Embedding Expertise: How can Academics Contribute to a Learning Health System? Part 3

In the previous post, I showed van der Vegt et al.’s SALIENT framework for end-to-end AI implementation, and one reason SALIENT resonates with me is the multidisciplinary expertise represented (boxes superimposed by me below) and the explicit alignment with regulations and the organization where the innovation will be implemented. These are principles I saw put into practice at my alma mater, the University of Utah, which has 60 years’ experience developing and implementing digital innovation, beginning with LDS Hospital, Intermountain Healthcare.

Frameworks like SALIENT lay out what needs to happen, but the question I struggle with is how to operationalize the frameworks. The Academic LHS suggests one important ingredient: capitalizing on embedded academic expertise in health system sciences. Some of you know of Scott Evans, who created the Antibiotic Assistant--a clinical decision support system implemented in the 1990’s that worked behind the scenes  to analyze data from the EMR and suggest the need for anti-infective therapy for a patient who may have gotten an infection while in the hospital. The system decreased length of stay, decreased hospital costs, and resulted in fewer adverse events. Although Scott was not a clinician, he spent his career embedded in the critical care unit at Intermountain and repeatedly credited this embedding to his success in translation of innovation. 

I stumbled across a 2003 thesis by Hu Xiao Xia of University of New South Wales (Jeffrey Braithwaite was his supervisor) titled Improving quality while reducing cost : an innovation journey whose objective was to “explore how Intermountain Health Care…innovates in implementing TQM [total quality management] organisation-wide to improve and manage clinical quality.” The main finding was that “innovation implementation at IHC was a journey, not a destination.” The journey has occurred over decades, involving many key individuals, key decisions and events, affected by internal context and the external environment (p. 66). That can’t be done from university offices down the street --it requires embedded collaboration.

Many current embedded collaborations inspire me:

1. Future practice at NYU Langone “stand[s] in the gap between academic research and digital practitioners, publishing and participating in thought leadership that is both scholarly and operational.” 

1. The University of Sydney and Sydney Kids have launched a collaborative Learning Health Initiative--they are building a data and analytics infrastructure to support learning health system projects.

1. The Shah Lab at Stanford has developed an evidence generation pipeline for AI, in their case, that goes from the lab to translation: The Program for AI in Healthcare conducts the research which the Applied Data Science team puts into practice.

1. The Vanderbilt University Medical Center LHS Platform is integrated into the health care system.  The Platform fully embeds pragmatic effectiveness trials into routine processes of care. 

We are putting our toes in the water by embedding two researchers in the Royal Melbourne Hospital to help evaluate the implementation of their newly lauched Digital Coordination Centre. We hope this relationship will grow as we work with an RMH fellow in the Learning Health System Academy this year on early identification of hospital-acquired complications associated with anticoagulants.

An article by Scarbrough and Kyratsis described what fosters embedding of innovation in healthcare, particularly digital innovation:

• True to the theme of this post, increased connectivity between innovators and implementers of innovations (I was surprised and happy they cited a paper I co-authored with the fabulous Andrew Balas on diffusion of innovation)

• Able to align with policy environments that truly support and reward sustainable embedding of innovation over more short-term outcomes 

• Adapting standards to be more inclusive of new and shifting forms of evidence on patient benefits that do not conform to the established format of randomized clinical trials 

• Development of new practices, such as the creation of regulatory sandboxes that connect policymakers with innovators and help to avoid pilotitis. 

Supportive infrastructure and embedded researchers are vital to an LHS, and in my text post I’ll talk about the need for system-level frameworks to align business-as-usual with LHS research.