APASL‐ACLF Research Consortium–Artificial Intelligence (AARC‐AI) model precisely predicts outcomes in acute‐on‐chronic liver failure patients

Abstract

Acute insults on underlying chronic liver disease (CLD) result in systemic inflammation, perturbed homeostasis, multiorgan dysfunctions and high mortality in acute-on-chronic liver failure (ACLF) patients. [1][2][3] Although multiple definitions exist for this syndrome, 4 a common denominator includes high short-term mortality. 2,5 The Asia Pacific Association for the Study of the Liver (APASL) defines ACLF in CLD patients without prior decompensation as an acute hepatic insult resulting in jaundice, coagulopathy, ascites or encephalopathy within 4 weeks. 2 This constitutes around 25% of all incident-ACLFs among hospitalized cirrhosis in the USA. 6 The syndrome is dynamic with a reversible component where patients may recover or improve from this syndrome and may not require definite treatment. 2 On the flip side, the definite treatment, that is liver transplantation, is limited by resources, organ availability and expertise. 2 Thus, it is pertinent to predict the outcomes precisely, which will help stratify patients for an appropriate level of care, communicate prognosis and guide the allocation of resources. Although APASL-ACLF Research Consortium (AARC) model is a valid tool for outcome predictions, 7,8 about 20%-30% of ACLF outcomes are poorly predicted with this model. 7,8 Also other models lack the specificity of predictions in APASL-ACLF patients. 7,8 Hence, a need for precision in outcome predictions persists in these patients.Artificial intelligence (AI) is gaining considerable momentum in healthcare, and machine learning (ML), a field in AI, is concerned with developing models or algorithms to interpret complex/large data. The applications of AI in gastroenterology and hepatology are increasingly recognized. 9,10 Recently, an ML model-extreme gradient boosting (XGB) was demonstrated to perform better than logistic regression to predict mortality in ACLF patients from Europe. 11

Application of machine learning in the prognosis prediction of malignant large bowel obstruction: a two-cohort study

“…Moreover, the LDA showed excellent performance, with an AUC of 0.92 (95% CI: 0.64-0.90) for OS. Both classifiers have proven to be reliable within the medical field (34)(35)(36)(37). To further optimize the model, we integrated clinical genomics and radiomics data.…”

Machine Learning Based Diagnostic Paradigm in Viral and Non-Viral Hepatocellular Carcinoma

“…Pearson's correlation coefficient was used to quantify the relationship between every pair of HCCs and Non-HCCs samples [108]. The model trained on these features when deployed obtained 84.7% sensitivity and 88.4% overall accuracy [109]. The feature extraction using the Pearson's correlation coefficient resulted in better accuracy than without the it.…”