Canonical text (the exact bytes fingerprinted)

TRUVACE RECORD VERSION
record: TRV-2026-0947
version: 1
kind: certified
reason: Certified into the record
timestamp: 2026-09-01T06:03:58.264149Z
status: published
lens: g_space
sector: health
headline: Explainable Plasma Proteomics-Based Machine Learning for Osteoporosis Diagnosis, Prognosis, and Protein Biomarker Discovery in the UK Biobank
dek: Osteoporosis (OP) is often underdiagnosed, highlighting the need for tools that can both detect existing disease and predict future risk; large-scale plasma proteomics combined with explainable machine learning enables integrated diagnostic and prognostic modeling while prioritizing clinically relevant protein markers. This study aims to develop and validate an explainable plasma proteomics machine-learning framework for osteoporosis diagnosis, future risk prediction, and biomarker discovery. We further tested w…
gain_title: Plasma proteomics XGBoost models with SHAP-selected markers improved discrimination of prevalent osteoporosis and prediction of incident osteoporosis from baseline samples in UK Biobank.
problem_title: (none)
trace_subject: (none)
gain_reading: Plasma proteomics XGBoost models with SHAP-selected markers improved discrimination of prevalent osteoporosis and prediction of incident osteoporosis from baseline samples in UK Biobank.
gain_evidence: both the diagnostic and prognostic XGBoost models showed robust discrimination for osteoporosis status and future risk, respectively. | Using only these SHAP-selected protein markers, the XGBoost model outperformed the full-proteome models and provided robust, simultaneous diagnostic and prognostic prediction of osteoporosis.
problem_reading: (none)
problem_evidence: (none)
quick_read: Using UK Biobank plasma proteomics, researchers developed SPX-OP, an explainable machine-learning framework that separately models prevalent osteoporosis and incident osteoporosis with XGBoost and SHAP, then evaluates a combined marker panel for baseline stratification. By publication date 2026-09-01, both diagnostic and prognostic models showed robust discrimination, and a model using only SHAP-selected proteins outperformed full-proteome models.

The work matters because osteoporosis is often underdiagnosed and needs tools that detect existing disease and predict future risk; transforming high-dimensional proteomic data into interpretable markers like FSHB, ADIPOQ, SOST, COL9A1, and CHAD could improve risk stratification. Uncertainty remains about generalizability beyond UK Biobank and clinical validation of the plasma biomarkers.
limitation: Findings require further validation before clinical use as plasma protein biomarkers for osteoporosis risk stratification.
tag: Evidence-backed gain
key_points: Developed SPX-OP framework that separately models prevalent OP and incident OP using XGBoost and SHAP on UK Biobank plasma proteomic data. | SHAP-derived markers including FSHB, ADIPOQ, SOST, COL9A1, and CHAD were linked to osteoporosis and enriched in bone development and remodeling and extracellular matrix organization pathways. | Model using only SHAP-selected protein markers outperformed full-proteome models for simultaneous diagnostic and prognostic prediction.
rundown: Researchers used UK Biobank plasma proteomic data to build SPX-OP, an explainable framework that trains separate XGBoost classifiers for prevalent and incident osteoporosis and applies SHAP to prioritize markers, then tests whether the union of markers supports integrated baseline stratification of normal, prevalent, and future incident states.

The SHAP-selected panel included proteins such as FSHB, ADIPOQ, SOST, COL9A1, and CHAD, which were enriched in bone-related pathways involving bone development and remodeling, extracellular matrix organization, and inflammatory processes, and enabled a reduced-feature model to outperform full-proteome models.
sources:
- peer_reviewed | The FASEB Journal | https://doi.org/10.1096/fj.202600647r | 2026-09-01
prev: 0000000000000000000000000000000000000000000000000000000000000000

sha256

f03b1f5a988c11cb6112a50ebd70e0a0cb09f092bf71b54d87b6601ba0fe948d

previous

0000000000000000000000000000000000000000000000000000000000000000