Hybrid Intelligent Modeling And Performance Optimization Of Biodegradable Ureteral Stents

Ureteral stents is a type of tubular medical devices that can dilate and supportobstructed or impaired ureters, which connect kidney and blade, to duct urine. Biodegradable ureteral stents, which can be completely degraded in body, and then be discharged in vitro with urine after recovery of the impaired ureter to avoid reoperation, have increasingly gained attention. Key features of biodegradable ureteral stent, including biocompatibility, mechanical robustness, and controllable biodegradability, are also critical topics in this research area. This dissertation focuses on analysis and optimization of mechanical properties of biodegradable ureteral stents via modeling of the relationship between mechanical properties of each component and that of bicomponent stents.(1) Based on the mechanical properties of current biodegradable stents and relevant clinical requirements, bicomponent stentswere braided using poly(lactic-co-glycolic acid)(PGLA) and poly(glycolic acid)(PGA). Parameters, such as weight ratio of each component and physical structurewere investigated.(2) Resistance to radialcompression determines the supporting effect of ureteral stents. Effects of physical structures and weight ratio of components on the radial compressionalproperties of bicomponent ureteral stents were studied for modeling of radial compression process of bicomponent ureteral stents with different structures braided with different braiding angles. Particle Swarm Optimization(PSO) algorithm was selected to obtain optimal ratio of component and braiding angles. Based on the parameters of the model of each experimental samples group,five cross validation was used to verify and obtain generalization model of ureteral stents.Experimental data was adjusted according to feedback toobtain generalization model with higher precision based on PSO algorithm. Using this generalization model, effect of braiding angleon radial compression process was quantified and justified using experimental data. This generalization model demonstrated higher precision on predicting radial compression properties of bicomponent ureteral stents.(3) In order to be implanted and fixed facilely, ideal ureteral stents should have high tensile modulus and tensile strength. Otherwise, stents could not function properly to duct urine, and could even block urethra if they break earlier than expected. Moreover, the stents should have acceptable tensile properties as they should be pulled out without breakage if inflammation or rejection reaction occurs after implantation.Effects of physical structures and ratio of components on the axial tensile properties of bicomponent ureteral stents were studied, and relevant models to establish relationship between physical structures, braiding angle and tensile properties, were developed. PSO algorithm was used to optimize braiding angle and materials based on reference performance. The PSO optimized tensile model showed good simulation for the middle and latter steps of tensile process, but much worse for the initial steps. To solve the problem, idea of multi-species was introduced into the PSO model. Information was shared among differentmulti-speciesgroups for cooperative evolution. CPSO algorithm was then proposed to simulate the initial, middle and latte steps. The simulation results of CPSO algorithm were justified after comparing with experimental data.(4) HSVRM regression model was employed to simulate braiding process of bicomponent ureteral stents. The process of material mixing and braiding was simulated using SVR model, while the post-braiding thermal treatment was considered as a high level SVR after the input variable augmented.Based on a fixed rate of the test and the environment, the stress- strain relationship of under testing the sample was treated for a group of time series The models were proved effective. Since there are no standards to evaluate the mechanicalproperties of ureteral stents, specifications of commercial stents are considered as standards. Based on the standards, optimal ratio of materials of bicomponent stents was calculated using PSO algorithm. Comparing to lab experiments, the results from modeling could not only ensure terminal characteristics of final products, but also make the transition performance converge to the reference performance.