Metagenomic fingerprints in bronchoalveolar lavage differentiate pulmonary diseases

1

A multimodal machine learning approach was developed to differentiate lung cancer from various pulmonary infections using mNGS data.
2

The study analyzed 402 bronchoalveolar lavage fluid samples, including lung cancer and three types of pulmonary infections.
3

Model VI achieved an AUC of 0.937 in the training cohort, indicating high diagnostic accuracy for distinguishing lung conditions.
4

A rule-in/rule-out strategy improved accuracy for differentiating lung cancer from tuberculosis, fungal, and bacterial infections.
5

Findings suggest mNGS-based analysis could serve as a cost-effective tool for early and accurate differential diagnosis in pulmonary disorders.

1