Shaw, Jawaid

INTRODUCTION: Nosocomial infections (NIs) 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, 3 ML approaches (random forest [RF], extreme gradient boosting, and neural networks [NNs]) were used to predict NI. Data were split 80:20 for training and testing stratified by the outcome. Models were compared using area under the receiver operating characteristic curve (AUC). RESULTS: In total, 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 of postadmission. 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). Although the RF model (AUC 0.69) showed good calibration (Brier score 0.09), outperforming extreme gradient boosting, neural network, 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, the RF model demonstrated only 75.4% sensitivity, 52.9% specificity, and 15.9% positive predictive value (PPV). DISCUSSION: 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.

Background: App-based technologies could enhance patient and caregiver communication and provide alerts that potentially reducing readmissions. However, the burden of App alerts needs to be optimized to reduce provider burnout. Aim: The purpose of this study was to determine subjective and objective burden of using the Patient Buddy App, a health information technology (HIT) on providers in a randomized multicenter trial, who completed a semi-quantitative Likert scale survey regarding training procedures, data and privacy concerns, follow-up details, and technical support. This randomized multicenter trial recruits cirrhosis inpatients and their caregivers, and randomizes them into standard-of-care, HIT (communication only via App) and HIT+visits (App+phone calls/visits) for 30 days after discharge. The alerts are monitored by providers through a central iPad. The reason(s) and number of alerts were recorded as the objective burden. A total of 1442 messages were sent as alerts from the 103 dyads (patient + caregiver) (n=206) randomized to HIT arms. The most common messages related to Hepatic Encephalopathy (HE) (high or low bowel movement=50% or orientation tests=37%). Twelve providers completed the surveys reflecting the following themes—92% and 100%, felt adequately trained and confident about educating the patients and caregivers before roll out of App and had no concerns related to data and privacy; 70%, felt that appropriate time was spent on pursuing reason for data not being logged; 60% each, had issues with availability of adequate technical support and connectivity. Conclusion: The Patient Buddy App randomized multicenter trial till date shows an overall favorable rating regarding training procedures/education, privacy concerns, and ease of message follow-up, from providers. However, it is important to gauge and address subjective and objective burdens of monitoring human resources in current and future HIT studies to avoid burnout and to ensure successful study completion.