Machine learning framework for classification of Adenocarcinoma and Squamous cell carcinoma using lung microbiome dataset for their early diagnosis

Non-small-cell lung cancer (NSCLC) accounts for 80-85% of lung cancer (LC) cases, mainly classified into Adenocarcinoma (AC) and Squamous cell carcinoma (SCC). A late diagnosis at an advanced stage, high metastasic rate, and the development of therapy resistance are probable reasons for approximately 95% of mortality. Owing to high heterogeneity and variances in subtypes, it is important to precisely classify them for treatment. The lower airways harbour a dynamic microbial population, however the disruption in the homeostasis of microbiome compositions was found to be correlated with the increased risk of LC. Therefore, we aim to classify AC and SCC using lung microbiome with machine learning (ML) and deep learning (DL) based algorithms for early diagnosis and prognosis to enhance survival. For this purpose, we used 149 and 145 lung tissue samples of AC and SCC patients, respectively. Additionally, the data on patient age, sex, smoking history, and environmental material was also considered. We then fitted a ML and DL models to identify NSCLC subtypes using microbiological data. ML-based models were developed to discriminate NSCLC subtypes based on their microbial information. Consequently, 17 features were identified as biomarkers for distinguishing AC from SCC with an accuracy of 81% in K- nearest neighbour and 71% in Deep neural network. Overall, this study proposed a framework where we can rely on taxonomic features along with ML and DL techniques to classify overlapped AC and SCC metagenomic data providing lung microbiome as a predictive and diagnostic biomarker in LC.