All researchProject overview

Spectroscopic sensing · 2026

Sepsis recognition and pathogen identification with SERS

Evaluating analyte-enriched blood spectra with super-operational neural networks for two distinct classification tasks.

Qatar University · Qatar Research, Development and Innovation Council NPRP12S-0224-190144

Workflow illustration. Blood from sepsis and control subjects is centrifuged to enrich the analyte, which is placed on silver nanorod slides for SERS acquisition. The spectra are preprocessed and classified to separate control samples from five sepsis-causing bacterial strains.
Figure 1 from Hassan et al. (2026). Sample collection and analyte enrichment (a), followed by SERS acquisition, preprocessing and classification (b).View full-size figureSource articleCC BY 4.0

Approach

Can SERS spectra from analyte-enriched blood recognise sepsis and identify the pathogen responsible?

The workflow pairs analyte enrichment and surface-enhanced Raman acquisition with SuperRamanNet, a one-dimensional classifier. The study used 653 blood samples for the primary analysis and an independent external cohort of 70 samples. It evaluated sepsis recognition and pathogen identification as separate tasks.

  1. Enrich the analyte from blood
  2. Acquire and preprocess SERS spectra
  3. Classify with SuperRamanNet

Reported results

Five-fold validation kept spectra from each blood sample together. Pathogen typing was also evaluated in the external cohort.

Sepsis recognition accuracy in cross-validation
99.67%
Six-class pathogen accuracy in cross-validation
98.84%
External pathogen typing accuracy
98.28%
Two scatter plots, t-SNE and UMAP, of SuperRamanNet embeddings. Points form six separate clusters, one for each class: E. coli, K. pneumoniae, P. aeruginosa, S. aureus, S. epidermidis and negative controls.
Figure 5 from Hassan et al. (2026). t-SNE (a) and UMAP (b) projections of SuperRamanNet embeddings show six well-separated clusters.View full-size figureSource articleCC BY 4.0

My contribution

I developed the computational workflow from spectra and preprocessing through data splitting, training, evaluation and interpretation, and helped document the experimental protocol.

Study scope

The evaluation concerns the sample preparation, acquisition, labels and cohorts described in the paper. Clinical deployment would require further validation.