Explainable Machine Learning for Longitudinal Multi-Omic Microbiome

Abstract

Over the years, research studies have shown there is a key connection between the microbial community in the gut, genes, and immune system. Understanding this association may help discover the cause of complex chronic idiopathic disorders such as inflammatory bowel disease. Even though important efforts have been put into the field, the functions, dynamics, and causation of dysbiosis state performed by the microbial community remains unclear. Machine learning models can help elucidate important connections and relationships between microbes in the human host. Our study aims to extend the current knowledge of associations between the human microbiome and health and disease through the application of dynamic Bayesian networks to describe the temporal variation of the gut microbiota and dynamic relationships between taxonomic entities and clinical variables. We develop a set of preprocessing steps to clean, filter, select, integrate, and model informative metagenomics, metatranscriptomics, and metabolomics longitudinal data from the Human Microbiome Project. This study accomplishes novel network models with satisfactory predictive performance (accuracy = 0.648) for each inflammatory bowel disease state, validating Bayesian networks as a framework for developing interpretable models to help understand the basic ways the different biological entities (taxa, genes, metabolites) interact with each other in a given environment (human gut) over time. These findings can serve as a starting point to advance the discovery of novel therapeutic approaches and new biomarkers for precision medicine.

The Promise of Explainable AI in Digital Health for Precision Medicine: A Systematic Review

“…Integrating explainable artificial intelligence with digital health data is gaining momentum in precision medicine, addressing the need for transparent and understandable models essential for clinical applicability [ 19 , 51 , 52 ]. As machine-learning models become more complex, interpretability is crucial in clinical contexts such as microbiome research [ 51 , 53 ].…”

“…Integrating explainable artificial intelligence with digital health data is gaining momentum in precision medicine, addressing the need for transparent and understandable models essential for clinical applicability [ 19 , 51 , 52 ]. As machine-learning models become more complex, interpretability is crucial in clinical contexts such as microbiome research [ 51 , 53 ]. Explainable artificial-intelligence applications can help predict an Alzheimer’s disease diagnosis in a pool of patients with mild impairments, showcasing how interpretable machine-learning algorithms can help explain complex patterns that inform individual patient predictions [ 54 , 55 ].…”

“…Methods of extracting explanations from complex models can aid in the discovery of new personalized approaches to therapy and new biomarkers [ 57 ]. For example, Bayesian networks may serve as a framework for visualizing interactions between biological entities (taxa, genes, metabolites) within a specific environment (human gut) over time [ 51 ]. A model agnostic approach to explainability is offered by Shapley additive explanations, which enhance understanding at both global and local levels, improve predictive accuracy, and facilitate informed medical decisions [ 56 , 58 ].…”

“…These challenges include transparency, trust, and interdisciplinary collaboration in advancing healthcare innovations. Explainable artificial intelligence has many qualities that bridge the gap between complex AI algorithms, such as deep learning, and their practical applications in healthcare, enhancing the acceptability and effectiveness of AI interventions in clinical settings [ 19 , 51 , 52 ]. By facilitating a better understanding of AI-driven predictions, explainable artificial intelligence enables healthcare professionals to make informed decisions, thus fostering a collaborative environment where AI serves as a supportive tool for opaque decision-making [ 40 , 45 ].…”

Recent advances in data- and knowledge-driven approaches to explore primary microbial metabolism

“…proposed a sparse neural encoder–decoder network which not only predicts metabolite abundances from microbiome data but also allows to interpret microbe–metabolite links from the hidden layer of the network [ 51 ]. Given the dynamic nature of microbiome composition in the human gut, several packages were developed to specifically analyze time-series metabolomics data, such as MDITRE [ 52 ] and CGBayesNets [ 53 ]. Both of these tools combine Bayesian approaches with deep learning to predict human-interpretable rules for host status given taxonomic information.…”