Risk prediction method for endotracheal tube in critically ill patients
Objective To explore the application mechanism of intelligent warning algorithm of tracheal intubation pipeline risk in patients with severe illness.Methods Surveillance video data of 12 patients with tracheal intubation in the emergency room and intensive care unit from June 8,2022 to September 29,2022 were selected from Peking Union Medical College Hospital.Skin detection technology and SuperPoint technology were used to obtain the set of key points on the patient's face,and the average moving distance of key points was calculated to determine the moving state of the patient's head.The edge detection technology and parallel line detection technology based on Hoff transform are used to obtain the pipeline boundary pixel set,and the Hausdorff Distance between the pipeline boundary pixel set is calculated to judge the pipeline moving state.The state of the patient's hand is judged by the change of the center point of the boundary frame of the patient's hand.The analysis results of all the surveillance videos of the patients were summarized to obtain the risk rating of the patient's tracheal intubation accident.Results Through the experiment,the accuracy of the mobile detection algorithm proposed in the paper reached more than 90%,and the patient risk rating was consistent with the doctor's annotation.Conclusion The intelligent early warning algorithm of tracheal intubation pipeline risk in severe patients can effectively evaluate the risk level of patients and improve the work efficiency of medical staff.
万方数据
维普
