LB076 - AI-BASED RISK PREDICTION OF PICC COMPLICATIONS IN PATIENTS RECEIVING PARENTERAL NUTRITION
LB076
AI-BASED RISK PREDICTION OF PICC COMPLICATIONS IN PATIENTS RECEIVING PARENTERAL NUTRITION
S. M. Lee1, J. Y. Park2,*
1Department of Biohealth & Medical Engineering, Gachon University, Seongnam-si, 2Department of Clinical Nursing, University of Ulsan, Seoul, Korea, Republic Of
Rationale: Parenteral nutrition (PN) is a life-sustaining intervention for patients with limited oral or enteral nutritional intake. The essential components of PN include carbohydrates, lipids, amino acids, vitamins, trace elements, electrolytes, and water, all of which can be delivered intravenously. Proper vascular access is essential for safe administration, and a PICC refers to a central venous catheter inserted via the peripheral veins and is relatively more suitable for PN infusion compared to traditional central venous catheters. Thus clinical use of peripherally inserted central catheters (PICCs) has been increasing. This study aims to develop artificial intelligence (AI)-based survival analysis and machine learning models to predict peripherally inserted central catheter (PICC) complications and identify significant risk factors for PICC related complications in parenteral nutrition patients.
Methods: This study was designed as a retrospective medical record analysis. Logistic regression, support vector machine, random forest, and extreme gradient boosting were used to develop discrete complication prediction models, whereas survival analysis models, including random survival forest, DeepSurv, and DeepHit, were used to create time-varying complication prediction models. Model performance was evaluated using accuracy for complication occurrence and the concordance index (C-index) and integrated Brier score (IBS) for catheter use.
Results: A total of 218 participants were included in the study. Patients diagnosed with cancer were 44.4% and 50.5% in the complication and control groups, respectively. Patients without a history of central venous catheter insertion and PICC procedures were 83.3% and 92.5% and 83.3% and 92.5% in the complication and control groups, respectively. For the catheter insertion site, the right upper arm was used in 61.1% and 95% of patients in the complication and control groups, respectively. Among the independent variables, the difference in catheter insertion site was statistically significant between the two groups (p < 0.001) but not in other independent variables. Complication prediction achieved a mean accuracy of 0.92. Among the survival models, DeepSurv exhibited the best C-index (0.61) but a relatively higher IBS (0.170). Significant complication risk factors included the catheter insertion site, catheter diameter, gender, cancer diagnosis, and timing of the PICC insertion decision. Left-arm insertion and larger catheter diameters were associated with higher complication risks.
Conclusion: This study is significant in developing a PICC complication prediction model to support clinical decision-making and explaining the model’s functioning using explainable AI (XAI) techniques. The proposed predictive models have the potential to significantly impact clinical practice by enabling early identification of patients at high risk for PICC-related complications. This can lead to more personalized and proactive care, reduce complication rates, and lower healthcare costs. Furthermore, the implementation of these models in clinical settings may enhance patient safety and improve outcomes.
References: Khan, A., Laing, E., Beaumont, A., Wong, J., Warrier, S., & Heriot, A. (2023). Peripheral parenteral nutrition in surgery–a systematic review and meta-analysis. Clinical nutrition ESPEN, 54, 337-348.
Sheng, Y., & Gao, W. (2024). Machine Learning Predicts Peripherally Inserted Central Catheters-Related Deep Vein Thrombosis Using Patient Features and Catheterization Technology Features. Clinical Nursing Research, 33(6), 460-469.
Kim, K., Kim, Y., & Peck, K. R. (2020). Previous peripherally inserted central catheter (PICC) placement as a risk factor for PICC-associated bloodstream infections. American Journal of Infection Control, 48(10), 1166-1170.
Disclosure of Interest: None declared