Kidney Week

Abstract: SA-PO0086

Integrating Standard Clinical Markers with Systemic Inflammation Response Index and Obesity Indices for Early Detection of Kidney Impairment: An XGBoost and SHAP Analysis

Session Information

• Artificial Intelligence: Kidney Imaging, Pathology, and Precision Phenotyping
  October 24, 2026 | Location: Exhibit Hall A, Convention Center
  Abstract Time: 10:00 AM - 12:00 PM

Category: Artificial Intelligence, Digital Health, and Data Science

• 300 Artificial Intelligence, Digital Health, and Data Science

Authors

• Kookanok, Chutawat, Interfaith Medical Center, New York, New York, United States

Chutawat Kookanok, MD

• Aiumtrakul, Noppawit, Mayo Clinic Minnesota, Rochester, Minnesota, United States

Noppawit Aiumtrakul, MD

• Poochanasri, Methavee, Phramongkutklao Hospital, Bangkok, Thailand

Methavee Poochanasri, MD

• Kulthamrongsri, Narathorn, University of Hawai'i System, Honolulu, Hawaii, United States

Narathorn Kulthamrongsri, MD

• Yongkiatkan, Panchanit, University of California Irvine, Irvine, California, United States

Panchanit Yongkiatkan, MD

• Win, Kaung Htet Hla, Interfaith Medical Center, New York, New York, United States

Kaung Htet Hla Win, MBBS

• Aung, Htoo Pyae, Interfaith Medical Center, New York, New York, United States

Htoo Pyae Aung

• Sule-saa, Samuel, Interfaith Medical Center, New York, New York, United States

Samuel Sule-saa, MBChB, MPH

• Sackey, Jeffrey Afrah, Interfaith Medical Center, New York, New York, United States

Jeffrey Afrah Sackey

• Tangprasittichok, Nathasith, Phramongkutklao Hospital, Bangkok, Thailand

Nathasith Tangprasittichok, MD

• Phichedwanichskul, Kitinan, Phramongkutklao Hospital, Bangkok, Thailand

Kitinan Phichedwanichskul, MD

• Tantisattamo, Ekamol, University of California Irvine, Irvine, California, United States

Ekamol Tantisattamo, MD, MPH, FASN

Background

The prevalence and burden of renal impairment represent a critical public health concern and early risk stratification are required. While traditional models rely heavily on established renal biomarkers, the predictive role of systemic inflammation and obesity indices remains an important area of investigation. This study aimed to develop an explainable machine learning framework to predict risk by integrating standard demographic and clinical markers with specific novel indices, the Systemic Inflammation Response Index(SIRI) and Waist-to-Hip Ratio(WHR).

Methods

Using NHANES data from 6,450 adults, missing values were handled by median imputation. Renal impairment was defined as an eGFR <60 mL/min/1.73 m² and/or UACR ≥30 mg/g. SIRI was calculated as (neutrophil × monocyte) / lymphocyte count. Three model configurations were developed using four machine learning algorithms—XGBoost, Random Forest, Support Vector Machine(SVM), and Naïve Bayes—with SHAP(SHapley Additive exPlanations) analysis to predict renal impairment risk. Model 1 included demographic, lifestyle, anthropometric, and blood pressure variables; Model 2 incorporated laboratory markers and SIRI; and Model 3 integrated variables from both Models 1 and 2.

Results

The XGBoost model demonstrated superior predictive performance compared with all other models(accuracy 0.90, ROC-AUC 0.88, PR-AUC 0.50). SHAP analysis identified age as the strongest predictor, followed by serum uric acid, metabolic indices, obesity indices, and inflammatory markers(SIRI and hs-CRP)(Figure 1). The SHAP beeswarm plot further illustrated the directional associations of predictors(Figure 2).

Conclusion

XGBoost demonstrated high accuracy for predicting renal decline. Using explainable AI techniques, this study identified SIRI and WHR as significant contributors to renal decline prediction. The model shows potential as a clinical decision-support tool, highlighting the importance of monitoring modifiable inflammatory and metabolic indices for early CKD prevention.