Eight out of ten drug treatments that work in mice fail in people. This project builds a computer model that bridges that gap for osteoporosis, a bone-weakening disease affecting millions. The problem is that mouse studies are poor predictors of human outcomes. Researchers currently rely on simple scaling to guess whether a treatment tested in mice will work in patients. This project creates two parallel "virtual twins"—one mouse, one human—that simulate how bones adapt to mechanical forces and drugs over time. The models are built from real data: medical images, cell measurements, and gait analysis. They predict which combinations of treatments—such as drug holidays or high-strain exercises—are most likely to strengthen bones in patients, without needing to test every option in expensive, lengthy, or ethically questionable animal and human trials. If successful, this framework could cut the failure rate of osteoporosis drugs entering clinical trials, saving time and money while getting effective treatments to patients faster. It could also be adapted for other musculoskeletal diseases. The work is applied, not fundamental—its goal is a practical tool for preclinical drug testing.
View original technical description
Eighty per cent of pharmaceutical interventions fail in patients even after being successful in animal studies. Musculoskeletal (MSK) diseases such as osteoporosis (OP) reduce dramatically the quality of life of millions of affected patients. Mice are the most common animal model to test new treatments. Nevertheless, the extrapolation of their effect onto patients and the identification of which new treatments should be tested in clinical studies is based on simple scaling approaches. In this project I will develop a new mechanistic computational framework that bridges between mouse and human, informed by in vivo experiments in mice, to discover optimal treatments in patients. I will create two parallel virtual mouse and human twins (VMHTs-OP), based on similar inputs (biomedical images, cell data, gait data) that will predict bone adaptation in function of biomechanical and/or biochemical stimuli. Each virtual twin will be based on advanced multi-scale computational models (multi-body dynamics, finite element and cell-population models) to predict bone adaptation over time and space due to OP and to new biomechanical and pharmacological treatments, identifying in silico the new combined treatments that are likely to be effective in patients, to be tested in future clinical trials. The models will be going through a comprehensive verification, validation and uncertainties quantification process in order to provide the required credibility for future preclinical applications. The model predictions will be validated against longitudinal mouse experiments and available longitudinal clinical data from known biomechanical or pharmacological interventions. Finally, the validated framework will be used to test in silico several combinations of treatments regimens (overlap, intermitted, drug holidays) and different interventions (microgravity, high strain exercises) that would not be ethically nor economically testable in animal and clinical trials.
Plain English summaries and category classifications on this site are generated by AI and may not perfectly reflect the original research.
Is something wrong? Let us know