Research On Medical Intelligent Auxiliary Infusion System Based On “Internet Plus”

As an important route of administration and treatment,intravenous infusion is still dominated by manual monitoring.However,this monitoring method has problems such as high nursing burden,low safety and low management efficiency,especially during the outbreak of the new crown epidemic,this problem has become particularly prominent.In response to this problem,this paper designs a medical intelligent auxiliary infusion system based on“Internet+”,which aims to reduce the nursing burden of guardians and improve the safety and management efficiency of infusion therapy.First of all,the research status of“Internet+”intelligent infusion and infusion monitoring methods is investigated and analyzed,the urgent problems to be solved are summarized,and the system design goals and overall plan are determined,on this basis,completed the analysis,design and selection of system machinery,software and hardware;secondly,in order to solve the problem that it is difficult for ordinary cameras to clearly capture the images of droplets falling at high speed,a method for the determination of droplet counting and infusion remaining amount based on geometric information is proposed.And after determining the detection target,combined with the actual infusion scene,an infusion detection network based on improved YOLOv5s is designed.Among them,the network input uses the Mosiac+Mixup hybrid data augmentation method to enrich the infusion dataset.The feature extraction side draws on the design idea of the Efficient Net V2 network,builds a lightweight backbone network based on the infusion scene,and also introduces the ACON-C activation function.The SE attention mechanism module is introduced at the feature fusion end to construct the FPN+PAN+SE feature fusion structure.The Cluster-NMS_W+D method is used in the prediction layer to improve the two major problems of easily missed detection due to occlusion of droplets and easy false detection due to similar characteristics of liquid level and dropper.Thirdly,combining the advantages of PID control and fuzzy control,a fuzzy PID drip rate control algorithm based on infusion scenario is proposed to improve the problem that conventional PID control cannot meet the precision requirements of complex drip rate control systems.Next,in order to realize remote monitoring and unified management of infusion therapy,the design and development of local,mobile and terminal human-computer interaction interfaces are completed according to the three development steps of functional requirement analysis,interface layout design and logic flow design.Finally,using the self-developed intelligent infusion prototype as the experimental platform,experiments on target detection accuracy,droplet counting,margin judgment accuracy,dripping speed stability,fuzzy PID dripping speed control algorithm control accuracy,and human-computer interaction effects were carried out.The experimental results show that the intelligent infusion system has the advantages of high infusion monitoring accuracy,good drip rate control effect,and friendly human-computer interaction experience,which can meet the application needs of most infusion scenarios.The safety of the infusion treatment is effectively guaranteed,the nursing burden of the guardian is reduced,and the management efficiency of the infusion treatment is improved.The Internet-based infusion monitoring system will pave the way for the establishment of a smart medical model in hospitals in the future.