Machine Learning Models Using Hospital Admission Characteristics Do Not Optimally Predict Nosocomial Infection Development in a Global Cirrhosis Cohort

BACKGROUND: Nosocomial infections (NI) in cirrhosis are associated with high mortality but could be preventable. Logistic regression (LR) models have failed to identify high-risk patients. We aimed to develop machine learning (ML) models to predict NI. METHODS: The CLEARED consortium consists of prospectively enrolled cirrhosis inpatients from >120 centers. Using day-of-admission clinical data, three ML approaches[Random Forest (RF), Extreme Gradient Boosting (XGBoost), and Neural Networks (NN)] were used to predict NI. Data were split 80:20 for training and testing stratified by the outcome. Models were compared using Area-under-Receiver-operating-characteristic curve (AUC). RESULTS: 8,263 patients (55.90±13.34 years; 64.1% men) from 127 centers in 37 countries were included. NI developed in 869 (10.5%) a median of 6 (4-11) days post-admission. Major NIs were respiratory (29.6%) and urinary tract infection (15.7%), and spontaneous bacterial peritonitis (13.5%). NIs occurred more frequently in patients from low/low-middle income countries and those with severe liver disease, alcohol etiology, and admission infections. NIs were associated with inpatient mortality (31.9% vs. 8.1%, p<0.001) and liver transplantation (4.8% vs. 1.9%, p<0.001). While the RF model (AUC 0.69) showed good calibration (Brier score 0.09), outperforming XGBoost, NN, and LR models (AUC 0.66 for all; LR comparison p = 0.043), no model achieved AUC ≥0.80 for clinical utility. At 10% predicted probability threshold, RF model demonstrated only 75.4% sensitivity, 52.9% specificity, and 15.9% PPV. CONCLUSIONS: NIs cannot be accurately predicted from day-of-admission data using ML models, even in a large, prospective, global cirrhosis cohort. Every hospitalized patient with cirrhosis should receive protocolized infection-control measures.

Authors

Institutions

Publication Details

Journal
The American Journal of Gastroenterology
Published
2026-04-27
DOI
https://doi.org/10.14309/ajg.0000000000004031
Primary Topic
Liver Disease and Transplantation
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
article

Machine Learning Models Using Hospital Admission Characteristics Do Not Optimally Predict Nosocomial Infection Development in a Global Cirrhosis Cohort

Minghua Su, Anoop Saraya, Jacob George, Patrick S. Kamath et al.
The American Journal of Gastroenterology
Liver Disease and Transplantation
article

Machine Learning Models Using Hospital Admission Characteristics Do Not Optimally Predict Nosocomial Infection Development in a Global Cirrhosis Cohort

Minghua Su, Anoop Saraya, Jacob George, Patrick S. Kamath, A Ullate De La Torre, Aloysious Aravinthan, Gustavo Pereira, Dibya Lochan Praharaj, Hanna Aberra, Katina Zheng, J Córdova-Gallardo, Shiv Kumar Sarin, Dinc Dinçer, Adrian Carlos Gadano, René Malé Velazquez, Helena Katchman, Ramazan Idılman, Scott Silvey, Ashok Choudhury, Garima Singhal, Brian J. Bush, X I A O Y U Wang, Henok Fisseha, Aabha Nagral, Dominik Bettinger, Radha K. Dhiman, Marco A. Arrese, Anand Vijay Kulkarni, Umang Arora, Hiang Keat Tan, Haibing Gao, Nabil Debzi, Xin Zheng, Qing Xie, Jawaid Shaw, Alberto Queiróz Farias, Andrés Duarte‐Rojo, Dinesh Kumar Jothimani, Sezgin Barutçu, Amey Sonavane, Harendran Elangovan, Zeki Karasu, Sunil Taneja, Sanjiv Mahadeva, Coşkun Özer Demirtaş, Cameron Gofton, Kessarin Thanapirom, Chuanwu Zhu, Sarai Gonzalez Hueso, Ponan Claude Regis Lah, Caiyan Zhao, CE Eapen, Wai‐Kay Seto, P C Hayes, Mario Silva, Mingqin Lu, Ningping Zhang, Yongfang Jiang, Mark Topazian, Adam Doyle, Daneille Adebayo, Gerry MacQuillan, Dilara Turan, Jasmohan S Bajaj, Sevda Aghayaeva, Neil Rajorya, Hailemicheal Desalgn, José Antonio Velarde-Ruiz Velasco, Florence Wong, Sumeet Asrani, Lei Wang, Chenghai Liu, Zhujun Cao, Bin Xu, James Fung, K Rajender Reddy
article en

Abstract

BACKGROUND: Nosocomial infections (NI) in cirrhosis are associated with high mortality but could be preventable. Logistic regression (LR) models have failed to identify high-risk patients. We aimed to develop machine learning (ML) models to predict NI. METHODS: The CLEARED consortium consists of prospectively enrolled cirrhosis inpatients from >120 centers. Using day-of-admission clinical data, three ML approaches[Random Forest (RF), Extreme Gradient Boosting (XGBoost), and Neural Networks (NN)] were used to predict NI. Data were split 80:20 for training and testing stratified by the outcome. Models were compared using Area-under-Receiver-operating-characteristic curve (AUC). RESULTS: 8,263 patients (55.90±13.34 years; 64.1% men) from 127 centers in 37 countries were included. NI developed in 869 (10.5%) a median of 6 (4-11) days post-admission. Major NIs were respiratory (29.6%) and urinary tract infection (15.7%), and spontaneous bacterial peritonitis (13.5%). NIs occurred more frequently in patients from low/low-middle income countries and those with severe liver disease, alcohol etiology, and admission infections. NIs were associated with inpatient mortality (31.9% vs. 8.1%, p<0.001) and liver transplantation (4.8% vs. 1.9%, p<0.001). While the RF model (AUC 0.69) showed good calibration (Brier score 0.09), outperforming XGBoost, NN, and LR models (AUC 0.66 for all; LR comparison p = 0.043), no model achieved AUC ≥0.80 for clinical utility. At 10% predicted probability threshold, RF model demonstrated only 75.4% sensitivity, 52.9% specificity, and 15.9% PPV. CONCLUSIONS: NIs cannot be accurately predicted from day-of-admission data using ML models, even in a large, prospective, global cirrhosis cohort. Every hospitalized patient with cirrhosis should receive protocolized infection-control measures.

The American Journal of Gastroenterology
Hebei Medical University (CN), Tongji University (CN), University Malaya Medical Centre (MY), Fujian Medical University (CN), All India Institute of Medical Sciences Raipur (IN), Universidade Federal do Rio Grande do Sul (BR), Central South University (CN), University of Nottingham (GB), Ankara University (TR), Sun Yat-sen University (CN), Chulalongkorn University (TH), Guangxi Medical University (CN), University of Freiburg (DE), Sanjay Gandhi Post Graduate Institute of Medical Sciences (IN), Pontificia Universidad Católica de Chile (CL), Christian Medical College, Vellore (IN), Capital Medical University (CN), University of Toronto (CA), Chennai Mathematical Institute (IN), Universidad de Guadalajara (MX), Jilin University (CN), Akdeniz University (TR), Singapore General Hospital (SG), Fudan University (CN), Wenzhou Medical University (CN), Royal Perth Hospital (AU), Tel Aviv Sourasky Medical Center (IL), Royal North Shore Hospital (AU), Royal Berkshire Hospital (GB), First Affiliated Hospital of Wenzhou Medical University (CN), Hunter Holmes McGuire VA Medical Center (US), Jaslok Hospital (IN), Sir Charles Gairdner Hospital (AU), Hospital Italiano de Buenos Aires (AR), Baylor University Medical Center (US), University of Malaya (MY), Ege University (TR), Third Hospital of Hebei Medical University (CN), Hospital de Clínicas de Porto Alegre (BR), Instituto Nacional de Ciencias Médicas y Nutrición Salvador Zubirán (MX), Shanghai University of Traditional Chinese Medicine (CN), Second Hospital of Shandong University (CN), All India Institute of Medical Sciences Bhopal (IN), Hospital General Dr. Manuel Gea Gonzalez (MX), Mengchao Hepatobiliary Hospital (CN), All India Institute of Medical Sciences Bhubaneswar (IN), Apollo Hospitals (IN), Institute of Liver and Biliary Sciences (IN), First Hospital of Jilin University (CN), First Affiliated Hospital of GuangXi Medical University (CN), Hospital Geral de Bonsucesso (BR), Zhongshan Hospital (CN), The First Affiliated Hospital, Sun Yat-sen University (CN), Wuhan Union Hospital (CN), Centro Médico ABC (MX), Asian Institute of Gastroenterology (IN), Westmead Institute (AU), Hospital das Clínicas da Faculdade de Medicina da Universidade de São Paulo (BR), Westmead Institute for Medical Research (AU), Mayo Clinic in Florida (US), Beijing YouAn Hospital (CN), Northwestern Medicine (US), Shuguang Hospital (CN), Hôpital Mustapha Pacha (DZ), Baku Engineering University (AZ), Centre Hospitalier Universitaire de Cocody (CI), St. Paul's Hospital Millennium Medical College (ET), The Fifth People’s Hospital of Suzhou (CN), Post Graduate Institute of Medical Education and Research (IN), Gaziantep University (TR), All India Institute of Medical Sciences (IN), Virginia Commonwealth University Medical Center (US), Marmara University (TR), University of Pennsylvania (US), University of Birmingham (GB), University of Hong Kong (HK), Philadelphia University (US), University of Edinburgh (GB)
No poverty
Openalex Percentile: Top 20%
Liver Disease and Transplantation
DOISourceOpenAlex
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.