INSIDE

Total knee arthroplasty (TKA) is the most common joint replacement surgery, with revision rates varying by surgical approach. Postoperative dissatisfaction (more frequent than after hip replacement) increases further in revision cases: it is often linked to persistent pain or limited

function, significantly impacting quality of life. Beyond infection risks, dissatisfaction has been associated with improper implant positioning, joint instability, mechanical complications (loosening, wear), unbalanced surrounding tissues, and patella overloading.

Postoperative follow-up, typically based on clinical exams and imaging, could be improved to better anticipate risks and adapt patient-specific treatments earlier. Biomarkers, useful for detecting inflammation, loosening, or wear, can be combined with other diagnostics. While wearable technologies allow ongoing health monitoring, implant-embedded sensors present regulatory challenges. Machine learning offers promise but demands large, coherent, high-quality patient datasets (still rare). Musculoskeletal (MSK) and finite element (FE) mechanical models are effective for predicting joint and patella loads, yet their clinical use is limited by computational cost and processing delays.

INSIDE proposes to center the digital twin (DT) on the implanted prosthesis within its human environment, leveraging existing multi-scale models to improve prediction of patient dissatisfaction. The project unites research teams with expertise in patient monitoring across different levels: patient / clinical, joint / implant, bone, and cellular. These multi-scale, multi-fidelity data will be merged using metamodels that account for both measurement and simulation uncertainty. Supported by a Technical Industrial Center and a DT company, INSIDE will develop a knee prosthesis DT updated with patient-specific data to predict complications early.

Instead of mining large datasets, INSIDE will integrate data across different levels, filling gaps with well-established knowledge. To generate sufficient comprehensive data, the project will focus on high-risk revision TKA patients, combining retrospective and prospective cohorts. Three technical work packages (WPs) will collect data and compute biomarkers. WP1 will gather clinical routine data (including medical imaging), enhanced by motion analysis of daily activities and biological samples (blood, synovial fluid, bone explants). WP2 will build a patient-specific MSK model using bone shape and motion data to estimate joint, ligament, and implant loads. A FE model, validated on implant mechanical testing with biomimetic lubricants, will predict wear and bone-implant interface pressure. WP3 will investigate biological responses to dynamic mechanical stimuli, measuring inflammation and wear biomarkers and analysing bone remodelling capacity using a FE model fed by bioreactor tests on bone explants. WP4 will integrate all scales to identify patterns and generalisable risk markers. Metamodels will enable rapid, uncertainty-informed prediction of post-op risks, reducing reliance on full mechanistic simulations. WP5 will ensure project credibility through training on model Verification & Validation, variability / uncertainty quantification and propagation, and FAIR-compliant data and frameworks to maximize sharing.

Surgeons and patients will be involved throughout the project to understand the benefit of using a DT in post-op follow-up. Finally, INSIDE aims to reduce post-operative complications and improve patient satisfaction. Its results will support better patient care, including during pre-op planning.

Scientific knowledge and data generated will be shared to strengthen collaborations and generalise outcomes to other orthopaedic implants. Establishing that dissatisfaction can be predicted from a small yet comprehensive cohort could reshape how multi-scale models are built for surgical outcome prediction.