Research On Optimization Of Stimulus Onset Asynchrony In ERP-BCI Based On Subject Independent Model And Dynamic Stopping Strategy

As one of the classic event-related potential based brain-computer interface(ERP-BCI)paradigms,the P300 speller has been proven by many clinical experiments to have good thinking decoding performance,and it can operate stably in normal subjects and disabled patients.However,considering the clinical use requirements,the information conversion efficiency of the current P300 speller is still low,and there is still much room for improvement of system efficiency.Recent studies have shown that the Dynamical Stopping Strategy(DSS)and the Subject Independent Model(SIM)method can effectively improve the coding efficiency and save the calibration cost.It is an effective means of improving the efficiency of the P300 speller system.However,for the Stimulus Onset Asynchrony(SOA),which is more important for coding efficiency,the current research is less involved.The exploration of optimizing SOA to improve system efficiency based on how to the use of DSS and SIM is not reported.In traditional research,the SOA of the discriminant model and the SOA set in the system using the discriminant model should be consistent.Under this assumption,it is difficult to integrate DSS and SIM with SOA optimization.In view of this,this paper first discusses the application of SID and DSS under different SOA conditions.In the study,the P300 speller offline data of 55 subjects was used to establish a Subject Independent Dynamical Stopping Model(SIDSM),and the model applicability is verified under 5 different SOA conditions on 14 new subjects online.The experimental results show that the SIDSM model can maintain the correct rate under different SOA conditions and achieve dynamic stop.The average recognition accuracy rate under non-modeled SOA conditions reached 93.06%,the average stop round was 5.97,and there was no significant difference between the recognition rate and the stop round under the modeled SOA condition.The experimental results show that the SIDSM established in this paper has good SOA generalization performance and can be applied to different SOA conditions without updating.Based on this,this paper further constructs a P300 speller system that adaptively adjusts SOA.The system can adaptively change the SOA setting according to the user's current output efficiency.The online experiment results of 7 subjects showed that the output efficiency of the subjects can be gradually improved with the use process.After 15 characters of operation,the average information transmission efficiency can be improved by about 25.73% compared with the initial situation.The results show that the dynamic adjustment of SOA based on SIDSM can effectively improve system efficiency.This paper provides a new idea for the development of efficient P300 speller system,and also establishes a certain technical foundation for the development of SOA dynamic adjustment strategy.