Project Details
Projekt
SP8 Multidimensional analytical approaches to improve LBP diagnosis and prognosis
Subject Area
Orthopaedics, Traumatology, Reconstructive Surgery
Term
since 2026
Project identifier
Deutsche Forschungsgemeinschaft (DFG) - Project number 439742772
Subproject 8 (SP8) is a newly initiated research project in the second funding period, formed in response to the scale and complexity of multidimensional data acquired. To ensure improved methodological rigor, interdisciplinary collaboration, and optimal resource allocation, an independent project structure is required. SP8 will apply advanced machine learning (ML), artificial intelligence (AI), and statistical methodologies to generate clinically actionable insights for the diagnosis and prognosis of chronic low back pain (cLBP). In preliminary work using first-period data, we developed an AI framework that demonstrated the diagnostic superiority of key multimodal features, particularly disc degeneration, mobility measures, and psychosocial factors, relative to broader, unfiltered variable sets. Building on this foundation, SP8 aims to sustain and extend state-of-the-art analytical and statistical oversight across subprojects, enhancing the robustness and reproducibility of findings. The overarching objective is to develop more personalized and targeted diagnostic tools through classification, subgrouping (clustering), and prognostic modelling, while providing methodological support for study design and data analysis across the research unit. We hypothesize that (1) integrating multimodal data via advanced ML will improve cLBP diagnosis and produce clinically robust tools that surpass single-modality approaches; (2) data-driven clustering will identify clinically distinct cLBP subgroups, characterized by functional, structural, and psychosocial profiles associated with pain, disability, and movement impairments; (3) the combination of key multimodal features and derived subgroups will yield interpretable and effective prognostic models for cLBP progression and onset; and (4) sophisticated statistical techniques, such as functional data analysis, will enable identification of subgroups based on longitudinal activity patterns, circadian deviations, and gait cycles. We will employ an ensemble stacking AI framework to expand our initial classifier, identify core diagnostic features, and enable external validation of subgrouping and prognostic models using independent datasets. In parallel, we will implement robust data management infrastructure in accordance with FAIR principles to support data sharing, interoperability, and reproducibility. Additionally, a web-based visualization platform will be developed to allow clinicians to interactively explore subgroup structures and prognostic outcomes. Through methodological rigor, external validation, and open science principles, SP8 will contribute to advancing precision diagnostics and global care strategies for individuals with cLBP.
DFG Programme
Research Units
