaceutical organizations are uniquely positioned to optimize the use of Big Data. Even in the case of retail, pharma may not be present in the actual transaction, but regulatory mandates require pharmaceutical and medical device companies to track what is sold and to whom, most of which has not been tracked historically. This detailed consumer information can further enhance the organization’s use of data to evaluate market opportunities and refine research, product development, manufacturing and logistics functions.

		The flow of information throughout a pharma enterprise is circular, with research and clinical data supporting product development, helping to guide manufacturing processes. Marketing, sales and distribution information provide insight into provider and patient interactions, and the newest form of data—social media&mdash;enables companies to determine the sentiment, or reaction to products, directly from consumers. Analysis of consumer sentiment can then determine new opportunities for research. (Fig. 1)

		It is no surprise, then, that pharma companies have access to greater amounts of information than ever before. While Meaningful Use (MU) Stage 1 increased the number of healthcare providers that capture patient information electronically, MU Stage 2’s mandate to exchange data among multiple providers has created a treasure trove of information in the form of structured and unstructured data, all of which can be used to complete clinical trials in a shorter time frame, conduct studies related to compliance and outcomes, or move more quickly toward personalized medicine. The same increase in volume and velocity of previously unavailable data is present throughout the pharma enterprise.

		Unfortunately, much of the data is captured in silos across the healthcare community. The missing piece for many pharmaceutical organizations is a platform that gathers information from partners and internal disparate sources and enables access to other users after it has been integrated, aggregated and harmonized in a central location. This foundation is critical to providing a range of information that can help executives reap rewards such as faster product time to market, cost savings, new efficiencies and greater customer understanding.

		The ecosystem for which data must be integrated includes the entire pharma enterprise, as well as providers and payers. Not only must a large volume of data from disparate sources be integrated, but the data must be aggregated and harmonized so it can be understood in the right context by a wide range of users.

		The traditional approach to integration is a point-to-point, script-driven process. As files travel from server to server or application to application, imbedded code directs the movement of the information. This approach works well when data transfer is limited to a few localized systems and users, when data volume is manageable and when data requires little “translation” for each user to access.

		In the recent past of IT development, it was common to hear of programs that performed “extract, transform and load” (ETL) functions, which worked well for the point-to-point applications just mentioned. Today, however, the combination of vastly larger datasets and new programming technology allows the user to “come to” the data, rather than the data coming to the user. You could call it “extract, load and transform” (ELT) instead. With Big Data, you extract and load the data into the repository (e.g. Hadoop, Casandra, etc.) and then when you figure out what questions you want to answer, you transform it and perform the research. This means that you no longer need to figure out all of the questions you are going to ask in advance of the load—a huge change.

		Effectively managing Big Data, however, requires more of a “hub and spoke” approach in which data is aggregated in a central repository, whether it is in a structured or an unstructured format, and can then have different rules applied, which ultimately empowers the various parts of the organization, or the “spokes.” This centralized approach enables access to information across the enterprise, no matter the source of the original data. An additional benefit of this model is less risk of broken script that must be tracked down and repaired to enable automation of data gathering.

		Moving to this model ensures that data integration, aggregation and harmonization take place in a much more strategic manner. Manual steps related to coding rules for point-to-point transfer of data are removed in favor of a higher level operational view. Also, access to data throughout the enterprise, and among partners, optimizes the use of analytics applications which are often limited by the amount of data and format available. When access to Big Data is automated, staff time spent on analysis is more effective because staff can focus on analyzing the graphical or methodical interpretations of data rather than overseeing the acquisition of data.

		Although redesigning a company’s information technology infrastructure to optimize the use of Big Data to provide quality business intelligence makes sense, a fiscally-sound approach is necessary. The most financially feasible option is the use of a cloud-based solution as the platform to integrate systems and aggregate and harmonize the data for use in appropriate applications. A cloud-based system also provides flexibility of use for the pharma enterprise—analysis of data can take place in the cloud, or portions of data can be compartmentalized and packaged for use in different parts of the enterprise. These options not only ensure access to the right data for specific applications but also enable the pharma organization to determine how and by whom data will be used. This is especially critical in the research and development departments, where proprietary information along with patient information demand additional levels of security.

		When all silos of information are integrated and accessible throughout the organization, data that has previously been unavailable to pharma can now be analyzed alongside traditional data. Social media is one example of new data that enhances point of sale data. With social media, now, pharma marketers and researchers have a tool that not only measures what is purchased but also what consumers think about the product. Traditional focus groups or review of sales data has provided some insight into consumer reactions, but social media provides real-time, product-specific feedback that enables a company to reevaluate a product, make marketing expenditure decisions or even adjust pricing if needed.

		Social media also informs the research area by providing an overview of consumer preferences and reactions to existing products. Research and development further benefits from integrated Big Data in a number of ways, including enabling many rather than a few researchers to work together at the same time. This expanded collaboration can help to fill clinical trials faster and document outcomes more quickly, which means products go to market in a shorter time frame.

		Moving products through development and manufacturing in a timely manner contributes to the financial success of a pharmaceutical organization, but there are two other potential financial opportunities for pharma as a result of integration and analysis of Big Data. Typically, diagnoses are determined by payer data, but access to patient information, along with historical sales data, prior to the submission of claims, enables pharma to anticipate needs more quickly to ensure the right medication is available at the time it is needed. Not only does this streamline manufacturing and distribution processes but inventory costs can be better managed with stronger insight into demand.

		Although healthcare providers have been more interested in the use of Big Data to monitor patient compliance with medication therapy, pharma can also benefit from access to adherence data. For example, a patient with a heart condition may fill the first prescription but neglect to fill subsequent prescriptions. An alert from the patient’s pharmacy to the physician prompts a follow-up conversation to determine the reason for no refills, to provide additional education and to obtain resources the patient may need to purchase the medication. Not only does this process result in a better patient outcome, and lower cost of care for a collaborative care organization, but pharma benefits from more predictable product sales.

		Another advantage of centrally-integrated data for the manufacturing and logistics facets of the pharma supply chain is the ability to obtain global point-of-sale information more quickly to improve efficiency of the manufacturing process. With different countries or regions regulating slightly different formulas for medications sold in their locations, the ability to see how demand in one region of the world changes compared to others allows pharma companies to better manage the production of drugs—in terms of formula “recipe” as well as volume&mdash;to create cost efficiencies in the manufacturing and distribution processes.

		Perhaps the business units that benefit the most from the ability to aggregate and analyze multiple layers of data are marketing and sales. Access to information not previously available, such as more direct consumer preference and adherence data, provides a more robust interpretation of trends to better identify opportunities to evaluate and improve sales.

		One of the major concerns associated with integration of data from different proprietors’ datasets is delineating who owns the data. With federally mandated requirements for data privacy and security, data ownership is an evolving discussion with different opinions on the use of protected health information—identified and de-identified.

		As the conversation about data ownership continues, no organization can ignore the need for multiple layers of security, including encryption and well-defined access controls. This is not to say access should be limited—that negates the concept of having access to more data&mdash;but appropriate access must be defined. In fact, a central integration platform that applies security protocols at a global level is an effective and fiscally responsible way to protect data from outside threats as well as unauthorized internal access. Partnering with a cloud-based provider ensures access to staff expertise that is difficult to maintain internally. Not only are security protocols continually updated but all processes required for seamless interoperability among all users are maintained on an ongoing basis to avoid use of obsolete solutions that may affect the integrity of data analysis.

		Developing and maintaining a platform to enable integration and aggregation of Big Data is not the expertise of pharma, but rather is the analysis of the information in order to provide actionable insight. Choosing to use an expert for the integration and harmonization of data is similar to choosing between changing the oil in your car by yourself or taking the car to an oil change center. Because changing oil is all the center’s staff does, they can perform the task better, faster and less expensively than you might when the cost of your time and supplies are included.

		The technological transformation of the healthcare industry has contributed to the dramatic increase in data that is available to the pharmaceutical industry. The industry recognizes the need to break through silos of information to enable enterprise-wide access to information but what impact can Big Data have if there’s no way to assemble and condition structured and unstructured information into actionable insight? Adopting a cloud-based, centralized integration platform provides a solution to manage Big Data to produce a more robust analysis of business trends and opportunities to support business decisions that ensure long-term financial and operational success.

Gary Palgon is the vice president of healthcare solutions for Liaison Technologies, which provides healthcare organizations with innovative solutions to complex integration and data management needs. He can be reached at gpalgon@liaison.com or on Twitter at @GaryPalgon.

Articles

Setting a foundation for long-term success: How the pharmaceutical industry can transform Big Data into actionable insight

Getting value from Big Data

	It is estimated that bringing a new drug to market costs $1.3 billion. In most cases, a pharmaceutical company has only about five years to recoup that cost and turn a profit. Accelerating time to market (TTM), therefore, is vital to extending the sales window and ultimately enhancing a company’s bottom line.

	Realizing faster TTM and quicker response to shifting market trends requires a dynamic and responsive relationship with data. Access to both point-of-sale and point-of-service data, for instance, can ease critical business decisions. By understanding patient adherence—which drugs are being prescribed and being used as prescribed&mdash;pharmaceutical companies can seek and capitalize on emerging opportunities for rebranding, repositioning, and new product development.

	Current clinical trial processes often contribute significantly to the time delays and high price tags associated with bringing drugs to market. Identifying qualifying clinical trial participants, for instance, traditionally is done one partnership at a time, and reconciling vast quantities of clinical trial data from disparate sources also usually relies on labor-intensive processes.

	By interacting directly with patient health information, however, pharmaceutical companies can more easily and more efficiently identify possible patients for clinical trials, thereby reducing the time and expense involved in patient recruitment. Similarly, pulling together data from various partners in a clinical trial and harmonizing it faster allows companies to analyze and respond more quickly.

	One obstacle preventing pharmaceutical companies from having a dynamic relationship with data is the diversity of data sources and formats that do not naturally harmonize with one another. The data are out there. The IT challenge is to bring it all together, and then derive from it actionable intelligence that meets business management needs. For example:

		Sales data from around the world are captured in different formats and different contexts, so it needs to be harmonized in order to accurately evaluate global sales. (For instance, a drug in one country may have a slightly different name in another country, or a drug volume in European markets might be expressed in metric terms vs. US standard terms.)

		Clinical trial information is gathered from multiple sources and needs to be assembled for analysis.

		In-house IT departments need to flow data for multiple drug trials to be administered simultaneously.

	Companies generally lack the immense infrastructure and expertise necessary to compile diverse data and achieve meaning from it. Yet those that can leverage sales data throughout the global marketplace, as well as patient health and clinical trial information from disparate sources, are better positioned to identify market trends and new operational opportunities.

	A cloud-based delivery model is uniquely suited to the challenges facing pharmaceutical companies. In part, this is because cloud-based technology provides leverage, scalability and technology refresh intervals that cannot be matched with traditional software models.

	What’s important to understand, however, is that cloud technology can support more than just infrastructure needs; it can also be used to solve business problems. Cloud technology supports three distinct layers of service that build upon each other:

		Infrastructure as a Service (IaaS). This bottom layer is what most people think of when referring to “the cloud.” Companies pay for space and usage—for hardware, Internet Protocol (IP) addresses, CPU space, memory and firewalls, for example&mdash;and expand or reduce these as necessary.

		Platform as a Service (PaaS). Although less familiar to most, PaaS is critical to any successful interaction with the cloud. This level provides the tools, capabilities and—more importantly&mdash;services that solve problems. PaaS includes things such as integration as a service, data as a service and information as a service. Integration Platform as a Service (iPaaS), for example, is an integration specialty that involves the ability to perform integration in the cloud.

		Software as a Service (SaaS). This level of the cloud hosts the applications—such as dashboards and charting programs&mdash;that help visualize the services provided in the cloud. Familiar examples of SaaS applications include human resources packages, sales automation dashboards and stock option apps.

	Cloud-based technology operates in the background of a company’s IT systems, allowing the organization to continue using its current systems, while gaining more functionality and scalability. Consequently, cloud-based solutions can be a more cost-effective proposition for data management and optimization.

	Data security and privacy are important considerations when using the cloud, especially when health information is involved. Vendors leveraging cloud-based technology are fully capable of ensuring data privacy and security by focusing on access control, security of stored data, and security of data in transit.

	Access control involves guaranteeing that only the right users have access to specified data. Features not only confirm the identity of a user, but also control what information each particular user can see and interact with. At the higher PaaS and SaaS levels of cloud use, such controls also can address the cloud-based service provider to ensure it only accesses what it should.

	Every level of the cloud, from IaaS through SaaS, requires access control. This activity becomes more complex as the service levels build on each other. To ensure appropriate levels of control, pharmaceutical companies should work with cloud-based providers to develop service level agreements (SLAs) that govern the specific access requirements present in the relationship. These agreements hold the cloud-based vendor accountable and better support effective control.

	The details involved with protecting stored data or data in transit should also be covered in SLAs. Protecting data during these intervals can be done in many ways, including through encryption, key management or tokenization. While encryption and key management have been around for awhile, tokenization is an emerging security model used successfully in many industries to reduce the risk of security breaches and protect confidential information.

	More specifically, tokenization is the process of replacing sensitive data with surrogates—called tokens&mdash;which are words, phrases, symbols or other elements that retain essential information about a piece of data without compromising its security. Out of context, a token is meaningless. And only certain individuals or systems can redeem the token for the original value. Consequently, tokens can be used safely by any cloud-based application or database without risk of exposing sensitive information.

	Although the cloud itself provides a flexible and scalable environment for meeting operational goals, it is the services that take place in the cloud that truly help an organization improve its data management processes. Pharmaceutical companies can obtain useful business intelligence effectively in the cloud through four distinct tasks:

 As the first step in the series of tasks, “connecting” refers to the ability to join disparate IT technologies to send information back and forth. There are different ways to exchange information (e.g., FTP, HTTP, or Secure SHell [SSH] protocol), but using common protocols can enable connectivity. Achieving such connectivity typically requires large capital expenditures for hardware, software, and other infrastructure—a costly endeavor in terms of money and resources. By contrast, using secure cloud-based technology to connect distant entities or dissimilar systems removes the burden of creating a technology infrastructure, and allows more efficient and inexpensive connectivity.

 Integration is tightly interwoven with connectivity. Instead of linking the physical IT “plumbing” or “pipes,” though, integration links the disparate data themselves by making sure they are ready for exchange by mapping and aligning interfaces and data. Data currently exist in a vast number of formats that must be made to flow seamlessly among different “standards” or IT systems. This can be challenging with many data partners and point-to-point connections, but there are different ways to support effective integration. For example, service-oriented architecture (SOA) interfaces might be used to allow multiple components to communicate with one another.

 If more data are being compiled for analysis by pharmaceutical companies, the data need to somehow be aggregated. This involves assembling data from various sources in a method that supports large volumes of information, referred to as “big data,” as well as to enable real-time access. Data aggregation, along with harmonization, provides the ability for analytics to identify operational trends.

 This is the most important step toward true data interoperability. It involves first “transforming” data when necessary into a common, usable format, “normalizing” it to minimize redundancy and dependency, and then “translating” it to equitable semantic definitions. Translation makes it understood, for instance, that “surname” and “last name” used in different fields on different forms mean the same thing. This is the stage at which similarities and differences are reconciled, and actionable information is ready to be obtained.

Case study: Integration Platform as a Service (iPaaS)

	The end goal of all these steps to uncover actionable data, of course, is to achieve business results. This is a look at how one large global pharmaceutical company used cloud-based services to realize several distinct objectives:

 — the corporation wanted to reduce the number of days required to flow clinical trial data used for analysis

 — it wanted to address its growing volumes of data and connections as it increased the number of clinical trials performed

 — it needed to map key clinical trial data to standard formats

 — it wanted to efficiently use offshore resources to accelerate and streamline processes

 — it wanted to improve data integrity, reliability and completeness of information in a clinical data repository

	Prior to tapping cloud services, this organization was particularly challenged by the harmonization process—readying the different types of clinical data to be analyzed by various systems. (See Figure 1: Clinical Trials Before)

	Through the cloud, the company was able to more easily accommodate multiple data standards and formats. It gathered data from multiple types of partners: clinical trial laboratories; the Food and Drug Administration (FDA); clinical research organizations (CROs); joint ventures; and electronic data capture applications (EDC). In the cloud, data were harmonized into a single, usable format that leveraged multiple integration methods and protocols, including APIs, Web Services, SDTM, EDI, and more. (See Figure 2: Clinical Trials After)

	As a result of this project, the company was able to accelerate data flow processes, and increase its capacity to handle simultaneous clinical trials with large numbers of participants. This helped the organization reduce costs, while enabling scalability for integration and data transformation. Ultimately, the company was able to speed drugs to market, and position itself to more easily assimilate new business.

	Pharmaceutical organizations should not overlook the operational utility of the cloud. The capability exists to perform complex data integration and management functions in the PaaS layer.

	Integration and data management performed as a service in the cloud creates several important efficiencies. First, it leverages existing infrastructure and expertise to obtain useful data quickly, which in turn helps speed drugs to market. Second, it allows companies to pay for data management as a service rather than as a technology installation, turning a substantial capital outlay into a more manageable operational expense. Finally, taking advantage of integration and data management as services in the cloud frees existing IT staff to focus on other mission-critical aspects of the data management effort, such as creating applications that further advance operational goals.

	Cloud-based services are not just for accessing additional CPU space or extra servers anymore. They can fundamentally change the way an organization approaches and solves complex data optimization problems, leading to greater efficiencies and long-term financial growth.

Gary Palgon is the Vice President of Healthcare Solutions for Liaison Technologies, which provides healthcare and pharmaceutical organizations with innovative solutions to complex integration and data management needs. He can be reached at gpalgon@liaison.com.

Cloud computing can be more than simply accessing databases and computing resources; it can be a pathway to business and research decisions