Purpose
Workload for healthcare workers is increasing and the healthcare system are facing several challenges such as difficulties to recruit, growth of the elderly population and an increasing demand for advanced methods in diagnostics. In a world that is rapidly changing, there is a need to find new innovative solutions to meet the challenges we see now and in the future. Implementing new technologies can be challenging, but we have now stepped into the future of Artificial intelligence (AI) and want to share our experience, results...
Methods and materials
As the first hospital trust in Norway, Vestre Viken hospital trust (VV) implemented an AI application for fracture detection in the daily routine at five hospitals. Gleamer Boneview is a CE-approved application used for detecting fractures, bone lesions, dislocations, and effusions. The AI results are categorized as positive, doubtful, or negative.How does the solution work? The application supports diagnostics for emergency patients with skeletal trauma. After finalizing the exam and sending to Boneview (Gleamer) the pictures are pseudonymized, stripped for identifying data and encrypt. Then...
Results
The results of the validation gave us an accuracy of 91% for AI detection compared to radiologists’ 95% for blinded data. The negative predictive value for AI was 94%, and for the radiologist 95%. Based on these numbers, we decided to change our patient pathway. Using AI for detecting fractures gives the radiographer a supporting tool in their assessment for where to send patients. If there is a negative finding, the patients are sent home. For positive findings, patients are sent to consultation and treatment....
Conclusion
Implementing an AI application for fracture detection has given major beneficial impacts for both patients and healthcare workers. Departments and institutions involved, including the community healthcare service, are experiencing positive effects of the implementation. Flagging of AI results in RIS gives radiologists a triaging tool that benefits patients that needs priority. This is an important tool for this application, and we assume this will have a bigger beneficial impact for future AI implementations. Boneview (Gleamer) improves patient pathway for 39.000 patients in VV annually.The implementation...
Personal information and conflict of interest
L. Tveiten:
Nothing to disclose
References
Kelen, G. D., Wolfe, R., D’Onofrio, G., Mills, A. M., Diercks, D., Stern, S. A., Wadman, M. C., & Sokolove, P. E. (2021). Emergency Department Crowding: The Canary in the Health Care System. Catalyst non-issue content,2(5). https://doi.org/doi:10.1056/CAT.21.0217