A Quantitative Prediction Model For Albumin And Urea Synthesis Of In Vitro Liver Tissues Based On Gradient Boosting Decision Tree

In recent years,in vitro liver tissue culture systems have been extensively studied,with many remarkable achievements made.Constructing liver tissue in vitro is a relatively complex process,involving the selection of cells and materials,as well as the ongoing optimization of culture conditions.Most of the past published studies combined theoretical and practical experience to design and conduct experiments,thereby contributing valuable research data to the field of liver tissue construction.However,few studies have mentioned the degree of influence of specific culture parameters on the secretion of functional indicators,and most researchers have only provided the measured values of liver function indicators for specific days.The functional secretion values for unknown days remain unclear,and a comprehensive and quantitative analysis of functional secretion data from past studies is lacking.The aim of this study is to discern the correlation between manufacturing parameters and functional secretion values,analyze key manufacturing parameters,predict functional secretion under specific conditions,and quantify the functional maintenance time of the liver tissue culture system.Therefore,based on the different in vitro culture methods,this study divided the culture platforms of in vitro liver tissues into four systems: 2D well plate,3D scaffold,self-assembled liver spheroids,and liver-on-a-chip.We selected research published in the past ten years,and further screened the required articles by means of keyword counting and unit unification.Next,this study extracted the manufacturing parameters and functional index secretion values in the liver tissue construction process respectively from the text and figures of the articles,respectively.After converting them uniformly into machine values,we found a relationship between the manufacturing parameters and functional indicators.Through the analysis of the influence of key manufacturing parameters on functional secretion,we found that the manufacturing parameters of cells and culture systems contributed more to the secretion of albumin and urea.In predicting functional indicators under specific conditions,the regression model can accurately distinguish the specificity of the manufacturing parameters of different control groups from the same article in predicting functional indicators,and can better reflect the overall trend of functional indicators.In addition,based on the analysis of measured functional data points from all studies under the four culture systems,this study defined the optimal and the longest functional maintenance time of the liver tissue culture system.Upon examining a selection of studies exhibiting high predictive accuracy in their comprehensive regression models,it becomes evident that most liver culture systems-those based on 3D scaffolds,self-assembled liver spheroids,and liver-on-a-chip-reach their optimal functional window within a span of two weeks.Moreover,these systems typically begin to exhibit a drop in secretory function a few days thereafter.To sum up,our regression prediction model can infer the influence of different manufacturing parameters on the secretion of albumin and urea,and can reasonably predict the secretion levels under specific conditions.Consequently,the above-mentioned analysis and prediction of functional indicators can serve as a supplement and expansion of previous research data,providing a valuable reference for the optimization of culture parameters in subsequent research.