VALIDATION POLICY

Model validation is essential in order to assure the reliability of simulation and the effectiveness of the use of the simulation tools.

Since only customized implant-joint constructs are modelled, the model responses must be tested case by case. This may lead to a considerable validation effort, but once validated for a particular set of conditions, the models can offer an unrivalled prognostic potential, specially for complex clinical problems.

orthoSIM’s validation strategy is based first on demonstrating the mechanical reliability of the virtual implant-joint constructs and then demonstrating the sensitiveness of these models to distinct clinical outcomes.

The first is done by means of mechanical tests in the lab, whereas the latter is done by means of both restrospective and prospective trials with patients.

The clinical validation process is a continuous process, since more and more patient cases shall be included in the validation, in order to estimate more accurately the effectiveness of each model. Clinical validation is usually performed by the model’s researchers or owners.

+ Learn about the Scientific Committee in orthoSIM

However, all validation tasks are supervised by an independent scientific committee with prestigious members of the scientific community selected from different European countries.

orthoSIM has a strict model certification policy that all models owners must comply with, in order to be selected and remain in the simulation service catalogue.

In order to be integrated into the orthoSIM platform, biomechanical models experienced a series of validation steps. In order to serve for clinical purposes they must be continuously validated according to a specific validation policy.

Validation policy

The model’s owners will submit a protocol detailing strategies for validating their models. This validation will refer both to the models of implants and to the models of parts of the human body.

This protocol must follow the validation requirements suggested (see hereafter) and must be approved by the orthoSIM’s Scientific Committee.

The ‘in vitro’ validation of the models consists of the comparison of their response with the behaviour of specimens tested. The ‘in vivo’ validation, by comparison with real clinical cases, is necessary for validation of the sensitiveness of these models to distinct clinical outcomes.

Different levels of validation (from level 0 to level 3) can be considered for any model. Integration of a new model into the orthoSIM portal requires the achievement of a minimum level of validation which depends on the purpose of the model.

• Level 0: Implants models will help manufacturers in analysing the behaviour of their implants by means of design simulations.

• Level 1: Instrumented bone and joint models will help implant designers and surgeons to compare the biomechanical behaviour of different configurations of implants.
• Level 2: The achievement of this validation level allows the models to offer simulation results for surgery planning, based on comparing the biomechanical behaviour of joints and implants with the real outcomes from patient cases.
• Level 3: The achievement of this validation level offers a superior degree of reliability in surgery planning.

The same model could pass through all validation levels in its integration’s life cycle, depending upon the number of patient cases tested. Clinically meaningful results can only be obtained from level 2 onwards.

The implant validation parameter selected is the overall system’s rigidity (K=F/d), where F is the maximum applied force or torque F and d is the displacement attained in the force application’s point.

The model is considered validated when the difference between experimental results and simulation results is lower than 10%.

Validation history

The spine model was developed and validated by ENSAM-LBM (Paris, France). The spine model has been parameterized to permit the simulation of the different pathologies and surgical injuries and to adapt the spine model to the patient spine morphology. The validation of the spine model has been done by comparing the in vitro test results and the model results under the same conditions. For the spine model validation the stiffness tests were done with cadaver specimens. Many different spine tests have been carried out to validate the pathologies.

The implant models have been generated and validated individually at the IBV (Valencia, Spain) for all the proposed implants. The implant model generation has been parameterized to allow the user to choose any geometry and configuration of the implant. For the implant model validation some implant tests have been done to measure its behaviour under different load modes. Once knowing the implant stiffness, the implant models have been adjusted until its behaviour was minimally different from the tests results.

Once implants and spine models had been validated separately, the models have been assembled and validated as an instrumented spine model, by comparing with the results of the in vitro tests of instrumented cadaver lumbar spines.

The validation of a FE Model consisted in a comparison of the behaviour of the model with the behaviour of the real implant system.

The chosen parameter for the validation of the spine implants is the global stiffness (K=F/d) of the implant configuration, where F is the maximum force or moment applied and d is the displacement of the application point of the load or the axial rotation of the assembly when the torque is applied.

The model is considered validated when the difference between the experimental and the model stiffness is lower than 10% for each load modes. Considering the lateral flexion behaviour is not the most relevant load mode (the standard doesn’t contemplate this load mode for the spine implant evaluation) and that it is the hardest stiffness to adjust, we have chosen as priority to validate the flexion-compression and torsion stiffness. And for some cases where the adjustments of the lateral bending worsen the other load mode stiffness, we consider the solution like validated even if the discrepancy for lateral bending is bigger than 10%.

The model is currently being clinically validated with real cases, both of good surgical practice and of failure. Results will be published in peer-reviews journals, in parallel with the market deployment of the service.