#199 Innovative development of artificial intelligence image identification platform for peritoneal dialysis catheter exit

Abstract

Abstract Background and Aims This study was conducted at the Peritoneal Dialysis Center of National Taiwan University Hospital, enrolling 440 peritoneal dialysis patients in 2024. (9.6%), including catheter exit and tunnel infections. Of note, 3 patients (0.6%) experienced severe infections that were not treatable with medication alone and required surgical intervention to remove the catheter and place a new outlet to continue peritoneal dialysis. Method From July 2023 to November 2024, during routine examinations at the patient's catheter outlet, photos of the wound were taken with the patient's consent. Conduct a pilot study and designate 200 images for training the model. Assessment criteria follow a catheter exit site scoring system that categorizes results as normal (score 0–3) or indicative of high risk for infection (score ≥4). In addition, experts also conducted manual recognition evaluation on 200 images and evaluated the accuracy of the image recognition software. This study was approved by the Ethics Committee of National Taiwan University Hospital (202306040RINA). Results The final software prediction accuracy reached 87.5%. In addition, the software was used to compare its image recognition results with experts, and the accuracy of determining whether there was infection reached 95%. And upload the identification results to the statistical form record. Conclusion This image recognition peritoneal dialysis catheter exit platform optimizes the care process for peritoneal dialysis exit site infections. It is hoped that the exit site infections rate can be reduced or detected in time so that patients can be more objectively and immediately assisted in assessing exit site infections and introducing home dialysis. The image recognition platform for peritoneal dialysis catheter outlet is used to promptly treat catheter outlet infection and provide personalized health education. Reduce the chance of surgical extubating due to exit site infection, reduce the need to switch to haemodialysis mode due to infection, and reduce the medical cost of treatment. Fees and other items.

Ähnliche Arbeiten