Mapping the Barriers to AI Implementation in Swedish Healthcare Key Barriers: Competencies, Data Accessibility, and Demonstrating Value of AI Products

Abstract

Artificial Intelligence (AI) has shown the potential to improve the quality of care for patients. However, barriers are preventing its adoption, and despite its potential, the integration of AI into Swedish healthcare remains limited. This thesis aims to identify the most impactful barriers to implementing AI in Swedish healthcare by identifying stakeholders, their interests, and their perceived barriers to the implementation. The data was collected through 19 semi-structured interviews. These interviewees were categorized into six stakeholders: clinicians, healthcare executives, MedTech companies, patients, policymakers, and researchers. Thematic analysis and tools from stakeholder analysis were utilized to synthesize overarching barriers in the system. The results revealed diverse perspectives on barriers for the stakeholders. The most prominent barriers identified were competencies, data accessibility, and demonstrating the value of AI products. Lack of competencies contributes to several other obstacles, and data accessibility was mentioned by all stakeholders. Demonstrating AI products’ value is fundamental to integrating AI into clinical practice. The thesis calls for further research to provide solutions to the identified barriers. If solved, the large-scale implementation of AI in Swedish healthcare will be one step closer.