Deep learning and host variable embedding augment microbiome-based simultaneous detection of multiple diseases

Abstract

Microbiome has emerged as a promising indicator or predictor of human diseases. However, previous studies typically labeled each specimen as either healthy or with a specific disease, ignoring the prevalence of complications or comorbidities in actual cohorts, which may confound the microbial-disease associations. For instance, a patient may suffer from multiple diseases, making it challenging to detect their health status accurately. Furthermore, host phenotypes such as physiological characteristics and lifestyles can alter the microbiome structure, but this information has not yet been fully utilized in data models. To address these issues, we propose a highly explainable deep learning (DL) method called Meta-Spec. Using a deep neural network (DNN) based approach, it encodes and embeds the refined host variables with microbiome features, enabling the detection of multiple diseases and their correlations simultaneously. Our experiments showed that Meta-Spec outperforms regular machine learning (ML) strategies for multi-label disease screening in several cohorts. More importantly, Meta-Spec can successfully detect comorbidities that are often missed by regular ML approaches. In addition, due to its high interpretability, Meta-Spec captures key factors that shape disease patterns from host variables and microbial members. Hence, these efforts improve the feasibility and sensitivity of microbiome-based disease screening in practical scenarios, representing a significant step towards personalized medicine and better health outcomes.