Growing expectation around getting more business value from data in a cost-effective manner is driving change across industries. As organizations collect data of increasing volume and variety, teams of data management analysts and data scientists are also formed to put this data to good use.

A typical data science project begins with extracting profile data (customer, product, organizations, stores, suppliers) from some master data management system and loading it into an S3 or some other data warehouse. The next task usually involves spinning up and managing servers to run analytics applications. After that, the data science team will develop an application to execute the query, and use various mechanisms to bring results and updates back to the master profiles. This process is often repeated if the data changes.

The problem with this approach lies with the significant budget and resources required. Data cleanup and data movement are complex endeavors. It’s difficult to integrate and synchronize master data management and analytic systems in both directions. Moreover, the accuracy of insights generated are debatable without a clean and reliable data foundation.

Let’s take the example of a healthcare organization that’s looking for better ways to engage with their members and offer them relevant wellness programs. Their goal is to focus on their patients’ overall well-being and quality of life. Healthcare organizational objectives include lowering readmission rates, ensuring adherence and delivering quality care. Delivering such care and targeting the right member with the right programs requires deeper understanding of their members.

Members of healthcare organizations are demanding consumer-like, personalized experiences, as well as forcing healthcare providers to shift towards a patient-centric approach. This approach focuses on building long-term relationships in order to understand patients’ needs, behaviors and preferences.

Now let’s look at retail. Many retailers are taking advantage of algorithms, like “collaborative filtering” to provide purchase recommendations for their customers. This method makes automatic predictions about a customer’s interests through the collection of preferences from many users with similar profiles. This allows organizations to create offers that are more relevant, compared with generic, demographic-based segmentation. Healthcare organizations task data science teams to design similar prototypes to send personalized promotions to their members.

Data scientists often waste much of their time in data cleanup and ETL activities, which is not the best use of their talent. Moreover, the time and resources required to setup hardware, applications and prototyping are quite prohibitive. It’s not uncommon for data science teams to spend three to four months developing any working model or application.

Rather than undertake such a resource-intensive operation, organizations need a platform that provides an easier way to blend data from all internal, external and third-party sources, matches and merges the data to create clean master profiles, correlates master profiles with transaction and interaction data and uncovers relationships across all data entities (member, hospital, primary care physician, insurance plan, family members) to produce a reliable data foundation.

Today, bidirectional connectors can continuously correlate master and transaction data within advanced analytics environments and surface aggregated insights into applications, enabling data scientists to garner insights into the needs of healthcare members and deliver real-time visibility into the relationships among physicians, caregivers, therapists and family members.

Deep integration between master data management and analytics execution can provide schema synchronization and integrations to provision the data, as well as bring insights back into master profiles. It can also consume a healthcare organization’s Medicaid claim transactions and identify double-dip payments and other compliance or fraud situations quickly.

Data scientists can use commercial analytic tools to quickly prototype and analyze their data with the right visualizations, leverage collaborative filtering algorithms for intelligent recommendations and develop relevant and timely program suggestions to improve patient engagement.

Since most use cases require correlating master and transaction data, you need the ability to work with extremely large data volumes. Therefore, the data management platforms used must be able to handle data at big data scale and provide the required on-demand performance.

Such a data management strategy and architecture enables a closed-loop among master data, operational data and analytics. Applications that read the data can be developed quickly, and advanced custom analytics, such as machine learning or graph clustering for market segmentation can be applied. Insights can then be delivered back into master data profiles or operational applications for immediate execution.

Data scientists will now be able to focus on building algorithms that provide value, rather than wasting time on data cleanup. They can prototype and produce the desired recommendation engines in just a couple of weeks. Cloud deployments can allow for scaling up performance as needed, monitoring and optimizing analytic job resource usage, while on demand scalability and performance keep costs in check.

In the healthcare example we discussed, organizations can gain deeper insights into their members’ preferences, provide them with personalized experiences and offer wellness programs that best suit their needs. During enrollments, healthcare systems will be able to recommend relevant physicians for new members, reduce double-dip payments and deliver holistic care that focuses on member wellness, rather than just treatment for short-term disease.

Register or login for access to this item and much more

All Information Management content is archived after seven days.

Community members receive:
  • All recent and archived articles
  • Conference offers and updates
  • A full menu of enewsletter options
  • Web seminars, white papers, ebooks

Don't have an account? Register for Free Unlimited Access