Clinical Research on Microecological Landscape for Infection Risk Stratification in Newly Diagnosed Patients with...

Area of Science:

• Microbiology
• Genomics
• Computational Biology

Background:

• Infections are a major risk for patients with hematological diseases undergoing chemotherapy.
• Antimicrobial resistance is exacerbated by non-specific antibiotic prophylaxis in neutropenic patients.
• Current infection risk screening methods are limited.

Purpose of the Study:

• To develop a prediction model for infection risk stratification in newly diagnosed hematological patients.
• To identify microbial features associated with neutropenia and infection risk.
• To guide precise preventive strategies and reduce antibiotic resistance.

Main Methods:

• Plasma metagenomic next-generation sequencing in 230 hematological patients.
• Analysis of microbial community profiles.
• Machine learning (random forest) for classifier construction and risk prediction.

Main Results:

• Distinct microbial features identified in neutropenic patients.
• Random forest model accurately predicted neutropenia (AUC=0.8324).
• Microorganism-based model predicted infection risk (AUC=0.942), improved with clinical data (AUC=0.953).

Conclusions:

• A microorganism-based prediction model offers effective infection risk stratification for hematological patients.
• Early identification of high-risk patients enables targeted interventions.
• The model can reduce prophylactic antibiotic use and mitigate antimicrobial resistance.