Multi-objective Optimize The Medicine Synthesis Of Central Experiment Design Based On Micro-genetic Algorithms

There are a lot of multi-objective optimize problems in the research of medicine and pharmacy, such as the optimum composition proportion in the process of medicine synthesis, the optimal extraction of medicine effective component the optimum allocation of public health resources, which is called policy-making criterion analysis. However, only by the scientific experimental designs, one or a group of optimal solutions can be found. Taking a scientific experiment means choosing rational test points and experimental designs, acquiring the optimal experimental conditions by statistic analysis so as to reach high yield, high quality and low consumption. Just as Atkinson and Donev point out that a well-designed experiment is an effective method to cognize the world.In multi-factor and multi-level experimental research, factorial designs based on the ANOVA model are full-test which analyses each factor in each level crosswise. This kind of method not only tests difference effect of each factor in each level but also examine the interaction among the factors. However, it is difficult to achieve experiment in reality when test number increased due to multi-factor or multi-level influence.Therefore, segmental factor experiment which picks out some typical test points derives from full-test. In1950s, Genichi Taguchi, the famous Japanese statistician, put forward orthogonal design, likewise, central experiment design which is used to fit Second-order response surface was invited in the sane time.As a highly effective experiment method, central experiment design could find the optimal test battery rapidly because of its scientific and efficient property. While confined by the level of previous experiment factors, it only show a roughly standard for each factor. If you want to find the best test battery when the objectives are optimal, you should choose multi-objective optimums in the process of contral. In traditional optimal ways, such as weighting or contouring map, multi-objectives are converted into a single objective or a series of objectives which also have larger subjectivity. However, micro-GA has global and efficient characteristics which could solve the disadvantage of the traditional multi-objectives optimization process. So in this topic we use the SGALAB(beta5.0) toolbox for Matlab by Yi Chen to evaluate the effects of micro-GA and test program in multidimensional solution and search the optimal test condition in the medicine synthesis of central experiment design and compare the optimizational effects between micro-genetic algorithms(micro-GA) and conventional optimization method. There are4parts of this dissertation:Part1:Introduce central experiment design and micro-genetic algorithms’(micro-GA) theory. Central experiment design is a series of design procedures which are suitable for much factors or levels. Central composite design is mainly used in5-level multi-factor experiment, Box-Behnken design is mainly used in3-level multi-factor experiment; Micro-GA has three types of elites which keep the diversity of population and noninferiority of solutions.Part2:Applicable scope and conditions of central experiment design, such as central composite design that includes central composite circumscribed, central composite inscribed, central composite face-centered and Box-Behnken design.Part3:Take a program test to micro-GA Multidimensional solutions. Test micro-GA by complex two-objective optimization test function.The results are Micro-GA is reliable in theory and the procedure is feasible by testing functions and could be used to resolve complex multidimensional solution problems.Part4:Use micro-GA coded by the Matlab2009a to optimize simulated data, pick out the optimal experiment conditions. Compare with the optimization effects between micro-GA and conventional optimization method.Use micro-GA to optimize the medicine synthesis condition of5-FuMS for chemo-embolization which includeing three mutually competed objectives-geometric mean diameter is above30nm, drug loading maximum, loading efficiency maximum, the values of factors such as gelation concentration,amount of emulsifiler,stirring speed are207.77mg·ml-1、0.27mg·ml-1、357r·min-1,the value of response such as geometric mean diameter, drug loading, loading efficiency are34.67μm,8.71%,77.09%. Compared with contouring map, loading efficiency increased by0.57%, geometric mean diameter increased by1.75μm,so the medicine synthesis condition for micro-GA is much more better.Use micro-GA to optimize the medicine synthesis condition of CLA for the intravenous delivery based on BBD, which includeing three mutually competed objectives-particle size(nm) minimum, ζpotential(mV) maximum, distribution (%) maximum, the values of factors such as tocopherol succinate, poloxamer188,0.1M NaOH are68.25%,0.52%,15.26%,the value of response such as particle size, ζpotential, distribution are135.75nm,33.67mV,98.12%. Compared with desirability function D, ζpotential increased by2.43mV,which amount to8.47%,CLA distribution increased by0.8%,so the medicine synthesis condition for micro-GA is much more better for experimentalist.To sum up, central experimental design are suitable for multi-factor or multi-level experiments. Micro-GA is reliable in theory and the procedure is feasible by testing functions; Micro-GA could search more optimal medicine synthesis condition than traditional optimal methods based on central experiment design.