Research On Individualized Optimization Of Implantable Cardiac Defibrillator And Shockable Rhythm Detection Algorithm

Sudden cardiac death(SCD)is mainly caused by ventricular arrhythmias(such as ventricular fibrillation).Due to it's high mortality,cardiopulmonary resuscitation and defibrillation are needed when it suddenly occurs.For high-risk patients with SCD,the treatment method is subcutaneous implantable defibrillator(S-ICD)for long-term treatment and prevention.Meanwhile,for patients with SCD outside the hospital,an automatic external defibrillator(AED)is usually used for emergency treatment.The SICD does not invade blood vessels and contacts the myocardium,which simplified the procedure and avoided the risk of complications.Furthermore,S-ICD can detect ventricular fibrillation in time and implement shock defibrillation,which widely used in clinical.However,S-ICD defibrillation energy is higher than T-ICD,causing myocardial damage.Moreover,the problems of over-sensing and in appropriate shock in S-ICD have caused psychological and physical pain in patients.What's more,the defibrillator's core shockable rhythm detection algorithm is crucial because it can quickly and accurately detect shockable rhythms and improve the survival rate of SCD.In order to solve the above problems,the main work of this paper is as follows:1.Six S-ICD electrode configurations were proposed and defibrillation electric fields under different configurations were simulated through the SCIRun platform besed on three swine models.Four defibrillation parameters were weighted and a comprehensive defibrillation efficacy index was obtained.2.A signal reconstruction algorithm based on BP neural network model was proposed,which use the standard twelve-lead ECG signal of the patient to reconstruct the corresponding subcutaneous three-lead vector of S-ICD.The feasibility of the algorithm is discussed and verified through the database.3.Research on shockable rhythm detection based on machine Learning,and the time and frequency domain features and complex nonlinear features of ECG signals are extracted.After the 10-fold cross-validation test,shcokable rhythm detection algorithms based on three machine learning classifiers were recommend.4.For the first time,a two-dimensional convolutional neural network model based on time-frequency diagram is proposed for automatic detection of shockable heart rhythm.The performance of the algorithm was verified on the international standard database by the ten-fold cross method.The accuracy,sensitivity,specificity of the algorithm is 98.82%,95.05% and 99.43% respectively,which was much higher than the 90% sensitivity and 95% proposed by the American Heart Association.In this paper,S-ICD is optimized before surgery,and the machine learning and deep learning methods are used to develop the defibrillator's shockable rhythm detection algorithm,which has achieved good results and effectively improved the defibrillator's Defibrillation efficiency.Due to the time and data limitations of this research,this paper only made a preliminary discussion and analysis,and the results and conclusions need to be further developed in animal experiments and statistical evaluation.