Synthetic blood-based infrared molecular fingerprints: artificial cohorts for methodological research
Leopold-Kerschbaumer, N.; Feiler, N.; Kepesidis, K.
Show abstract
Infrared molecular fingerprinting of human blood samples provides a powerful, minimally invasive approach for disease detection and health monitoring. However, ethical and legal constraints often limit the sharing of real patient data collected from clinical studies. In this work, we present a synthetic dataset of blood-based infrared molecular fingerprints, generated using multivariate Gaussian models fitted on real measurements from a large case-control study targeting various cancer types. The synthetic dataset retains the statistical and physical properties of real molecular fingerprints, enabling the development and validation of analytical methodologies without compromising patient privacy. We demonstrate that the provided artificial dataset can serve as a proxy for real data in methodological research, facilitating reproducibility and collaboration in biomedical spectroscopy. This approach offers a practical solution for overcoming ethical barriers in clinical data sharing in spectroscopic biomarker research.
Matching journals
The top 4 journals account for 50% of the predicted probability mass.
Similar papers in this journal
- RamanMAE: Masked Autoencoders Enable Efficient Molecular Imaging by Learning Biologically Meaningful Spectral Representations 94%
- Optical photothermal infrared imaging using metabolic probes in biological systems 93%
- Joint automatic metabolite identification and quantification of a set of 1H NMR spectra 93%
Similar papers in this journal
- An interactive mass spectrometry atlas of histone posttranslational modifications in T-cell acute leukemia 90%
- A peptide-centric quantitative proteomics dataset for the phenotypic assessment of Alzheimer's disease 89%
- A clinical microscopy dataset to develop a deep learning diagnostic test for urinary tract infection 89%
Similar papers in this journal
Similar papers in this journal
- Machine learning assisted mid-infrared spectrochemical fibrillar collagen imaging in clinical tissues 95%
- The effects of skin tone on photoacoustic imaging and oximetry 93%
- Application of Transfer Learning for Rapid Calibration of Spatially-resolved Diffuse Reflectance Probes for Extraction of Tissue Optical Properties 92%
"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.