CardioNIR investigates whether near-infrared spectra of dried blood spots, combined with proteomic and metabolomic profiling, can capture the dynamic metabolic state of an animal during cardiopulmonary bypass — and whether those signatures translate into a usable, low-burden intraoperative monitoring tool.
Headline empirical results so far include an R² of 0.81 for anion gap from NIR alone, a metabolic exhaustion threshold around 35–40 minutes on bypass, and a reusable, open-source 22-module preprocessing and modelling pipeline. The project sits at the intersection of analytical chemistry, surgical physiology, and high-dimensional statistical learning.
The four sub-analyses below decompose the work into:
- Pipeline — the open-source preprocessing and modelling framework.
- NIR-PAD — peripheral arterial disease modelling from NIR spectra of dried blood spots, including replicate averaging and conformal prediction.
- Proteomics — Olink CVD III analysis of 92 cardiovascular proteins across bypass phases.
- Metabolomics — UHPLC-QqTOF metabolomic phenotyping and integration with the spectral and proteomic layers.
CardioNIR — Olink Proteomics
CardioNIR
Proteomics
Olink
Swine Model
CardioNIR — Predictive Modeling Pipeline
CardioNIR
Machine Learning
tidymodels
Pipeline
R
CardioNIR — Trans-Myocardial Metabolomics
CardioNIR
Metabolomics
LC-MS
Swine Model
Network Analysis
No matching items