Int J Mol Sci. 2024 Aug 6;25(16):8586. doi: 10.3390/ijms25168586

PMID: 39201273

Jorge J Cebolla, Pilar Giraldo, Jessica Gómez, Carmen Montoto, Javier Gervas-Arruga

Highlights: This study aims to identify protein biomarkers for the early detection of skeletal complications in patients with Type 1 Gaucher disease using machine learning. The research identified 18 potential biomarkers, emphasizing that some of these can be measured through blood tests. The goal is to develop new tools for clinical decision support systems using machine learning and systems biology techniques.

Background: Gaucher disease is a rare autosomal recessive disorder caused by glucocerebrosidase deficiency. It is closely associated with skeletal complications, significantly impacting patients' quality of life. Developing early diagnostic methods has the potential to improve treatment processes. Currently, imaging studies and genetic testing are considered the gold standard; however, identifying protein biomarkers for early diagnosis is a highly promising and forward-looking approach. Machine learning and artificial intelligence have revolutionized biomarker discovery. By uncovering meaningful patterns in complex datasets, machine learning plays an innovative role in biomarker discovery.

Objective and methods: The study used the patented Therapeutic Performance Mapping System (TPHS), which integrates systems biology, machine learning, and pattern recognition techniques. A comprehensive literature review was conducted through PubMed and Medline to identify candidate proteins. Additionally, gene expression data were gathered from the NCBI Gene Expression Omnibus, ArrayExpress, and Omics Discovery Index databases. Statistically significant genes were identified and mapped to candidate proteins. Findings were mathematically modeled using the TPHS system. Candidate proteins were filtered using various statistical methods, and the final candidates were analyzed using the Human Protein Atlas database to identify proteins measurable in urine or blood.

Results: The study identified 18 protein biomarkers as potential candidates, including PDGFB and PTH, which can be measured through blood tests. Gaucher disease is a condition where early diagnosis is critical for effective treatment. Although existing methods provide highly accurate diagnostic results, biomarkers measurable in blood or urine are essential for even earlier detection. This kind of machine-learning-based approach highlights the importance of single or combined protein biomarkers for early diagnosis. The use of artificial intelligence to uncover protein markers that can enable early diagnosis of Gaucher disease and other rare diseases is a highly important and promising application.