The promise and perils of 'Big Data': focus on spondyloarthritis

This review will describe the available large-scale data sources to study spondyloarthritis (SpA), enumerate approaches to identify SpA and its disease-related manifestations and outcomes, and will outline existing and future methods to collect novel data types [e.g. patient-reported outcomes (PRO), passive data from wearables and biosensors].

In addition to traditional clinic visit-based SpA registries, newer data sources, such as health plan claims data, single and multispecialty electronic health record (EHR) based registries, patient registries and linkages between data sources, have catalyzed the breadth and depth of SpA research. Health activity tracker devices and PRO collected via PROMIS instruments have been shown to have good validity when assessed in SpA patients as compared to legacy disease-specific instruments. In certain cases, machine learning outperforms traditional methods to identify SpA and its associated manifestations in EHR and claims data, and may predict disease flare.

Although caution remains in the application of newer data sources and methods including the important need for replication, the availability of new data sources, health tracker devices and analytic methods holds great promise to catalyze SpA research.