Online Detection Of Multiple Change Points And Association Analysis Of Pathological Data Based On Adaptive Random Strategy

In recent years,with the rapid development of information technology,a large amount of data has emerged in all walks of life.For example,finance,medical care,education and other industries generate a large amount of data every day.Enterprises can use these data for advertising push,market segmentation,data analysis and various product innovation.With the rapid growth of data volume,higher requirements are put forward for data processing methods.At present,data flow mutation detection technology has also gradually received widespread attention,but the traditional detection technology can not meet the current demand.Therefore,the online detection method is derived from this,and with the higher and higher requirements on the performance of data analysis methods,it has gradually become a hot topic in current research.At present,sliding window is one of the commonly used techniques in multi-break point detection of data stream.Using sliding window,sequential data stream can be divided into multiple sub-sequences,and the sub-sequences in each window are analyzed separately to achieve the purpose of multi-break point detection.In the sliding window model,the selection of the sliding window length will have a certain influence on the detection result of the mutation point.Too small sliding window length will increase the time consumption of the algorithm,and too large sliding window length will cover the differences between data flows.In addition,most of the traditional data stream mutation detection techniques are offline algorithms,which cannot detect data streams online.Based on offline algorithm and sliding window model,an online detection model of multiple mutation points based on adaptive online model and random overlapping model is proposed in this paper,and the multi-path pathological signal association network is constructed by using the proposed model,and the correlation between the signals is analyzed,which has achieved good results.First of all,this paper introduces TSTKS algorithm in detail,combined with the sliding window model can better detect multiple mutation points of data stream.On this basis,the buffer model is introduced,which uses the buffer to receive timing data in real time,and then puts the data into the data receiver.The sliding window is used to slice the data,and TSTKS algorithm is used in each sliding window to carry out online detection of multiple mutation points.Secondly,according to the trend of data fluctuation in different stages,an adaptive online model is proposed,which can dynamically adjust the buffer length by using the situation of mutation points detected in consecutive multiple buffers.According to the change of the number of mutation points detected in the buffer,when the number of mutation points detected in the continuous multiple buffers decreases gradually,the length of the buffer is enlarged.When the number of mutation points detected in continuous multiple buffers increases,the buffer length can be reduced.In addition,the length of the sliding window will be adjusted with the length of the buffer.When the buffer length decreases,the length of the sliding window will decrease proportionally;when the buffer length increases,the length of the sliding window will also increase.Meanwhile,in order to further improve the detection effect of the algorithm on mutation points,the random overlapping model was introduced into the sliding window model to improve the accuracy of the algorithm.It can be seen from the test results that the time consumption of the adaptive online model is significantly lower than that of the fixed buffer model,and the random overlapping model based on the adaptive online model can further improve the accuracy of the mutation point detection algorithm.Finally,the adaptive online model and random overlapping model proposed in this paper were used to construct a multi-path association network of pathological signals and carry out association analysis of pathological data.Experimental results show that the associative network based on the adaptive online model and random overlapping model proposed in this paper has better effect than the fixed buffer model and fixed sliding window model,no buffer model and fixed sliding window model.