H. Cao, F. Zhou, M. A. Deherrera, M. Wu, Edwards Lifesciences, Irvine, CA
<>Conventional life prediction is based on deterministic approach using median survival data. Reliability analysis often makes use of log-normal or Weibull distribution in order to derive a reliability-based life prediction. In particular, Weibull distribution has been regarded as the most appropriate function to describe fatigue life. In implantable medical device application, the reliability requirement is high and the median survival data may not provide the necessary confidence for some life-critical structure applications. The current study is undertaken to provide a rigorous sensitivity analysis on the fatigue life distribution function and to compare the model with carefully controlled experimental data. In particular, the focus is on the examination of the most suitable life distribution function. The overall goal is to present a statistical framework for predicting the durability performance of structure-critical medical devices with high confidence that is required by the clinical practice and consistent with international standards and regulatory requirements. <>The experimental program consists of fatigue testing of superelastic Nitinol component specimens at conditions of relevant mean and alternating strains in a simulated physiological environment. Testing would be conducted using a sufficiently large sample population of statistical significance up to very high cycles to satisfy the clinical requirements for implantable medical device applications. Fatigue stresses in the specimen are determined using ABAQUS finite element method with an appropriate superelastic material model. The specimen fatigue life data are analyzed using a maximum likelihood estimation methodology with various statistical distribution functions. This provides a practical methodology for life analysis involving the application of superelastic material in implantable devices where long-term structural integrity and reliability are desired.
Summary: The current study is undertaken to provide a rigorous sensitivity analysis on the fatigue life distribution function and to compare the model with carefully controlled experimental data. In particular, the focus is on the examination of the most suitable life distribution function. Testing would be conducted using a sufficiently large sample population of statistical significance up to very high cycles to satisfy the clinical requirements for implantable medical device applications. Fatigue stresses in the specimen are determined using ABAQUS finite element method with an appropriate superelastic material model. The specimen fatigue life data are analyzed using a maximum likelihood estimation methodology with various statistical distribution functions. The overall goal is to present a statistical framework for predicting the durability performance of structure-critical medical devices with high confidence that is required by the clinical practice and consistent with international standards and regulatory requirements.
