Machine learning as an artificial intelligence application in management of chronic hepatitis B virus infection

Abstract

Let’s review the role of gut microbiota in pathogenesis of chronic hepatitis B infection as addressed in by Zhu et al. Zhu et al used high-throughput technology to characterize the microbial ecosystems, which led to an explosion of various types of molecular profiling data, such as metagenomics, metatranscriptomics, and metabolomics. To analyze such data, machine learning (ML) algorithms have shown to be useful for identifying key molecular signatures, discovering potential patient stratifications, and, particularly, for generating models that can accurately predict phenotypes. Strong evidence suggests that such gut microbiome-based stratification could guide customized interventions to benefit human health. Supervised learning includes designing an algorithm to fix a pre-identified problem. To get an answer, ML software must access data that have been nominated. On the other hand, unsupervised learning does not address any pre-defined problems. Bias should be eliminated as much as possible. In unsupervised learning, an ML algorithm works to identify data patterns without any prior operator input. This can subsequently lead to elements being identified that could not be conceived by the operator. At the intersection between supervised and unsupervised learning is semi-supervised ML. Semi-supervised learning includes using a partially labeled data set. The ML algorithm utilizes unsupervised learning to label data (that has not yet been labelled) by drawing findings from the labeled data. Then, supervised techniques can be used to solve defined problems involving the labeled data. Reinforcement learning, which is similar to supervised learning in the meaning, is goal-oriented. Reinforcement learning does not need labeled data, instead, it is provided with a set of regulations on a problem. An algorithm will carry out operations to try to answer questions involving the problem. Based on obtained data of gut microbiota, various therapeutic modalities can be applied: Prebiotics, probiotics, postbiotics, engineered bacteria, bacteriophage, and novel microbe-materials therapeutic system and fecal transplantation. In conclusion, ML is an artificial intelligence application that helps in providing new perspectives on tailored therapy. Furthermore, assessing the impact of gut microbiota modification is a critical step in advanced liver disease management. These new artificial intelligence techniques although promising, still require further analysis and validation in future studies.

Keywords: Artificial intelligence; Machine learning; Gut microbiota; Hepatitis B virus; Infection

Core Tip: There is substantial evidence indicating that stratification based on the gut microbiome could facilitate personalized interventions aimed at enhancing human health. It became essential to characterize the microbial ecosystems, resulting in a surge of various types of molecular profiling data, including metagenomics, metatranscriptomics, and metabolomics. In the analysis of such data, machine learning algorithms have proven to be effective in identifying crucial molecular signatures, uncovering potential patient stratifications, and especially in creating models that can reliably predict phenotypes. Machine learning may be supervised, unsupervised, semi-supervised or reinforcement type. Using a method for explaining individual classifier decisions for complex microbiota analysis may help in developing personalized treatment.