Visualized shared mental modeling: an adapted practical framework for interdisciplinary teams solving complex problems in healthcare

Abstract

Healthcare systems and the scientific community are increasingly recognizing the need for multidisciplinary, inter-organizational perspectives and skill sets to tackle the most pressing challenges of our time to improve human health. However, as our approach to scientific inquiry is evolving, our methods of working across teams remain largely unchanged. Using a case example of a human-centered design project focused on improving care transitions during reentry to the community from incarceration, this paper details the importance and function of visualized shared mental models for multidisciplinary teams addressing complex multisystem problems. This paper contributes a practical framework for developing a stakeholder-validated, visualized shared mental model (vSMM). This framework provides a pragmatic process that enables multidisciplinary teams to develop shared understanding, navigate complexity, and coordinate effectively across systems. By describing both process and application, this paper provides a tool to enhance collaboration in health-oriented research, quality improvement, innovation, and implementation processes.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12913-025-13588-7.

Introduction

In the United States, addressing today’s complex health problems increasingly requires the collaboration of diverse stakeholders from multiple disciplines across distinct organizations and heterogeneous systems. Recognizing such demand, initiatives focused on improving human health have increasingly called for approaches that apply multidisciplinary team science and leverage diverse perspectives. For example, the Robert Wood Johnson Foundation (RWJF) offers the Systems for Action grant, which requires community organizations to lead a project conducted in partnership with academic researchers [1]. The Patient-Centered Outcomes Research Institute (PCORI) requires their funded projects to engage in meaningful and sustained partnerships with patients, communities, and stakeholders [2]. Several grant initiatives of the National Institutes of Health (NIH) have also encouraged the inclusion of community stakeholders, persons with lived experience of the targeted health problem, and intended recipients of interventions on the proposed project team or as study advisors (e.g., NIH HEAL Initiative, National Institute of Neurological Disorders and Stroke Strategic Plan) [3, 4]. While the science of team science specific to human health has grown in the U.S. over the past two decades as a new interdisciplinary field to enhance the effectiveness of multidisciplinary health research teams and optimize funding outcomes [5], the U.S. health improvement community still receives minimal formal training on how to effectively implement operational teamwork. As a result, health improvement teams often lack a common, practical framework for developing and sustaining shared understanding within increasingly diverse and frequently transient team contexts.

In this U.S. context, Soft Systems Methodology provides a flexible approach to addressing complex, ill-structured problems where predictive modeling is difficult and no simple solutions exist [6]. Such “wicked,” “messy,” “fuzzy,” or “soft” problems commonly involve multidisciplinary stakeholders with distinct—and often conflicting—perspectives and goals [7, 8]. Soft Systems Methodology aims to understand these complex problems within local contexts from multiple stakeholder vantages, using structured tools (e.g., CATWOE checklist) and visualization techniques (e.g., rich pictures, conceptual model building) to facilitate inquiry, dialogue, understanding, and action [6, 9]. Soft Systems Methodology has been adaptively applied with much variability within healthcare systems [9]. Its principles suggest the potential for adapting accessible frameworks to support multidisciplinary, inter-organizational collaboration between healthcare and non-healthcare systems.

In Systems Thinking, mental models refer to the internal representations of a system that are cognitively held by individuals and formed based on their experiences, beliefs, and comprehension of that system [10]. Mental models are often implicitly held and evolve as individuals encounter new information. A shared mental model (SMM) is the collective mutual understanding achieved among team members about the team and its goals, roles, tasks, and processes [11]. SMMs can facilitate effective teamwork among heterogeneous actors, strengthen consensus decision-making [12], and enhance team performance [13]. Decades of research across multiple disciplines, including systems science, engineering, organizational science, sociology, anthropology, and environmental science, inform SMM as a construct (Table 1 of SMM definitions by discipline). SMMs encompass a team’s shared understanding of roles, tasks, goals, and processes among a team and represent the aggregate mental representation of its individual team members mutually held within the individual-level cognitions of the team members [21]. A well-constructed SMM is the cornerstone of a high-performing team and enables effective teamwork, including smoother communication, coordinated action, more efficient decision-making, and creative innovation [12, 22] to both leverage and produce cohesive alignment of goals, perspectives, language, and comprehension across differences. While teams often use visual aids (e.g., drawing, diagrams) to support their development of SMMs, SMMs are not necessarily visual [23]. Visualization has been shown to support knowledge building and transfer, and the development and presentation of SMMs in teams [6, 24–26].

Table 1.

Example definitions of “shared mental models” by discipline

Discipline	SMM Definition	References
Design Thinking	“SMM relate to a collective comprehension between all individuals in a team concerning several aspects of teamwork such as tasks, goals, and skills” “SMM embody knowledge structures that unite individuals in a team. The shared structures of SMM lay open a path on how individuals may perform as a team in their surroundings”	Redlich et al., 2017 [11]
Management	“A means by which organisations and individuals create and share meaning, thereby enabling a common understanding and the development of knowledge” “Frameworks of value and belief systems which act as the basis for analysis of any new ideas, concepts, policies and cultural developments being considered by a team”	Davidson & Blackman, 2005 [14]
Industrial Design Engineering	“Team members’ overlapping mental representation of key elements of the team’s task environment”	Bierhals et al., 2007 [15]
Psychology	“Organized knowledge structures, or sets of concepts and the associations among them” “Mental models are defined in terms of both content knowledge and, importantly, the structure (or organization) of that content knowledge”	Ross & Allen, 2012 [16]
Human-Computer Interaction	“When individual team members’ mental models align—when they have similar understandings of their shared task and each other’s role in it—then this ‘shared’ mental model will allow the team to perform better because they will be able to more accurately predict the needs and behaviors of their teammates”	Andrews et al., 2023 [17]
Information Systems	“Knowledge similarity within a team”	Espinosa et al., 2002 [18]
System Dynamics	Integration of “partial representations of a complex situation” that are often localized by department or discipline	Vennix, 1999 [19]
Industrial and Organizational Psychology	“An organizing structure of the relationships knowledge among the task the team is engaged in and how the team members will interact”	Salas, Sims, & Burke, 2005 [20]

In the absence of formal training on the conduct of team science, it is easy to falsely presume that assembling different perspectives on a team is sufficient to produce a shared and complete understanding of the target problem [17]. Team members may be prone to overly rely on assumptions (i.e., unspoken, untested hypotheses), remain unaware of their knowledge gaps, overemphasize one perspective of the problem, and risk marginalizing the perspectives of non-dominant team members or those with divergent input [27]. Such team dynamics can negatively impact the scientific process by promoting communication inefficiencies, unproductive conflict, incomplete or false comprehension of the target problem, and, at worse, ineffective or unsustainable outcomes.

The objective of this paper is to present a framework for developing a visualized SMM on multidisciplinary health services teams solving complex health problems through inter-organizational quality initiatives, research, or program development and evaluation. We define a visualized SMM (vSMM) as the visual externalized representation of a team’s SMM that is capable of being accessed beyond an individual-level cognition. We posit that vSMMs can enhance the conduct of team science and transcend the end of Plan-Do-Study-Act or grant lifecycles to be translated to new audiences for ongoing impact. This paper begins by describing the methods for developing the framework, then the framework and processes for developing vSMMs, then presents a case example of a quality improvement project that used this framework to develop a vSMM of reentry care transition processes in New Hampshire (NH).

Methods

The vSMM framework presented here was developed through an iterative, deductive, and inductive process. It is conceptually informed by a narrative review of literature on SMM development in health and complex team environments. It is practically informed by interdisciplinarily-trained team members—experts in systems thinking, systems engineering, health services research, and human-centered design—who led the development of this framework during the conduct of a quality improvement (QI) initiative focused on improving existing reentry care transition processes between the NH state prison system and one community-based health system.

Narrative reviews have many forms and lack prescribed methods for conducting them [28, 29]. We conducted a targeted search inclusive of peer-reviewed literature, conference papers, and book chapters in PubMed and Google Scholar to identify core constructs and processes related to SMM development across disciplines, including theoretical and psychological perspectives. Additional sources were identified through author expertise (i.e., ISK, AMF) in systems thinking and shared mental modeling. Documentation from QI team meetings (e.g., meeting agendas, notes, digital whiteboards, collaborative virtual workspace) was reviewed by MFS and ISK alongside the literature to identify literature-grounded methods used by the QI team during the construction of a visualized SMM. The resulting framework synthesizes practice-based strategies with literature-based constructs for developing vSMMs within multidisciplinary teams working on complex health issues.

Framework for developing a visualized shared mental model (vSMM)

Figure 1 presents the framework for developing a visualized SMM (vSMM).

Fig. 1.

A process framework for developing a visualized shared mental model: 1) convene an interdisciplinary team, 2) facilitate dynamic and iterative learning, with an output of the visualized shared mental model. Green arrows indicate positive activities that support dynamic and iterative learning. Red indicates antagonists addressed through this framework

Convene an interdisciplinary team (Figure 1.1)

Team selection

The first step in developing a vSMM is to convene a heterogeneous team that brings together individuals with diverse expertise, perspectives, and lived experiences relevant to the target issue. While individuals may prefer to work independently or restrict collaborations within their own discipline or institution for the sake of convenience or perceived efficiency, the complexity of most real-world healthcare challenges today demands the integration of multiple viewpoints. Further, according to critical systems thinking, heterogeneity across multiple dimensions (i.e., lived experience, field, discipline, role, and system) is essential to reveal hidden assumptions, discern and critique systemic boundaries, and enhance the ethics and inclusivity of system improvement processes and outputs [30]. Such teams are better positioned to construct a more expansive, accurate understanding of the target issue at hand [19, 30, 31].

Effective team formation is neither arbitrary nor passive; it requires deliberate planning to identify and recruit individuals whose knowledge, skills, assets, and influence are relevant to the target issue [5]. Ecosystem mapping offers a practical strategy for identifying the disciplines, institutions, entities, and individual roles that either currently influence or could influence the issue if meaningfully engaged [32]. Such mapping helps project leaders to critique the boundaries of targeted systems and consider which actors should be directly represented on the team, who may be at risk of exclusion, and also where alternative mechanisms (i.e., advisory boards, consultation panels, or community town hall meetings) may provide supplemental perspectives [30]. Ecosystem mapping further facilitates purposive sampling of necessary disciplines, and snowball methods can be employed to leverage the networks and insights of recruited team members to identify additional contributors from outside 1’s immediate professional sphere [27, 32].

During team selection, project leaders should thoughtfully consider the positionality of individuals in relation to both formal power and experiential knowledge. It is essential to include individuals are directly impacted but not necessarily in positions of organizational authority (i.e., lived experience of the targeted issue). As informed by critical systems thinking, inclusion of lived perspectives is critical to challenge dominant narratives, uncover systemic blind spots, and co-produce solutions that are more acceptable, feasible, and responsive to the needs of impacted populations [30]. In tandem, project leaders should recruit members with the formal authority or influence needed to enact change within organizations or systems. Such individuals can increase the likelihood that team-generated ideas are adopted and sustained at the organizational or system level.

The psychological profiles of team members and the team as a whole are also crucial determinants of the ultimate success of vSMM development. Cognitive biases, such as confirmation bias or anchoring effects, can cause individuals to resist integrating new information that challenges their preexisting beliefs [27]. Social phenomena, including passivity, social loafing, or groupthink, may further obstruct team processes, undermining access to the diversity of thought necessary to refine and align mental models [33, 34]. These challenges highlight that simply convening a group of diverse individuals is insufficient. Team leaders must consider psychosocial dynamics from the outset, ensuring the selection of members who demonstrate openness, cognitive flexibility, and a willingness to engage in collaborative learning [35, 36]. Additionally, leadership must prioritize modeling and fostering humility, trust, psychological safety, and productive communication norms from the earliest stages of team formation [37–40].

Team grounding

Once the team is assembled, project leaders should orient team members to relevant project or funding-related requirements (e.g., required activities, constraints on project scope). To prepare to engage in dynamic, iterative learning, teams should co-create initial agreements for communication and conflict management, with the understanding that these can be revisited as the project evolves. Agreements should consider how to maximize equitable participation of team members to ensure marginalized perspectives will be voiced, considered, and integrated into team processes.

Project leaders should consider potential accommodations that may be needed for team members, especially those with lived experience and non-academic backgrounds. Project leaders work with such team members to co-develop strategies to support sustained participation. Such strategies may include funding for protected time, use of teleconference platforms to ease meeting access, recognizing experiential knowledge as expertise, offering activities or events that foster rapport building, and providing benefits that partners identify as meaningful [41]. By addressing these factors early, project leaders can establish a foundation for mutual respect and belonging to mitigate impostorism and promote collaboration.

Drawing upon Critical Systems Thinking, teams members should establish an early practice of reflexive dialogue by discussing their roles, organizational context, access to power, and positionality within the broader systems shaping the targeted issue [30]. Such discussions can reveal baseline knowledge and assumptions about the systemic and institutional factors and dynamics influencing the targeted issue (e.g., laws, policies, hierarchies, norms), including power structures that may resist change, perpetuate status quo thinking, or produce marginalization. Establishing a reflexive team practice reinforced by the project leader can begin to foster psychological safety for future knowledge exchange and constructive critique [39].

Finally, the team should consider selecting a theoretical framework to guide collective inquiry. While the project leader or academic team members may lead in identifying potential fits, the full team should participate in the final selection. This foundational work prepares the team to engage in more generative, dynamic processes for information gathering, exchange, and visualization, as discussed in the following section.

Facilitate dynamic and iterative learning (Figure 1.2)

Gather, share, and receive information unobstructedly (Figure 1.2A)

Heterogeneous teams bring with them distinct viewpoints, disciplinary paradigms, ontologies and cognitive frames, all of which must be surfaced and reconciled to build a collective understanding of the target issue [19, 31]. To facilitate this process, the team must establish structures and practices that promote the unobstructed exchange of information and ideas, ensuring that all members’ contributions are recognized, understood, and integrated [37, 42].

First, the project leader must actively encourage team members to openly share their individual mental models [38, 43], which have been shaped by past experiences, education, socioeconomic backgrounds, cultural values, and personality traits. As team members engage in interactive gathering and exchange of information, their individual models are iteratively reshaped, gradually converging towards a shared understanding of roles, tasks, and processes related to the target issue. This convergence process can be broken down into several stages of team-learning behaviors reliant upon unobstructed information sharing [21]. The first stage—construction—involves the explicit articulation of each member’s perspective and forms the initial basis for interaction [43]. Following this, collaborative construction (or co-construction) occurs, wherein team members engage in the negotiation of thoughts that have been made explicit by accepting, rejecting, or editing each other’s ideas [21]. Teams should experience constructive conflict during co-construction, whereby differences in interpretation are made known and resolved through clarifications and discussion of opposing views [21]. Targeted engagement and reconciliation of divergent viewpoints facilitates the meaningful evolution of individual mental models by allowing new, holistic ideas to emerge that were not initially available to any single team member [21, 43].

Psychological safety—the bedrock of unobstructed communication—is a shared belief that interpersonal risks are safe within the team setting. In a psychologically safe team environment, team members feel secure sharing and receiving knowledge, perspectives, uncertainties, questions, and divergent viewpoints without fear of judgment [37]. Leaders play a central role in cultivating this environment, modeling behaviors that value inclusivity and constructive discourse while framing mistakes as opportunities for collective learning [38, 40]. Leaders should also establish clear communication norms—such as structured opportunities for feedback, active solicitation of input from quieter team members, and agreed-upon protocols for managing disagreements—that reduce barriers to participation and facilitate an equitable exchange of ideas [37, 42, 44].

To prevent dominant voices or existing power dynamics from obstructing communication, the team should implement mechanisms to ensure all members have opportunities to share information [45]. This may include structured conversation formats (e.g., turn-taking, round-robin sharing) and more informal techniques like “calling in” quieter members to contribute their perspectives during discussions [31, 42]. Additionally, regularly scheduled team meetings, brainstorming sessions, and debriefings provide dedicated spaces for reflection, information exchange, and recalibration of the team’s understanding. These forums should emphasize critical dialogue, allowing team members to challenge, refine, and iteratively align their individual mental models as new information emerges [19, 37, 46].

Critically, project leaders and team members must make a concerted effort to surface and examine tacit assumptions [43]. Unspoken assumptions—whether about the problem, team members’ roles, or disciplinary paradigms—can silently shape the way individuals interpret and share information [47]. When left unexamined, assumptions can obstruct the free flow of ideas, introduce bias, misunderstandings, and missed opportunities for integration [48]. By intentionally bringing these assumptions to light, team members can critically evaluate their validity, identify potential conflicts or misalignments, and converge upon a shared understanding [27]. Externalizing what is typically implicit requires deliberate practices, such as structured dialogue and facilitated reflection [46, 49]. For example, team discussions can include questions specifically designed to elicit unstated beliefs or challenge underlying assumptions. Keen leaders can also listen for verbalized assumptions for the team to address in real-time. Such efforts allow all team members to have an opportunity to articulate their reasoning and make their thinking accessible to others [39]. Moving from the implicit to the explicit not only improves the quality of shared information but also enhances transparency and trust within the team and reinforces psychological safety, as team members feel confident that their viewpoints are understood and valued [37]. By prioritizing the externalization of assumptions, teams can reduce ambiguity in their communications (Fig. 1a-c), promote unobstructed information-sharing that accurately reflects the true complexities of the problem at hand, and align their mental models more effectively [49].

Unobstructed information gathering, sharing, and receiving should be an ongoing, adaptive process that facilitates the evolution of the team’s understanding of the target issue. Such information sharing allows teams to openly discuss and negotiate a shared conceptual lexicon of the target issue (Fig. 1a-c), which clarifies how key concepts relate to one another rather than requiring uniform application of specific terminology. It supports mutual understanding across diverse team members by respecting sociocultural and disciplinary differences in language, while helping members recognize when they are referring to similar or overlapping ideas. Unobstructed information sharing simultaneously helps to inform the co-construction of explicit visual representations of implicit ideas about the target issue (Fig. 1a-b). Regularly revisiting, updating, and renegotiating information and individual interpretations of the information helps to ensure that the team’s emerging vSMM (Fig. 1, vSMM output) remains dynamic, accurate, and responsive to new insights [19, 37].

Bound information into a visualization (Figure 1.2B)

By externalizing and concretizing abstract concepts, boundary objects make it easier for team members with unobstructed communication (Fig. 1a-b) to negotiate meanings, understand differences, and align around a shared conceptual lexicon (Fig. 1b-c), thus facilitating the iterative emergence of the vSMM [50]. A boundary object is any artifact, document, term, or concept that serves as a point of reference, enabling people with diverse perspectives to collaborate effectively without full alignment of individual mental models [51]. These objects are flexible enough to adapt to the specific needs and viewpoints of different stakeholders, yet robust enough to maintain a common identity across these perspectives. Boundary objects act as mediators in the collaborative process, bridging the gap between individual mental models and the collective mental model that the team is striving to achieve by facilitating knowledge exchange, shared meaning-making, and collective learning on teams [52].

Research has shown that the effectiveness of boundary objects in fostering SMMs depends on their capacity to be both plastic and resilient; they must be sufficiently interpretable to accommodate the specific needs of various team members while maintaining enough consistency to allow everyone to remain on the same page. Carlile argues that boundary objects are particularly valuable in complex, high-stakes environments (i.e., healthcare) where the integration of diverse knowledge is critical for success [22]. In such settings, boundary objects serve as common reference points that help teams efficiently align their individual mental models, improve coordination, reduce misunderstandings, and enhance overall team performance. Beyond facilitating communication (Fig. 1a-b), boundary objects directly help to construct shared knowledge and shared conceptual lexicon (Fig. 1b-c). By providing a tangible focus for discussion, boundary objects enable team members to externalize their thought processes, which can then be collectively critiqued, refined, and integrated into the team’s vSMM [53]. For example, in multidisciplinary teams, where members may have vastly different terminologies, methods, and conceptual frameworks, boundary objects such as diagrams, prototypes, or standardized forms can help translate and align these diverse elements into a coherent shared understanding [22].

Visualization of boundary objects through tools such as diagrams, flowcharts, and models improves upon non-visual alternatives (e.g., shared text-based documents like meeting minutes or standard operating procedures [54]). These improvements are achieved by providing a living external representation of the team’s complex ideas, processes, and relationships that can be easily understood and interpreted by all team members, despite differences in disciplinary backgrounds [55]. As such, visual tools enable team members to collaborate without necessarily having to fully understand each other’s specialized knowledge or vocabulary while making abstract concepts or key concepts more concrete and equitably accessible [51]. This helps to reduce cognitive load and increases the information processing capacity of the group by externalizing information that would otherwise be held and processed internally [19].

In fields such as systems science and organizational behavior, visual tools like causal loop diagrams, network maps and systems modeling (e.g., SysML), can help teams see their concepts and analyze complex systems, identify interdependencies, and predict the potential outcomes of different actions [19]. By externalizing these abstract ideas, visualizations facilitate team engagement in collective problem-solving and build a shared understanding that guides their actions. Such simplification of complex information is crucial in high-stakes environments where quick, accurate decision-making is essential. For example, in healthcare teams, visual tools like clinical pathways or decision trees can help ensure that all team members share the same understanding of patient care processes, which can improve coordination and reduce the likelihood of errors [24].

Visualization also enhances the iterative development of SMMs by enabling teams to continuously refine their collective understanding (Fig. 1c-b) and track the evolution of their vSMM. As team members interact with visualized boundary objects, they can more readily identify gaps in their knowledge, challenge assumptions, and propose modifications that lead to a more accurate vSMM [19]. This dynamic process of feedback and adjustment supports teams working on complex tasks where conditions and requirements may change over time. This dynamic process of feedback also strengthens trust and unobstructive sharing (Fig. 1b-a). The process of creating and negotiating visual representations may also foster a sense of co-ownership among team members, which can reinforce commitment to the team’s goals and increase willingness to engage in collaborative problem-solving (Fig. 1b-a) [26, 56].

Translate and combine information into a shared conceptual lexicon (Figure 1.2C)

Equally important is the team process of negotiating and producing a shared conceptual lexicon to ensure mutual comprehension of information and ideas being discussed, as well as an understanding of how key concepts relate to one another without flattening or erasing sociocultural and disciplinary differences in language. In multidisciplinary teams, members often bring field-specific jargon and conceptual frameworks shaped by their individual disciplines and lived experiences [27]. While team members may assume they are using words or concepts with shared meaning, these assumptions are often tacit and can hide misalignments. To overcome these barriers, teams must iteratively engage in clarification and consensus-building processes to co-develop a shared vocabulary that transcends any single discipline [22]. This shared vocabulary enables collaborative meaning-making by reducing ambiguity and supporting the integration of diverse inputs into the team’s vSMM [37].

The process of developing a shared conceptual lexicon is not merely about agreeing on terminology; it also involves addressing deeper layers of conceptual alignment, as described in ontological frameworks [57]. Without careful attention to how terms and concepts map to shared understanding, teams risk creating superficial agreements that fail to resolve fundamental differences in meaning-making. Schoonenberg and colleagues emphasize four critical ontological properties—soundness, completeness, lucidity, and laconicity—to ensure that the language expressed represents the concepts it is intended to convey [57]. These properties, while initially framed for systems modeling, offer a useful lens for understanding how teams develop shared conceptual lexicons.

Soundness ensures that all elements of the shared vocabulary correspond to real concepts within the team’s mental model. For example, if a team uses the term “patient engagement”, the concept must be grounded in shared, explicit definitions that reflect the domain it describes. The absence of soundness creates excess or irrelevant terminology (i.e., construct excess), which can obscure the team’s conceptual clarity. Completeness ensures that all critical concepts within the team’s mental model are represented within the shared conceptual lexicon. Without completeness, essential ideas or processes may remain unarticulated (i.e., construct deficit), leaving critical gaps in understanding and producing mental models that do not represent reality. Lucidity ensures that each term within the shared vocabulary has a single, unambiguous meaning, whereas the absence of lucidity leads to overloaded terms that represent multiple concepts simultaneously (i.e., construct overload). For example, consider the term “access” in a health improvement context. One team member might interpret “access” as physical proximity to care facilities, while another may define it as the availability of insurance coverage. Despite using the same term, the lack of lucidity obscures these conceptual distinctions, creating ambiguity and undermining mutual understanding of communicated concepts. Finally, laconicity ensures that no concept is redundantly represented, avoiding construct redundancy, or the unnecessary overlap of concepts that can cause confusion and fragment the team’s mental model [58].

The absence of one or more of these ontological properties limits the effectiveness of the shared conceptual lexicon and, by extension, the team’s ability to integrate information and collaborate meaningfully. For example, a team may develop a shared conceptual lexicon that is both sound and complete—accurately representing all relevant concepts—but still struggle because key terms lack lucidity; this team may unknowingly use overloaded terms with different implicit meanings, resulting in ambiguity that can impede progress. To address these challenges, teams must make their assumptions explicit and engage in curious, transparent, and patient dialogue to clarify and align meanings. Psychological safety, as discussed in Section “Gather, share, and receive information unobstructedly (Figure 1.2A)”, is a necessary prerequisite for this work, as team members must feel safe to freely share, test, and negotiate their interpretations without fear of judgment (Fig. 1a-c) [37]. Successful engagement in the negotiation of a shared conceptual lexicon can increase psychological safety and trust (Fig. 1c-a). Similarly, boundary objects (Sect. “Bound information into a visualization (Figure 1.2B)”) such as diagrams, models, or other external visualizations can serve as shared scaffolds for teams to iteratively organize and refine their conceptual lexicon (Fig. 1b-c [55]).

Shared mental odel output: externalization and preservation of the shared mental model (Figure 1.3)

The product of the above iterative, dynamic process is the visualized preservation and codification of the team’s SMM. The final working version of the vSMM serves as a cross-sectional formalized artifact that externalizes the team’s collective understanding of the target problem, its interdependencies, and—depending on scope of the project goals—the pathways toward the intended solution. A major benefit of an externalized, visual representations of team SMMs is its resistance to the knowledge degradation that occurs with internally held concepts. The vSMM persists when teams inevitably disperse at the end of a project.

In their simplest, most accessible form, visual representations of SMMs can be hand-drawn [24], such as with pen and paper, or on whiteboards. Hand-drawn sketches have the advantage of requiring few resources and allowing teams to quickly begin drafting their mental models for negotiation [59]. However, as hand-drawn visuals can be less mutable and difficult to disseminate, teams should consider transferring visual representations to technology-based platforms, or using such platforms entirely for mental model construction. Technology-based tools enhance the accessibility, mutability, and dissemination of visualizations. Free, easy-to-access options include shared document platforms (e.g., Google Docs, Microsoft Word Online) while paid alternatives like Zoom whiteboards and Miro boards allow for real-time collaboration and asynchronous refinement. Tablet-based tools, such as GoodNotes, Notability, or No Magic Cameo Systems Modeler, can also support sketch-based model building and live screensharing during team meetings. Teams working on complex systems should consider collaborating with a systems engineer to employ systems modeling tools, such as SysML (Systems Modeling Language), which enable the creation of interactive visual representations using a systems modeling language that mirrors the structure of the English language with a subject+verb+operand (e.g., web-based dashboards, dynamic flowcharts, simulations). SysML diagrams provide a structured approach to capturing relationships, behaviors, and hierarchies in complex systems, allowing teams to visualize the interdependencies and dynamicity of their processes in user-friendly and iterative ways. By depicting complex information in a format that supports clear communication [56], interactive models enable teams and their stakeholders to explore scenarios, test assumptions, and view the cascading effects of decisions within the system.

While there is limited empirical research on effective stakeholder validation methods for SMMs [60, 61], teams should consider using an inclusive, consensus-based validation process that leverages multiple methods to determine when they have arrived at the final working version of their vSMM [30]. During internal and external validation, teams can consider using surveys or interview methods to assess alignment between the team’s vSMM and real-world conditions and document feedback. Internally, team members can collectively review the vSMM to ensure accuracy and maximal representation of team perspectives. Validation methods may include iterative team discussions, structured walkthroughs, or exercises focused on identifying gaps, challenging assumptions, and clarifying ambiguities [62, 63]. Teams can also present the vSMM to stakeholders external to the team for structured feedback and validation. This external review serves as a check on whether the vSMM is sufficiently interpretable, robust, and actionable for audiences outside the team [64]. Teams can also convert their vSMMs into interactive websites or platforms to enable facilitate dissemination to larger audiences of stakeholders for broader feedback. Upon team and stakeholder validation and integration of feedback, the final visual representation can be considered a validated, cross-sectional vSMM—a tangible, shareable artifact that reflects the team’s collective understanding of a target issue at a specific point in time. While the future applicability of a vSMM may be limited as the issue or context evolves, vSMMs can serve as a foundation for future iterations, ongoing collaboration, decision-making, and dissemination of ideas, as well as a reference point to assess the evolution of perspectives and systems relevant to an issue [55].

Case example: shared mental modeling of reentry care transitions

In this section, we describe an application of shared mental modeling to a complex health problem—reentry care transitions among adults exiting the state prison system in NH. As of 2023, 1.9 million people were incarcerated in United States prisons and jails [65]. Incarceration disrupts access to essential resources for health, social, and economic well-being and exposes individuals to health-harming conditions that increase the risk for trauma, medical and psychiatric illness, social isolation, economic adversity, and premature death [66]. Chiefly, federal policies have historically restricted access to community-based healthcare systems—whose care standards are accountable to mandatory regulatory oversight—through policies like the Medicaid Inmate Exclusion Policy, which bans Medicaid coverage during incarceration [67, 68]. Underfunded and lacking mandatory healthcare standards, carceral facilities struggle to deliver quality care and facilitate reentry care coordination for an incarcerated population that experiences disproportionately higher rates of chronic medical conditions (1.4 times higher in state prisons [69]) and behavioral health conditions ( > 2 times higher in state prisons [70]; than the general population. In the absence of continuous healthcare financing, mandated quality standards, and clear guidelines to support transitions of care, reentry from prisons and jails to the community poses significant risks of worsening morbidity and death. The significant risk of premature death upon release from incarceration has been consistently documented both in the United States [71–73] and internationally [74]. A recent survival analysis of a national U.S. cohort found mortality risk post-release from incarceration was 1.4 times higher for all-causes and 3 times higher for overdose deaths compared to the general population [75]). In community-specific analyses focused on the first weeks post-release, all-cause mortality has been up to 12.7 times greater and overdose mortality 40–129 times greater than the general population [71, 72]. While NH has not calculated mortality risk among its incarcerated and reentering populations, preliminary qualitative data from reentering adults in NH mirror the national narrative of high morbidity and mortality risk amidst a lack of formal coordination processes to link adults with community-based healthcare during reentry.

In recent years, federal, state, and local policymakers increasingly acknowledged the public health crisis imposed by U.S. mass incarceration, including disproportionate health-harming impacts on populations with substance use disorder (SUD) and mental illness and elevated mortality risks during the reentry period. The SUPPORT Act (Substance Use Disorder Prevention that Promotes Opioid Recovery and Treatment for Patients and Communities Act), enacted in 2018, aims to improve SUD prevention, treatment, and recovery efforts while enhancing Medicaid’s role in addressing SUD. Medicaid Reentry Sect. 1115 waivers—a key provision of the act—focus on improving care coordination, care continuity, and therefore mortality risk during reentry by providing up to 90 days of Medicaid coverage before release from incarceration [67]. Waivers prioritize cover case management (including peer support), medication treatment for SUD, and a 30-day supply of all prescribed medications at release [67]. As of February 2025, the Centers for Medicare and Medicaid Services has approved 18 states for 5-year reentry Sect. 1115 waiver demonstrations, effectively—though temporarily—waiving the 1965 federal ban on Medicaid services during incarceration for populations with SUD and serious mental illness in these states (Centers for Medicare & Medicaid Services, 2025). In 2022, the NH Department of Health and Human Services (DHHS) developed and submitted the state’s application for the Reentry Sect. 1115 waiver. Concurrently, the project leader began to pursue funding to examine and improve reentry care transition processes between the NH state prison system (Department of Corrections, DOC), community healthcare, and social services in NH. Herein, the authors illustrate the development of a vSMM of existing reentry care transition processes in NH among a diverse, multidisciplinary team drawn from distinct health, social, and legal systems with no prior experience working together within or across their organizations.

The goal of the Returning Home quality improvement (QI) project was to improve the reentry care transition process for adults released from the NH state prisons. This goal was established by the project leader in consultation with a community advocate with lived experience of incarceration and the NH DOC Commissioner, based on preliminary needs identified during the project leader’s active qualitative study of reentry experiences with reentering NH adults. During the proposal phase, the project leader selected the Dynamic Biopsychosocial Model as the initial theoretical foundation of the project’s inquiry [76]. The project was funded through the Susan and Richard Levy Healthcare Delivery Incubator at Dartmouth College and Dartmouth Health as one of 4 selected projects in the 2022–2023 cohort. Selected projects received staff support and funding to follow a human-centered design process to develop a rapid, sustainable, scalable, and transformational healthcare redesign [77]. The Dartmouth Health Institutional Review Board reviewed and approved this project as non-human subjects research consistent with quality improvement (STUDY 00019204).

Convene an interdisciplinary team (Figure 1.1)

Team selection

The project leader used a snowball method to assemble a multidisciplinary team and project advisors representative of heterogeneous lived perspectives, skill sets, and systems interfacing with reentry processes in NH. Team participants and project advisors were recruited from September to December 2022. The project leader leveraged established professional relationships to directly recruit team members with and without lived experience of incarceration, as well as to identify second-degree contacts through her existing professional network. She connected with second-degree contacts by strategically requesting formal introductions from mutually known colleagues. She also requested introductions from known contacts in leadership positions to build connections and credibility among potential team members belonging to systems external to hers (e.g., NH DOC Commissioner introductions to administrative leaders in other NH DOC divisions; Dartmouth Health Vice President of Population Health introduction to administrators in other health systems). She had existing professional relationships with two community-based colleagues with lived experience of incarceration and leveraged their professional connections to identify an additional team member with lived experience. All team members with lived experience worked in non-research fields. Additionally, she targeted specific expertise (e.g., systems engineering, telehealth operations) by searching institutional faculty databases and inquiring within her professional network to identify individuals who could facilitate further introductions. Funding staff reviewed the team composition and recommended additional contacts with underrepresented expertise, skills, and perspectives that they perceived would benefit the project (e.g., qualitative researcher, medical students).

As part of the selection process, the project leader met with each contact individually to understand their baseline understanding of the intersection of reentry and health. She assessed their role and proximity to NH reentry processes—as in, directly engaged, indirectly involved, or having relevant knowledge without engagement—and determined the applicability of their expertise and skills in improving reentry care transition processes. During each meeting, she also took note of contacts’ communication style, the nature and level of enthusiasm and commitment expressed, and rationales for participating to begin considering how to integrate diverse dispositions. The final assembled team attained diversity across multiple dimensions, including field, discipline, role, and systems (Table 2).

Table 2.

Team participant characteristics (N = 14)

Characteristics	Total N (%)
Background
Person with lived experience	3 (21%)
Student	2 (14%)
Healthcare Provider	5 (36%)
Healthcare staff (non-clinical)	1 (7%)
Academic faculty	3 (21%)
Community representative	1 (7%)
Other	2 (14%)
Gender
Man	5 (36%)
Woman	9 (64%)
Age
20–29	3 (21%)
30–39	6 (43%)
40–49	4 (29%)
50–59	1 (7%)
Education
High School	2 (14%)
Bachelor’s Degree	1 (7%)
Master’s Degree	5 (36%)
Doctoral Degree	5 (36%)
Race
Asian	3 (21%)
Black or African American	3 (21%)
White	8 (57%)
Other	1 (7%)
Ethnicity
Hispanic or Latino	1 (7%)
Typical Meeting Attendance
Less than or equal to 1 time/month	2 (14%)
2–3 times/month	1 (7%)
Weekly	11 (79%)

Note: *Denominator exceeds N = 14 as participants have multiple backgrounds

In addition to the team, the project leader selected project advisors with direct and potential involvement in reentry care transition processes. Her primary project advisor (Dartmouth Health’s Vice President of Population Health) was recommended by the funding staff due to her focus on population health, her broad professional network, inclusive of health systems and community-based organizations across the state of NH, and her expertise in community-academic health partnerships. The project leader used her own professional network to identify a Program Director of an established reentry care transition clinic located in another New England state to serve as a secondary advisor. The project leader additionally identified and invited three community-based organizations focused on addressing social determinants of health—a recovery community organization, a SUD treatment navigation organization, and a health insurance navigation organization—to provide feedback and accountability during the project year.

Team grounding

The project year commenced in January 2023 with a 4.5 hour teleconference-based orientation facilitated by the Levy Incubator funding staff to welcome all funded teams. The orientation introduced teams to the human-centered design framework and provided foundational didactics on human-centered design approaches, the keys to successful health services innovation, and translational lessons drawn from entrepreneurship. The orientation additionally focused on team development, using an interactive workshop to facilitate team member introductions, icebreakers, cooperative agreement on the conditions for psychological safety, and an open discussion of individual work styles and preferences. Team members were encouraged to reflect on their preferred roles and individual strengths related to people-focused, process-focused, or product-focused tasks. These preferences were visually displayed on a whiteboard and served as the basis for strategizing how to leverage and complement each other’s strengths effectively.

During initial team meetings, members shared insights about their organizational contexts, their specific roles, and the extent of their engagement with or connection to other organizational domains. Early reflections included initial perspectives on reentry processes, relevant contextual factors such as institutional culture, policies, and laws, and candid discussion of their positions and access to power and influence within these systems.

The project leader prioritized engagement with team members with lived experience by regularly soliciting feedback and providing individualized one-on-one meetings as needed to address scheduling conflicts or clarify project matters. From the onset, lived experience was emphasized as a critical, highly valued form of expertise. These early efforts established expectations for mutual respect, trust, and psychological safety and prepared the team for reflexive dialogue and empathic listening. This groundwork set the stage for the dynamic, iterative processes of knowledge gathering, exchange, and visualization that followed.

Facilitate dynamic and iterative learning (Figure 1.2)

From January to December 2023, team participants met weekly for 90 minutes via teleconference (i.e., Zoom; 47 meetings total). These weekly team meetings were led by the project leader with meeting management support from Incubator staff. In parallel with these team meetings, the Incubator staff met weekly with the project leader to help plan the upcoming team meeting and debrief the last team meeting, including discussion of evolving team dynamics, strategies to moderate the participation of specific members (e.g., re-balance the input of disproportionately vocal members and quiet members), and targeted review of the project Gantt chart of the projected timeline for expected milestones. Given staffs’ role as observers in team meetings, the project leader also used debrief/planning meetings as an opportunity to regularly solicit feedback on how to improve her team leadership and facilitate dynamic, iterative learning on the team.

Gather, share, and receive information unobstructedly (Figure 1.2A)

With Incubator staff support, the project leader used the first team meetings to establish initial structures and practices to promote unobstructed exchange of information and ideas. To record and share information without obstruction, the project leader set up a shared folder (i.e., Microsoft Teams) to upload and organize all project documents with unrestricted access granted to team members, project advisors, and funding staff. Incubator staff facilitated access for all team members, including those outside of Dartmouth institutions, and assisted in orienting team members to the folder contents. The structure of team meetings included protected time for mental health check-ins, open discussion for brainstorming, collaborative agenda-setting, and solicitation of feedback on meeting organization and project progress.

Within this meeting structure, several practices were deployed to cultivate a willingness to engage in unobstructed information sharing. The project leader used norm setting and behavior modeling to promote the adoption of team formation behaviors. Established norms included project co-ownership, non-hierarchical leadership, verbalized dissent, and inclusion across differences. These norms were established by the project leader through explicit statements and behavioral modeling. Examples include verbal correction if a team member refers to the project as “your project” and the use of statements about being equal contributors, co-owners, and co-creators of the project until these sentiments became echoed by the team. Modeled behaviors included: reflexivity; transparent admission of knowledge gaps, assumptions, and mistakes; invitation of critique and disagreement; targeted request for feedback from silent members; and validation of others’ expertise and areas of leadership [49]. The project leader targeted additional norming behaviors to non-academic team members who were less familiar with academic processes. Such behaviors included explicit validation of their thoughts and ideas, verbal assurances that the expressed thoughts of all team members are equally relevant, positive and engaging verbal responses when divergent thoughts are expressed, the modeling of comfort in receiving divergent thoughts, and the model of directly engaging disagreements. The project leader practiced these norms and modeling behaviors with deliberate warmth, respect, and humility, operationalized through an authentic delivery of welcoming, calm verbal tones, virtual eye contact, honest sharing of her emotions and perspectives, empathic responses to those of her team, and curiosity for others’ experiences. She also met individually with team members who missed meetings or desired individual time to discuss their thoughts to facilitate meaningful participation of all members. Importantly, the established norms, behaviors, and their style of delivery were co-championed by Incubator staff and two team members (systems engineer; mixed-methods researcher) to facilitate rapid dispersion, adoption, and reinforcement across the heterogeneous team.

Within this team context, members collaborated in mixed-methods information gathering from January to April 2023 to develop a cohesive understanding of reentry care transitions. Methods included a rapid literature review using MEDLINE and Google Scholar to learn about existing reentry care transition programs and interventions in the United States (e.g., Transitions Clinic [78]. In parallel, the team collectively brainstormed questions for key stakeholder types involved in pre- and post-release processes (see Additional file 1) and met with 29 stakeholders to discuss real-world care transition processes from the NH DOC, perceived challenges, and opportunities for improvement. Notes from these stakeholder meetings were discussed during team meetings to support iterative sensemaking. For lived experience insights, a subset of the team also read de-identified transcripts (N = 19) from a concurrent, IRB-approved study of reentry among adults with opioid use disorder exiting NH prisons and jails (Dartmouth IRB STUDY00032371). No formal qualitative analysis was conducted as part of the team’s process to develop the vSMM; a formal content analysis of the stakeholder discussions will be reported separately.

Throughout this data gathering process, team members’ thoughts and ideas were gathered for group interrogation through audiovisual recording of the weekly team meetings, meeting minutes recorded by Incubator staff, and agenda co-creation during meetings. Upon the completion of data gathering and initial data review, the systems engineer team member with expertise in health system redesign led the team through an orientation to systems thinking approaches to health services innovations and processes for visualizing information (see Section 2.2.2) and harmonization of conceptual lexicons (see Section 2.2.3).

The project leader developed a modified version of the Quality of Patient-Centered Outcomes Research (QPCOR) Partnerships Instrument [79], which was administered by Incubator staff to team members at the project midpoint in July 2023 to assess the perceived quality of the team’s partnership (Table 3). Team members reported mean scores above 8 on all items, indicating that team members perceived a high quality of patient-centeredness and community-engagement in their collaboration.

Table 3.

Quality of patient-centered outcomes research partnerships (QPCOR) instrument, modified for diverse teams (N = 14), July 2023

Measure Statements	Mean (SD)
1) I have a clear understanding of the purpose of the project.	9.0 (1.71)
2) I feel listened to.	9.1 (1.54)
3) I feel prepared to be an equal partner in the project.	8.4 (2.13)
4) Team members are knowledgeable/willing to learn about how my expertise/experiences are relevant to the project.	9.4 (0.93)
5) I believe that I have choices in how I can be a part of the project.	8.5 (1.83)
6) I feel accepted by all members of the project team.	9.3 (0.99)
7) The project team uses language that is consistent with my values and culture.	9.5 (0.76)
8) Team members are thinking of ways we can continue to work together in the future.	9.0 (1.75)
9) I feel comfortable engaging with the members of the project team.	9.3 (1.20)
10) I feel like my views are incorporated into the project.	9.3 (1.14)
11) My participation in this project directly benefits my other work or activities.	8.8 (1.63)

* Scores range 0–10. Scores less than 7 indicate a need for intervention

Bound information in a visualization (Figure 1.2B)

The systems engineer led the team through an iterative, participatory process of visualizing and organizing the knowledge gathered from the interviews using free-form diagramming in a Miro board (www.miro.com) as a visualized boundary object (Figure 2). The systems engineer shared screen during weekly meetings and constructed preliminary models of NH DOCs reentry processes with the team in real-time drawing from verbalized content from the team’s live discussion of the interview results, literature results, and accounts provided by team members with lived experience of reentry from NH DOC. In addition, the systems engineer and project leader met individually with team members directly involved in reentry processes for targeted clarification and model refinement.

Fig. 2.

Visualized boundary object in Miro Board

The free-form diagrams in the Miro board were converted to formal systems modeling language (SysML) models. An engineer then programmed the EnVision System (Figure 3), which uses Python to translate SysML artifacts into interactive HyperText Markup Language (HTML) web-visualization packages [80]. The EnVision System translated the SysML models into an HTML webpage that provides navigable visualization of the model of NH DOC reentry processes for use with multidisciplinary audiences.

Fig. 3.

Stakeholder engagement process with the EnVision system. (reprinted with authors’ permission)

The HTML web visualization of the NH reentry model was distributed to team members via email, and written feedback was solicited. The systems engineer then presented the HTML web visualization to team members and solicited verbal feedback regarding its accuracy, integrating feedback into revisions.

Translate and combine information into a shared conceptual lexicon (Figure 1.2C)

Throughout the project year, the systems engineer supportively coached team members to make their assumptions explicit for public interrogation, clarification, definition, and reconciliation of terminology. She consistently reinforced this practice by modeling curious interrogation of expressed ideas until all members arrive at and confirm a shared understanding matched by a shared conceptual lexicon that clearly and accurately reflects the concept expressed. Examples of the team’s construction of a shared conceptual lexicon include their co-construction of a definition of the targeted problem, team roles and reentry contexts, and team engagement in systems thinking and modeling. Team members with lived experience also identified a new theoretical framework for the project—the eight dimensions of wellness [81]—which the team collectively agreed more comprehensively addressed reentry-relevant factors than the initial framework selected by the project leader.

To define the targeted problem, the project leader and the Incubator staff used the first team meeting to engage team members in a problem generation exercise, whereby each team member independently drafted a sentence to describe the problem to be targeted during the project year. The team leveraged a visual text-based boundary object by displaying all sentences on a virtual whiteboard, which was then used to verbally negotiate word choice, meaning, and order until the team reached consensus on a final problem statement.

The team also engaged in knowledge sharing about their respective roles related to reentry care transitions in NH and their experiences with the target problem. In sharing about their roles, team members taught each other terminology relevant to their roles and explained commonly used terms that convey different meanings in different contexts. For example, a few community-based team members initially used the term “jail” to refer to prison; upon consistent correction from the project leader and prison-based team members, they developed a shared conceptual understanding of the difference between jails and prisons and aligned their language with the correct use of the term “prison.” The process of sharing, listening, clarifying, and negotiating the meanings and terms used for team member roles and reentry contexts facilitated the level-setting of team members’ understanding of reentry from the NH DOC, the potential players involved, and the knowledge gaps present on the team. This process informed the team’s development of a shared spreadsheet (i.e., text-based boundary object) to list all roles and representatives to contact for interviews for data gathering (see Section “Gather, share, and receive information unobstructedly (Figure 1.2A)”) and the team’s development of the interview guide. The team used a collaborative approach to developing the interview guide during team meetings, in which questions were constructed based on team-identified gaps in knowledge and finalized by consensus. This approach provided another opportunity for terminology and meanings to be debated, negotiated, and reconciled across team members.

During the systems engineer’s orientation to systems thinking mentioned in Sect. “Bound information into a visualization (Figure 1.2B)”, she presented an overview of the purpose and methods used in systems thinking, definition of a “system” and its components, an explanation of key terminology (e.g., form, function, allocation) used during systems modeling [82, 83] and examples of their application to models of healthcare delivery. These concepts and terms were reinforced during the visualization of NH reentry processes to support the construction of the final working version of the vSMM.

Figure 4 shows views of the final working version of the vSMM. The team constructed a vSMM that allowed members to navigate from a birds-eye view of the reentry process to more detailed views of specific reentry activities. Figure 4A shows the top-level (Level 0), which presents the four highest-level steps in the reentry process. Figure 4B focuses in on the activity of releasing an individual from incarceration (red box) at its highest level of detail. Figure 4C focuses in on one release activity—parole (green box)—at its highest level of detail. Finally, Fig. 4D focuses in on one parole activity—approving the parole plan (pink box)—at its highest level of detail. The model can be found at https://sustainablehealth.med.wayne.edu/Projects/incarceration/index.html.

Fig. 4.

Visual representation of the visualized shared mental model at 4 levels: level 0 (A), level 1 (B), level 2 (C), and level 3 (D). Note: reentry from the NH state prison system follows a relatively linear set of steps with clearly defined start and end points. However, vSMMs can present both linear and non-linear system processes

Discussion

This paper set out to provide a practical framework for developing visualized shared mental models for heterogeneous teams addressing complex health problems. This framework articulates the iterative bidirectional nature of consensus-based meaning-making that must occur within multidisciplinary teams to ultimately achieve a collective mutual understanding about the team and its targeted goal. While participatory modeling of SMMs is more commonly used in technical disciplines (e.g., engineering), their adoption has been limited among health-related disciplines. Additionally, existing modeling approaches for developing SMMs use visualizations that are not easily accessible or comprehensible to multidisciplinary teams that include non-academic members (e.g., fuzzy-logic cognitive maps). Therefore, this paper offers two main contributions, it (1) extends existing participatory modeling approaches by introducing visualization with systems modeling language to enhance and sustain team SMMs, and (2) introduces health-related disciplines to the concept of SMMs and the process of developing SMMs about complex issues and within team environments. These contributions may serve to enhance the conduct of team science in all health-oriented fields, including but not limited to intervention design and delivery in health services research, quality improvement, innovation, and implementation science.

The vSMM framework extends existing participatory modeling approaches, strengthening the development and sustainability of team SMMs. In traditionally non-externalized SMMs, the SMM is cognitively held within each team member’s mind and tacitly assumed to be shared across team members, which fails to detect misaligned conceptual lexicons, conflicting assumptions, or information gaps. vSMMs advance non-externalized SMMs by enabling transparency of data integration among the entire team and preservation of the team’s collated knowledge. While reality can never be comprehensively represented [84] and complete alignment of individual SMMs across perspectives and value differences may not be attainable, vSMMs enable teams to co-develop sufficiently shared understanding to support coordinated, equity-oriented action. Such sufficiency is attainable through reflexivity, dialogue, mutual respect, curiosity, and empathic listening, particularly critical for teams working across disciplines, organizations, and lived experiences [39, 49].

The use of vSMMs supports an ongoing, iterative process of knowledge integration, reflecting Soft System Modeling’s emphasis on evolving insights, multiple stakeholder perspectives, and adaptation to changing contexts [85]. The sustainable nature of the visual outputs generated by visualization increases the likelihood that products, tools, and methodologies produced by a single team can be transferred to, adopted, and revised by other teams or contexts in future projects. In this way, vSMMs do not represent finalized, static representations of a complex issue; rather, vSMMs function as an adaptive blueprint of a complex issue, enabling subsequent teams of researchers, practitioners, and innovators to maximize resources by starting from advanced positions of knowledge rather than redundantly starting anew. Such may support the generation of higher-quality research questions, enhance the rigor and reliability of the team’s outputs, and reduce redundancy across projects. vSMMs may also help to maximize the return on investment for health systems funding quality improvement initiatives, as well as public and private funding for health services research, program development, and innovation.

Our case example of a vSMM for reentry care transitions demonstrates a process that can be used by multidisciplinary, heterogeneous health-oriented teams to align their mental models when addressing complex issues. While our case example uses vSMM to develop a baseline shared understanding of a targeted problem before taking action through a quality improvement project, vSMMs can be developed to support all stages of a project, including ideation, planning, execution, analysis, dissemination, and implementation. We believe vSMMs have immense potential applicability during the current formalization of the field of implementation science. For example, the Returning Home team could have developed a vSMM to model the QI intervention itself, including its forms, functions, and their allocations. vSMM could have then been used as part of an implementation strategy to train clinical sites to deliver the intervention, or as part of an implementation monitoring tool to assess the clinical site’s fidelity to the intervention model. Khayal and McGovern have previously demonstrated the utility of systems thinking and system modeling for implementation science, including for implementation planning, the design of implementation strategies, and the development of implementation modifications for health interventions [86]. Our paper builds on this prior work by formalizing a framework for developing vSMMs, which can and should be leveraged as a tool to advance the conduct of implementation science.

While vSMMs offer much value to teams, their use has some potential limitations. First, a vSMM’s validity depends on team composition; low team heterogeneity or attrition may produce a final model confounded by unknown missing information. The risk of attrition is particularly salient for community-academic partnerships and projects engaging persons with living experience of the targeted phenomenon. To ensure maximal participation, project leaders should identify anticipated barriers to participation during team formation and accommodate non-academic team members’ needs, such as by funding protected time, meeting via teleconference, acknowledging community expertise to mitigate impostorism, and providing partner-valued benefits to prevent exploitation [41]. Second, vSMM’s durability has caveats. vSMMs are living documents that never reach completion and require iteration to keep pace with knowledge evolution, just like we iterate cognitively held SMMs as we gain new knowledge [55]. Third, teams developing vSMMs should include a team member with expertise in systems thinking, which is a structured cognitive approach that recognizes a complex problem as a dynamic system with the interrelated forms, functions, and allocations that produce and reinforce complexity [83]. A systems engineer—or at least a practitioner trained in systems thinking—provides essential quality control by ensuring rigorous analysis of the problem’s dynamic nature, systematic assumption testing, cohesive integration of team knowledge, and expands the team’s access to sophisticated visualization tools (i.e., system modeling language) to preserve vSMMs for future use [82]. Healthcare programs are slowly introducing systems thinking into health-related curricula.

Future research should investigate scalable approaches for integrating vSMMs into various sectors, particularly healthcare, public health, and policy implementation. Our case example leveraged systems modeling language as a tool for visualizing SMMs, but it is possible that other modeling tools can be used effectively to achieve the same level of rigor and accessibility for multidisciplinary teams. Examining how different teams operationalize vSMMs in diverse contexts will provide valuable insights into best practices for leveraging visual representations for shared mental modeling in multidisciplinary collaborations. Future work should also explore strategies for maintaining and adapting vSMMs over time to ensure their long-term utility in dynamic research and implementation environments.

Conclusion

The vSMM framework leverages team member diversity and differences as integral assets that—through effective facilitation toward psychological safety and systems thinking—can yield shared mutual understanding of a phenomenon that approximates reality. The vSMM framework democratizes engagement for multidisciplinary heterogeneous teams, strengthening the integration of disparate areas of expertise to enhance the validity and robustness of a team’s understanding of real-world problems and systems. vSMMs enable teams to communicate their knowledge and ideas with greater clarity to those beyond their team (e.g., potential partners, adopters, funders, policy-makers), increasing their potential to drive high-impact collaborations, research, innovations, implementation, or policy change. All multidisciplinary teams seeking to address complex issues should consider leveraging the vSMM framework to enhance the nature and quality of the inputs, conduct, and outputs of their collective effort.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Author contributions

ISK and MFS conceptualized the framework with critical feedback from AMF. ALD conducted the literature review with supervision by MFS and ISK. MFS administered the modified version of the Quality of Patient-Centered Outcomes Research (QPCOR) Partnerships instrument and analyzed the results. MFS developed the manuscript outline with feedback from ISK, ALD, and AMF. ALD prepared Table 1. MFS prepared Tables 2 and 3. ISK prepared Figs. 1 and 2. JMJ prepared Figs. 3 and 4. All authors provided critical feedback to support table and figure presentation. MFS drafted the initial manuscript, which was critically reviewed and revised by all authors. MFS finalized the manuscript content with supervision by ISK. ISK formatted the manuscript for LaTeX. All authors read and approved the final version of the manuscript

Funding

This work was supported by an American Cancer Society Award (RSG-22–128-01-HOPS), the Susan & Richard Levy Healthcare Incubator at Dartmouth College and Dartmouth Health, National Institutes on Drug Abuse training grants (R25DA037190 and R25DA035163), a Health Resources and Services Administration training grant (T32HP32520), The Dartmouth Institute Health Equity Faculty Fellowship, the Class of 1974 Health Equity Scholars Program, and the Growing Convergence Research Program of the National Science Foundation (OIA 2317877). The funders played no role in the study design, data collection, analysis, and interpretation of data, or the writing of this manuscript.

Data availability

The survey results are described within the manuscript. Given the small sample size and the identifiability of individual responses, individual responses are not publicly available.

Declarations

Ethics approval and consent to participate

The Dartmouth Health Institutional Review Board reviewed and approved this project as non-human subjects research consistent with quality improvement (STUDY 00019204). The project was performed in accordance with the Helsinki guidelines and declarations. All participants provided informed consent to participate.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.