IRP AI for Broadband Spectroscopy

Title: AI for broadband spectroscopy: How machine learning can help to find biomarker fingerprints to determine kidney haemodialysis efficiency?

Duration: January 2024 - December 2024
Researchers: Roderik Krebbers, Amir Khodabakhsh, Simona M. Cristescu
Funding: Interdisciplinary Research Platform
Project partners: IPG Photonics Corporation, Radboudumc

Loss of kidney function is a life-threatening condition typically treated with haemodialysis or a kidney transplant. Chronic kidney disease often goes unnoticed until there is significant impairment in kidney function. Therefore, early detection of kidney failure is crucial. Unfortunately, there are currently no established biomarkers that allow early identification of deteriorating kidney function. However, evidence suggest that certain compounds associated with deteriorating kidney function are present in exhaled breath. As breath analysis is non-invasive, this approach is particularly feasible in children. This project aims to develop a machine learning model capable of accurately obtaining the concentrations of various compounds in exhaled breath samples, by using the absorption spectrum. The model is expected to work even in the presence of noise and spectral interference.