Abstract: TH-PO024
Classification of Arteriovenous Access Aneurysm in Hemodialysis Patients Using Artificial Intelligence Application
Session Information
- AI, Digital Health, Data Science - I
November 02, 2023 | Location: Exhibit Hall, Pennsylvania Convention Center
Abstract Time: 10:00 AM - 12:00 PM
Category: Augmented Intelligence, Digital Health, and Data Science
- 300 Augmented Intelligence, Digital Health, and Data Science
Authors
- Dong, Zijun, Renal Research Institute, New York, New York, United States
- Zhang, Hanjie, Renal Research Institute, New York, New York, United States
- Wang, Lin-Chun, Renal Research Institute, New York, New York, United States
- Ren, Sarah, Renal Research Institute, New York, New York, United States
- Tisdale, Lela, Renal Research Institute, New York, New York, United States
- Han, Maggie, Renal Research Institute, New York, New York, United States
- Mohib, Shahzre, Renal Research Institute, New York, New York, United States
- Rosales M., Laura, Renal Research Institute, New York, New York, United States
- Prakash-Polet, Sindhuri, Azura Vascular Care, New York, New York, United States
- Douglas, Denzil, Azura Vascular Care, New York, New York, United States
- Starakiewicz, Piotr, Azura Vascular Care, New York, New York, United States
- Preddie, Dean C., Azura Vascular Care, New York, New York, United States
- Thijssen, Stephan, Renal Research Institute, New York, New York, United States
- Kotanko, Peter, Renal Research Institute, New York, New York, United States
Background
Arteriovenous (AV) access aneurysms in hemodialysis (HD) patients may lead to severe and potentially life-threatening consequences, such as rupture. To address this issue, we developed an artificial intelligence-based app that utilizes images of the AV access to classify AV aneurysms (Zhang, Clin Kidney J, 2022). Our objective was to assess the correlation between the classification results generated by our aneurysm classification app and the independent clinical examination performed by physicians specializing in access care.
Methods
As described recently (Zhang, Clin Kidney J, 2022), AV accesses were photographed, and the images were transferred to the cloud, where they were classified as "Advanced" or "Not Advanced" by a convolutional neural network algorithm (Fig. 1A). We compared classifications made by physicians who were blinded to the app results (i.e., the ground truth) and those generated by our app.
Results
We studied 28 subjects (Fig. 1B). Twenty out of 22 aneurysms were accurately classified as “Not Advanced” and 5 out of 6 as "Advanced" (Fig. 1C), resulting in an accuracy of 89%, sensitivity of 83%, specificity of 91%.
Conclusion
Our preliminary results show that an AI-powered app can classify AV aneurysms with actionable accuracy in a demographically diverse HD population. If these results are corroborated in a larger patient population, our tool has the potential to support the development of a robust aneurysm monitoring system, enabling timely detection, facilitating referrals, and avoiding emergency interventions associated with aneurysms.