"There is a revolution happening in big data analytics, creating demand for business users to be their own data scientist," said Brian Roy, director of platform engineering and architecture at Zephyr Health, in a statement. "In development of the Zephyr analytics platform, we found Neo4j to be the best open solution that met our needs, offering flexibility, scalability and a viable community committed to creating the building blocks that accelerate our business."
Zephyr Health uses Neo4j to enable its cloud-based Zephyr analytics platform, providing graph analytics across a diversity of data. The Zephyr analytics platform allows pharmaceutical makers, medical device manufacturers and other health care customers to discover connections across their data that can advance their R&D, clinical trials and marketing. For instance, Zephyr's engine helps pharmaceutical companies find the right doctors for a clinical trial by linking private and public data—such as specialty, geography and clinical trial history.
Zephyr Health's vision was to create a platform that would unify data in a way that could provide customers with deep insights into the dynamics of a given market. For this, Zephyr needed to combine a wealth of data from both its customers and public sources of information. It then needed to build an application capable of querying this mix of multisourced data. Two factors caused Zephyr to look beyond traditional database technologies for its solution. First was the real-time analytic component: making sense of connections across these diverse sets of data in real time. The second was the diverse and changing nature of their data, for example with data from doctor's surveys, which caused new attributes to come in regularly.
Zephyr solved this problem by storing the connected data in a Neo4j graph and exposing it to business users via its own domain query language. In a week or two of development, the Zephyr team was up and running with data on millions of physicians and hospitals. The company is now seeing its customer base and data volumes grow exponentially. The Zephyr analytics platform is running faster than expected and meeting and exceeding market demands.
"Graph databases are effective for every industry we know of—from telco and financial services, to logistics and hospitality, to online dating and health care," Eifrem said. "The success that Zephyr Health has seen with Neo4j underscores the transformative business power that results when graphs are made available to everyone."
Eifrem said there are primarily two types of organizations that are likely to use graph databases: big Web properties and commercial vendors.
He said attention paid to graph databases from larger, more established players like IBM, Oracle and Teradata sheds light on the technology, which stands in Neo's favor.
Teradata announced its Teradata Aster SQL-GR graph engine late last year. The Aster SQL-GR engine enables native processing of large-scale analytic graph queries and prebuilt graph functions and can be used for customer churn, product affinity, fraud detection and recommendation engines. For example, telecommunications providers can use graph analysis to look for high-traffic connections among selected users, which provides clues to fraudulent calling activity. Eifrem said financial services institutions, particularly investment banks, are using Neo4j to analyze fraud as well.
Eifrem said Neo4j has a leg up on bigger players like Teradata because of its maturity and robustness.
"Databases require calendar time to be robust," he said. "It takes time. These guys are all new. We've been running 24/7 production for eight years. We also have an ecosystem advantage over everyone else. We're much more popular than the others out there."
Database monitoring site DB-Engines ranks Neo4j as clearly the most popular graph database available today, Eifrem said.