Artificial intelligence (AI) is a promising generation of digital technologies. Recent applications and research suggest that AI can not only influence but also accelerate innovation in organizations. However, as the field is rapidly growing, a common understanding of the underlying theoretical capabilities has become increasingly vague and fraught with ambiguity. In view of the centrality of innovation capabilities in making innovation happen, we bring together these scattered perspectives in a systematic and multidisciplinary literature review. The aim of this literature review is to summarize the role of AI in influencing innovation capabilities and provide a taxonomy of AI applications based on empirical studies. Drawing on the technological–organizational–environmental (TOE) framework, our review condenses the research findings of 62 studies. The results of our study are twofold. First, we identify a dichotomous view of innovation capabilities triggered by AI adoption: enabling and enhancing. The enabling capabilities are those that research identifies as enablers of AI adoption, underscoring the competencies and routines needed to implement AI. The enhancing capabilities denote the role that AI adoption has in transforming or creating innovation capabilities in organizations. Second, we propose a taxonomy of AI applications that reflects the practical adoption of AI in relation to three underlying reasons: replace, reinforce, and reveal. Our study makes three main contributions. First, we identify the innovation capabilities that are either required for or generated by AI adoption. Second, we propose a taxonomy of AI applications. Third, we use the TOE framework to track trends in the theoretical contributions of recent articles and propose a research agenda. © 2023 The Authors.

Introduction Absence of structured guidelines to navigate the complexities of implementing AI-based applications in healthcare is recognized by clinicians, healthcare leaders, and policy makers. AI implementation presents challenges beyond the technology development which necessitates standardized approaches to implementation. This study aims to explore the activities typical to implementation of AI-based systems to develop an AI implementation process framework intended to guide healthcare professionals. The Quality Implementation Framework (QIF) was considered as an initial reference framework.Methods This study employed a qualitative research design and included three components: (1) a review of 30 scientific articles describing differences empirical cases of real-world AI implementation in healthcare, (2) analysis of qualitative interviews with healthcare representatives possessing first-hand experience in planning, running, and sustaining AI implementation projects, (3) analysis of qualitative interviews with members of the research groups network and purposively sampled for their AI literacy and academic, technical or managerial leadership roles.Results The data were deductively mapped onto the steps of QIF using direct qualitative content analysis. All the phases and steps in QIF are relevant to AI implementation in healthcare, but there are specificities in the context of AI that require incorporation of additional activities and phases. To effectively support the AI implementations, the process frameworks should include a dedicated phase to implementation with specific activities that occur after planning, ensuring a smooth transition from AI's design to deployment, and a phase focused on governance and sustainability, aimed at maintaining the AI's long-term impact. The component of continuous engagement of diverse stakeholders should be incorporated throughout the lifecycle of the AI implementation.Conclusion The value of this study is the identified processual phases and activities specific and typical to AI implementations to be carried out by an adopting healthcare organization when AI systems are deployed. The study advances previous research by outlining the types of necessary comprehensive assessments and legal preparations located in the implementation planning phase. It also extends prior understanding of what the staff's training should focus on throughout different phases of implementation. Finally, the overall processual, phased structure is discussed in order to incorporate activities that lead to a successful deployment of AI systems in healthcare. © 2025 Nair, Nygren, Nilsen, Gama, Neher, Larsson and Svedberg.

Introduction: Interstitial lung diseases (ILDs) are chronic, progressive conditions that affect the lung parenchyma and require continuous monitoring to assess treatment response and clinical course. Chest computed tomography (CT) is the primary examination used for this follow-up, but precise, longitudinal quantification of pulmonary changes remains a challenge in clinical practice. The absence of objective tools to measure disease progression hampers comparisons across serial examinations. Artificial intelligence (AI)–based technologies offer new possibilities for quantifying pulmonary patterns over time, supporting medical decision-making with greater precision. This paper presents a case study on the collaborative development of an AI tool aimed at the longitudinal quantification of pulmonary changes in patients with ILD, integrating clinical and technological expertise across an international network of institutions.

Objectives: To develop an artificial intelligence tool capable of quantifying pulmonary changes in order to support longitudinal clinical follow-up and assist medical decision-making.

Materials and Methods: This is a case study conducted within an international collaboration among academic, clinical, and technology institutions in Brazil and Sweden. The tool was developed using anonymized chest CT scans from patients with a confirmed diagnosis of interstitial lung disease. Development is carried out by interdisciplinary teams working together on defining clinical criteria, curating data, training the model, and evaluating results. The process is supported by regular meetings among partners, enabling joint decision-making and continuous adaptation of the tool to identified clinical needs.

Results: The tool is currently in the training phase, using anonymized chest CT examinations from patients with ILD. Preliminary tests have been performed and are being reviewed by radiologists on the clinical team to verify the consistency of the segmentations generated by the model. Thus far, initial results indicate that the system has the potential to contribute to longitudinal disease monitoring by providing quantitative support for the analysis of serial examinations. The next stage of the project involves continuous model refinement and expanded clinical validation in a controlled setting.

Conclusions: The development of this AI tool represents a promising advance in supporting the longitudinal follow-up of interstitial lung diseases. By standardizing the quantification of changes on CT examinations, the technology seeks to provide additional input for clinical decision-making. The collaborative approach between technical and clinical institutions has been essential to ensure that the solution meets real needs in medical practice. Preliminary results are encouraging, and next steps involve further model improvement.

Introduction: The incorporation of artificial intelligence (AI) into healthcare has driven the development of tools that support clinical decision-making, especially in specialties such as radiology. However, creating AI-based Clinical Decision Support Systems (CDSS) requires intensive collaboration among professionals from different fields. This interdisciplinarity, while essential, generates cognitive and operational tensions arising from distinct ways of thinking, professional values, and success criteria. Prior research shows that developers tend to emphasize algorithmic accuracy and technical performance, whereas clinical professionals prioritize tacit knowledge. These differing perspectives lead to divergent interpretations of the same problems. Although many studies address implementation challenges, few investigate tensions in the early phases of development. Understanding how these tensions emerge, are interpreted, and are negotiated can inform more effective collaboration strategies and the development of technologies better aligned with clinical practice.

Objectives: To understand how healthcare and technology professionals perceive and handle the tensions that emerge during the development of an AI-based CDSS.

Materials and Methods: This study adopts a participatory action research approach conducted during the development of a CDSS focused on pulmonary radiology. The project involved collaboration among an academic institution, a medical-technology company, a public hospital, and a technology research unit in Brazil and Sweden. Data collection took place over six months.

Results: Preliminary analysis revealed a set of recurring tensions between clinical and technical professionals during CDSS development. The main tensions concerned: (1) the types of knowledge valued; (2) expected timelines for delivery and validation; (3) the use of sensitive data; (4) division of responsibilities; (5) alignment between technical and clinical workflows; and (6) management of professionals’ time and resources. These tensions were not perceived uniformly across groups. Tensions were also observed around image labeling: while viewed as a simple step by technical teams, clinicians perceived it as burdensome and misaligned with their clinical duties. These mismatches reflected not only operational differences but also distinct views of what counts as effective collaboration. Moreover, tensions tended to overlap and intensify at specific moments, complicating alignment between teams.

Conclusions: The findings indicate that developing AI technologies in healthcare involves tensions among different understandings of value, safety, and collaboration. Grasping how these tensions emerge can help anticipate friction points and promote more effective alignment strategies. Identifying such mismatches from the earliest phases is essential for improving the integration of digital technologies into clinical practice.   

This research is an interdisciplinary work between data scientists, innovation management researchers, and experts from a Swedish hygiene and health company. Based on this collaboration, we have developed a novel package for automatic idea detection to control and prevent healthcare-associated infections (HAI). The principal idea of this study is to use machine learning methods to extract informative ideas from social media to assist healthcare professionals in reducing the rate of HAI. Therefore, the proposed package offers a corpus of data collected from Twitter, associated expert-created labels, and software implementation of an annotation framework based on the Active Learning paradigm. We employed Transfer Learning and built a two-step deep neural network model that incrementally extracts the semantic representation of the collected text data using the BERTweet language model in the first step and classifies these representations as informative or non-informative using a multi-layer perception (MLP) in the second step. The package is AID4HAI (Automatic Idea Detection for controlling and preventing Healthcare-Associated Infections) and is made fully available (software code and the collected data) through a public GitHub repository (https://github.com/XaraKar/AID4HAI). We believe that sharing our ideas and releasing these ready-to-use tools contributes to the development of the field and inspires future research.

Innovations that employ novel technologies can be problematic, particularly when sources of uncertainty can lead to severe financial, social, and reputational losses. Nowhere is this more evident than in innovations across Medical Device (MD) developments. In healthcare, described by stricter safety requirements, adapting the MD process to ongoing regulatory principles intended to balance benefits and risks is often elusive. We investigate a single case study comprising four medical device technology developments implemented in products and services. Does this study explore how a healthcare firm changes its medical device development process to adaptive regulations? Our study offers three contributions. First, we contribute to the innovation literature by proposing a flexible MD process in which safety standards are continuously revised, and development stages are regulated differently. For example, when the legal and regulatory aspects of emerging technologies are unpredictable, unknown firms are encouraged to de-risk the early-stage potential problem in adopting the emerging technology. Second, we contribute to the literature on Technology Development by showing how introducing digital technologies innovation requires a significant change in the culture and mindset of the organization. In the healthcare industry, where rules and procedures are hindering risky and uncertain investment, nurturing the culture of people towards risk-taking and learning from failure is a crucial dimension of digital transformation. Third, we propose a combined process that leverages traditional MD phases. The change management theory suggests a way to enact a digital transformation in a hyper-regulated environment.

Based on the results of a multiple case study of seven manufacturing firms, a contingency framework for harnessing fuzziness in the front end of innovation is proposed by delineating two discrete capability paths through which new product ideas are developed into corroborated product definitions. The study illustrates that ideas characterized by high levels of fuzziness benefit from following an exploratory path, where the creative potential of fuzziness is embraced by deploying problem-formulation and problem-solving capabilities. In contrast, ideas at low levels of fuzziness benefit from following an exploitative path, where fuzziness is tolerated by drawing upon idea-refinement and process-management capabilities. When the fuzziness level of the idea and the set of capabilities to develop the idea are poorly aligned, the idea-development process is either inefficient or runs the risk of stalling. These findings have theoretical and practical implications for the front end of innovation and new product idea development. © 2021 The Authors. Published by Elsevier Ltd.

Disruptive innovation is widely recognised as a bold enterprise. Risks and uncertainties drive incumbent firms to seek alternative solutions to find disruptive ideas. Machine learning emerges as a powerful tool to reduce uncertainties while processing vast amounts and types of information. However, incumbents encounter immense difficulty in codifying tacit knowledge into effective algorithms and often end up with incremental or tactical outcomes despite bold aspirations. Using the literature on problem finding, we explore the development process of machine learning for ideation. Our action research conducted on a healthcare firm provides theoretical and managerial contributions. First, this study suggests that ideation for disruptive innovation benefits from machine learning by facilitating a heuristic search in which a group of actors evaluate plausible hypotheses rather than seek logically accurate conclusions. Previous studies on ideation stress directional search. Second, we propose an ideation process centred on problem formulation to identify disruptive innovation based on its inherent characteristics (e.g., radical functionality and discontinuous technical standard). Third, we discuss the challenges of adopting algorithm-based systems in the ideation — a process well known for being fuzzy.

Background: Significant efforts have been made to develop artificial intelligence (AI) solutions for health care improvement. Despite the enthusiasm, health care professionals still struggle to implement AI in their daily practice.

Objective: This paper aims to identify the implementation frameworks used to understand the application of AI in health care practice.

Methods: A scoping review was conducted using the Cochrane, Evidence Based Medicine Reviews, Embase, MEDLINE, and PsycINFO databases to identify publications that reported frameworks, models, and theories concerning AI implementation in health care. This review focused on studies published in English and investigating AI implementation in health care since 2000. A total of 2541 unique publications were retrieved from the databases and screened on titles and abstracts by 2 independent reviewers. Selected articles were thematically analyzed against the Nilsen taxonomy of implementation frameworks, and the Greenhalgh framework for the nonadoption, abandonment, scale-up, spread, and sustainability (NASSS) of health care technologies.

Results: In total, 7 articles met all eligibility criteria for inclusion in the review, and 2 articles included formal frameworks that directly addressed AI implementation, whereas the other articles provided limited descriptions of elements influencing implementation. Collectively, the 7 articles identified elements that aligned with all the NASSS domains, but no single article comprehensively considered the factors known to influence technology implementation. New domains were identified, including dependency on data input and existing processes, shared decision-making, the role of human oversight, and ethics of population impact and inequality, suggesting that existing frameworks do not fully consider the unique needs of AI implementation.

Conclusions: This literature review demonstrates that understanding how to implement AI in health care practice is still in its early stages of development. Our findings suggest that further research is needed to provide the knowledge necessary to develop implementation frameworks to guide the future implementation of AI in clinical practice and highlight the opportunity to draw on existing knowledge from the field of implementation science. ©Fábio Gama, Daniel Tyskbo, Jens Nygren, James Barlow, Julie Reed, Petra Svedberg. 

Firms increasingly use artificial intelligence (AI) to innovate. Yet, the current literature offers fragmented guidance regarding its use in innovation processes. This study argues that the exponential use of AI has resulted in a mass of disorganised knowledge, creating confusion and frustration surrounding how managers navigate from conventional to digital innovation processes. A systematic literature review was carried out to examine studies investigating AI innovation published over the last decade (2011–2021). The results suggest that AI is present across all innovation phases and that firms have created three unique capabilities: AI-enabled ideation, AI-enabled development, and AI-enabled commercialisation. This article enriches the innovation management literature, and it equips managers with practical guidance in the use of AI.