Improvement science as the engine for the next generation of learning health networks.

While the evidence-based care movement has traditionally relied on randomized controlled trials (RCTs), these methods often lack the flexibility and external validity required for complex, real-world clinical settings. To overcome these issues, this paper proposes an Improvement Science Informed Learning System (ISILS) for Learning Health Networks. The ISILS operates as a continuous, self-reinforcing cycle. Its design integrates methods for identifying and overcoming unwanted variation in outcomes, determining effective practices, and rapidly implementing those successful practices across network member systems. This actor-oriented organizational structure promotes knowledge creation across diverse sites, enabling patients, clinicians, and researchers to co-produce health. Using the management of pediatric Crohn's disease as an illustrative case, this paper illustrates how the ISILS can identify superior therapeutic drug monitoring strategies and improve network-wide remission rates. Ultimately, this integrated approach ensures that best-practice care is a predictable output of a designed and active learning system.