Conference Agenda

This paper develops an analytical framework that combines the strategy-as-process approach with public management literature on collaborative innovation. The strategy-as-process school seeks a behavioural and historically informed explanation of strategic change - and resistance - in whole organizations and over time (Pettigrew, 1987). It emphasizes the role of prior interactions, stakeholders’ interests and perceptions, and often explores organizations’ wider context (Ferlie & Ongaro, 2022).

Historically there has been an emphasis in the strategy-as-process school on change processes within large and stand alone organizations – rather than more diffuse networks, now viewed crucial for tackling complex problems in the public sector (Alford & Greve, 2017; Cinar et al. 2024; Torfing, 2019). Also, empirical studies have all been based on English cases. We aim to counter these limitations by attending to the network level of strategy for collaborative innovation, in a non-UK context.

The empirical research is part of a larger EU-H2020 project (2018-2022) of strategic renewal, co-governance and co-creation in public organizations across Europe (Ferlie et al. 2024; Van Gestel et al. 2023). In this paper, we investigate the strategy-as-process in a comparative analysis of collaborative innovation within one country (The Netherlands), studying the linkages between strategic management and network innovation in the social and employment services, nationally and in six regions. We particularly examine a strategy called the Jobs Agreement (2013-2025) to innovate public services for disadvantaged jobseekers.

The methodological approach reflects a processual, contextual and comparative case study design (Langley et al. 2013) where purposefully selected networks are investigated in their environment and over time (Ferlie et al. 2013). Comparative case studies (here of six regions) are recommended because of their relatively high internal validity and larger external generalizability than possible with a single case (Yin 1994). Data for this paper were derived from documentary study and 44 in-depth (individual and group) interviews with 74 participants (politicians, managers, unions, employers and client associations, professionals and policy experts) across the country. Based on a process-based strategic analysis (Pettigrew, 1987), the findings indicate which type of innovation occurs in the network settings and where and why not or to a lesser extent.

The paper contributes to the literature in two ways:

- we add a network perspective to the strategy-as-process literature, bridging its insights with theoretical and conceptual understandings of collaborative innovation in the public sector (Bryson et al. 2015; O’Toole, 2015).

- we contribute to strategic management literature by comparing network governance focused on similar public sector problems but in different (regional) contexts.

Utilizing a behavioural and historically informed explanation of strategic change (and resistance to it) (Ferlie & Ongaro, 2022), we could trace the impact of national reform history and differing institutional design, collaboration and leadership factors in the local/regional change process for innovative results. We thus specify the use of the strategy-as-process approach at the network level, and add a strategy-as-process perspective to the often a-historical and a-processual studies of collaborative innovation in the public sector.

Background

Artificial Intelligence (AI) is increasingly seen as a strategic driver of innovation in healthcare systems, offering the potential to improve clinical effectiveness, optimize resource use, and support decision-making. One promising application is the use of AI-powered Clinical Decision Support Systems (CDSS), particularly in primary care, where general practitioners (GPs) often face high workloads, limited time, and incomplete information (Rajkomar et al.). In Italy, the NHS shows notable regional differences in prescription habits and healthcare consumption that are linked to local clinical cultures and practices. These disparities result in inefficiencies, inequalities, and reduced effectiveness in public healthcare spending.

To address these issues, the Italian Ministry of Health, through its technical agency Agenas, has launched a nationwide initiative to provide GPs with a CDSS powered by AI. This tool aims to improve diagnostic accuracy, support personalized care plans, strengthen chronic disease management, and enhance preventive services. While this initiative aligns with broader goals of digitalizing territorial healthcare, the presence of technology alone does not ensure successful adoption (Roppelt et al.).

Objectives and Research Questions

This study explores the multi-level challenges and opportunities associated with the adoption of AI-driven CDSS in primary care. The research focuses on two key levels:

• Professional level: It is crucial to explore how GPs perceive AI prior to implementation. Clinicians’ interpretations of new technologies play a fundamental role in their development, deployment, and appropriation (Leonardi & Barley, 2010). AI might be seen as a support tool that enhances decision-making, or as a threat to autonomy and the doctor-patient relationship.

• Organizational level: Healthcare organizations must adapt service models to integrate CDSS into existing clinical workflows without disrupting care continuity. This involves prioritizing patient needs, ensuring coordination among providers, and aligning the CDSS with operational strategies.

Based on this twofold focus, the study will answer the following research questions:

1. How do General Practitioners perceive AI prior to its adoption—as a technological tool and in terms of its impact on the patient relationship, professional autonomy, and identity? Which conditions do they consider necessary for its successful adoption?

2. What are the expected benefits of the AI-CDSS for healthcare managers, what change management strategies do they propose, and what role do they attribute to Local Health Authorities in facilitating its adoption in primary care?

Methodology

A qualitative, multi-level research design will be used, including:

(i) semi-structured interviews with key GP opinion leaders across Italy to explore expectations, barriers, and enablers of CDSS adoption;

(ii) survey to managers of Local Health Authorities to examine organizational strategies, change management, and training needs.

Expected Results

The study will generate practical insights for promoting CDSS adoption at both professional and organizational levels. It will identify drivers of AI acceptance, suggest engagement strategies for GPs, and offer recommendations on care model redesign, training, and incentives. Ultimately, it aims to contribute to a more effective, sustainable, and equitable healthcare system enhanced by AI.