Team Structure and Quality Improvement in Collaborative Environments Narine Manukyan Department of Computer Science University of Vermont Burlington, Vermont 05405 Narine.Manukyan@uvm.edu Margaret J. Eppstein Department of Computer Science University of Vermont Burlington, Vermont 05405 Maggie.Eppstein@uvm.edu Jeffrey D. Horbar Vermont Oxford Network & Department of Pediatrics, UVM Burlington, Vermont 05401 horbar@vtoxford.org Abstract—∞ Teams comprising diverse individuals have been shown to increase the collective creativity in jointly solving prob- lems. However, in contexts where the purpose of collaboration is knowledge diffusion in complex environments, it is not clear whether team diversity will help or hinder effective learning. For example, in organized quality improvement collaboratives (QICs), healthcare institutions exchange information on clinical practices and outcomes with the aim of improving health out- comes at their own institutions. However, what works in one hospital may not work in others with different local contexts, due to non-linear interactions among various treatments and practices. While there is limited evidence that some QICs have resulted in improved care, it is not yet clear what factors contribute to the effectiveness of these team collaborations. In this study, we use an agent-based model to study how different strategies of team formation, including team diversity and size, affect quality improvement in simulated collaborative environments. We show that, in this context, teams comprising similar individuals outperform those with more diverse teams, and that this advantage increases with the complexity of the landscape and level of noise in assessing fitness. Furthermore, we show that larger teams of relatively homogeneous agents perform better than smaller teams, and that effective learning through team collaborations is dependent on the level of knowledge of team members’ performance levels. Thus, our results suggest that groups of similar hospitals should collaborate as a single team and openly share detailed information regarding their clinical practices and outcomes. To facilitate this, we propose a virtual collaboration framework that would allow hospitals to efficiently identify potentially better practices in use at other institutions similar to theirs, without any institutions having to sacrifice the privacy of their own data. Our results may also have implications for other types of data-driven diffusive learning, such as in personalized medicine. Keywords—Collaborative learning; knowledge diffusion; qual- ity improvement; complex environments; agent-based modeling; team diversity; team learning. I. INTRODUCTION Knowledge sharing has the potential to benefit all parties involved. Much recent research has focused on studying knowledge sharing among teams of individuals collaborating to jointly solve problems [1]–[8]. In this context, diverse teams have been shown to offer some advantage. For example, in [9] the authors show that groups of diverse problem solvers This work was funded in part by the NIH Eunice Kennedy Shriver National Institute of Child Health & Human Development award 1R21HD068296. can outperform more homogeneous groups of higher-ability problem solvers, because diverse individuals bring different perspectives and heuristics that aid in the creativity of the collective intelligence. Similarly, in [10] the authors show that teams with higher numbers of newcomers perform better because newcomers add to the diversity of the team. However, when the purpose of collaboration is knowledge diffusion in complex environments rather than knowledge creation, it is not clear whether diverse teams help or hinder performance. For example, many clinicians are now participating in organized quality improvement collaboratives (QICs), in which teams from different healthcare organizations exchange infor- mation on current practices and outcomes. Nonprofit institu- tions such as the Vermont Oxford Network [11], [12] act as facilitators for these QICs. Team members identify potentially better practices in use at teammates’ institutions and then try them out in the local context of their home institutions [13], [14]. In this type of collaborative environment, the goal is for all hospitals to improve their own performance by learning from the experiences of others in their teams. However, what works in one hospital might not work in others with different local contexts, due to non-linear interactions among various treatments and practices. Indeed, it is becoming increasingly recognized that such complex interactions are not uncommon in healthcare [15]–[19]. While there is positive but limited evidence that QICs can result in improved quality of care [20], it is not clear which factors contribute to the effectiveness of teamwork in QICs [21]–[23]. The primary goal of this contribution is to study how differ- ent strategies of team formation affect quality improvement in healthcare through information sharing and learning. Wright [24] introduced the concept of visualizing biological evolution as search of a “fitness landscape”, where an individual’s position in the landscape is determined by its N heritable characteristics (“features”) and the height of the landscape at any given location corresponds to the reproductive success (“fitness” ) of the individual. The distance between individuals on the landscape corresponds to the dissimilarity in their features. In [25], the authors adopted this landscape search analogy for modeling quality improvement in healthcare. In this context, features represent clinical practices and treat- ments and fitness represents the probability of positive patient