Implementation Intelligence in Cardiology: Real-World Use of AI Tools and Barriers to Uptake

Abstract

Background: Artificial intelligence (AI) has also proliferated at a very fast rate in the fields of cardiology where it aids in diagnostics, risk forecasting, and imaging analysis, along with workflow automation. Although the pace of technological change is increasing, its application in practice varies, and the unique obstacles affecting the adoption by the clinician are not comprehensible. Objective: To determine the current issues of AI tools use by professionals in cardiology in practice, determine the facilitators and obstacles to their implementation, and understand the willingness to expand use to more aspects of clinical practice. Methods: Eight cardiology centers have been considered in the case of a mixed-methods study performed in 2021-2023. The survey on 512 clinicians (cardiologists, fellows, nurses, and imaging specialists) was used to gather quantitative data. The survey evaluated the familiarity of AI, its usage frequency, perceived utility, and institutional support. The semi-structured interviews that were provided as a part of the qualitative research on the problem of workflow and trust-related issues involved 46 participants. The descriptive statistics and the thematic analysis were used to summarize these findings. Results: One-third of the participants said that it have regularly been using AI tools with the most frequent ones being cardiac imaging and risk stratification. The most perceived to be the key facilitators were accuracy, workflow effectiveness and enabling institutional infrastructure. The lack of training (62%), fears of transparency of algorithms (54%), workflow disruption (49%), and medicolegal uncertainty (41%) were the greatest barriers. Interview stories contributed the theme of mistrust in black-box models and lack of integration with any of the existing electronic health systems. Conclusion: Although there is wide excitement, adopting AI into cardiology is limited in reality due to knowledge gap in training, transparency issues, and workflow mismatch. Overcoming these obstacles by designing clinical-based application platforms, better model clarity and unified integration systems will play a critical role in managing scalable and responsible AI practice in cardiovascular healthcare.

Ähnliche Arbeiten