Theoretical Study On Accurate Diagnosis And Treatment Decisions Based On EEG Monitoring Large Data Of Single Epilepsy Cases

Epilepsy is a clinical syndrome with abnormal discharge of neurons caused by a variety of reasons. In an epileptic fit, the major clinical manifestations included sensory and motor disturbances, consciousness, mental and behavior disorders, autonomic nervous dysfunction, etc. Epilepsy is a kind of nervous system diseases which seriously jeopardizes human health, especially youth groups. It also has become a critical public health and clinical problem that we must deal with. Currently, clinical treatment of treating epilepsy mainly includes medical treatment and surgical treatment, etc. However, on the aspect of medical treatment, from the clinical medicine we found that the same type epilepsy cases with the same kind and the same dose of drugs may have different curative effect, some are effective and safe, and the other are invalid, and have even serious adverse reactions. Namely, the therapeutic and side effects of AEDS efficacy have significant individual differences. In the surgical treatment, there are many cases still having epileptic seizure after surgical removal of epileptic, which seriously hindered the application of surgical treatment. Therefore, for surgical treatment, the key is to accurately locate the position of the primary lesion.Traditional research and evaluation of medical treatment or surgical treatment method are generally based on multiple randomized clinical trials (RCT). However, the research subjects of RCT is a group of patients, while clinicians face individual case. In other words, RCT represents the universality (or average effect) of a thing, but each epilepsy case has its own specificity. Consequently, many RCT results do not always be used directly in the treatment of epilepsy in clinical practice. This method which only compare the average difference between the two groups and determine the efficacy of the method, ignores individual differences in therapeutic effect completely. Thus. evidence-based medicine based on RCT is difficult to promote the use of personalized treatment. Therefore, to investigate suitable personalized treatments for epilepsy, get rid of the drawback of RCT which only pay attention to average effect, and develop methods of accurate personalized treatment for epilepsy have been research focus of clinical diagnosis and treatment of epilepsy.The arrival of biomedical big data era prompted that clinical research must reconsider the theory and method system to adapt to the research demands of precise medicine, translational medicine. Precision Medical concept provides an unprecedented opportunity for the development of scientific, effective and accurate personalized diagnosis and treatment. However, the present precision medical research design ideas are based on precise classification of large numbers of cases the. Namely, large numbers of patients were divided into numerous relatively homogeneous subgroups, then suitable personalized treatments for epilepsy were investigated. Therefore, it is a "relative personalized" concept, rather than "absolute personalized" method. In fact, the absolute personalization should be fully implemented "treatment strategy for each person" rather than "treatment strategy for each group". That is, in absolute personalized clinical study design and statistical analysis, you must get rid of traditional Epidemiology and Biostatistics methods based on frequency school, and abandon the thinking patterns which inferences the curative effect information of population by a group of sample cases efficacy information. Instead, for a single feature of each particular case, personalized clinical research was applied to obtain precise personalized diagnostic method.In therapeutic strategy clinical studies of a single case, there are still some theoretical issues of research design and statistical analysis. In the study design, studies are still divided into experimental studies and observational studies two categories. Currently, method suitable for experimental research design is Single case of randomized controlled trials, or N-of-1 trials for short. However. for precise therapeutic strategy clinical observational studies of a single case, there are still lacking of effective design and statistical analysis method. But the case-crossover design for analytic epidemiology observational studies provides the basic idea for a single case observational study.Therefore, this study was based on the design ideas of case crossover design and the EEG monitoring big data information. Through EEG monitoring data mining and implementation of a single, specific cases of epilepsy, accurate treatment decisions design and analysis theory single of a single epilepsy cases will be found; and for the key problems that must be resolved to of epilepsy clinical personalized surgical treatment-"focus the primary lesion location accurately", suitable clinical personalized precise treatment methods of a single case should be studied.In the study design, this study based on a case-crossover design idea. During the process of long-range, multi-lead EEG monitoring, sets of time period EEG information recording in seizures and normal times were collected alternately, thereby EEG information collected constituting a contrast information set of EEG information recording in seizures and normal times of a single case. In the statistical analysis, focusing on the problem of localizing the intracranial abnormal electrical activity lesions, the model of localizing the intracranial epilepsy lesions was constructed based on independent component analysis (ICA) theoretical approach and EEG monitoring big data. Independent components of EEG information recording in seizures and normal times of the single case were extracted, by comparing independent components extracted from EEG information recording in seizures and normal times. Independent components of EEG information related to epileptic seizures were found, and then epilepsy lesions was located according to coefficient of the independent components in demixing matrix (i.e. the combination coefficient that electrode monitoring EEG signal on the head compose independent components in demixing matrix). For the key problem of surgical treatment for epilepsy-precise positioning for epilepsy primary focus, based on the above model of localizing the intracranial epilepsy lesions and the Granger Causality Test model, model of identifying the connectivity between epilepsy lesions and positioning for epilepsy primary lesions was proposed further. The model first built an auto regressive distributed lag model with EEG monitoring information of electrodes on the different positions, and then determined the regulatory relationships between different EEG signals in seizures and normal times through the Granger causality test method. The possible position of epilepsy primary lesions and regulation pathway of Epileptic discharge was by comparing analysis results.The main results:1. The model of localizing epileptic focus for single epilepsy case:The model of localizing epilepsy lesions was constructed based on independent component analysis (ICA) and EEG monitoring big data. And the blind source separation strategy and method of epilepsy EEG monitoring data was proposed. The aim of localizing epilepsy lesions realized by comparing independent components extracted from EEG information recording in seizures and normal times.6 EEG data recording in seizures and normal times analysis results indicates that the epilepsy lesions of case W was mainly in the position of electrodes F3、C3、P3. The results provided some guidance on making a plan of surgery treatment for epilepsy.2. Model of identifying the connectivity between epilepsy lesions and positioning for epilepsy primary lesions:The model first built an auto regressive distributed lag model with EEG information monitoring in seizures and normal times, based on the above model of localizing the intracranial epilepsy lesions and the Granger Causality Test model, model of identifying the connectivity between epilepsy lesions and positioning for epilepsy primary lesions was proposed.6 EEG data recording in seizures and normal times analysis results indicates that for case W, the connectivity between brain regions of 18 electrodes in seizures was stronger than it in normal times. That is to say the regulatory relationships between different EEG signals in normal times was less than in seizures. However, the number of regulatory relationships between different EEG signals in seizures exploded. Meta-analysis was applied to analyze 6 results of Granger Causality Test, the results indicates that the number of regulatory relationships between different EEG signals in seizures increased, and the direction of regulatory relationships was mainly from left brain regions to right brain regions. It can be seen, the left brain regions may be the possible position of primary epileptic focus.Main conclusion:1. The model of localizing epilepsy lesions was constructed based on independent component analysis (ICA) and EEG monitoring big data. And the blind source separation strategy and method of epilepsy EEG monitoring data was proposed. Analysis results indicates that the epilepsy lesions of case W was mainly in the position of electrodes F3、C3、P3. Compare to traditional studies focusing on the automatic identification of epileptic EEG signal, our model pays attention to the localizing the position of epileptic focus and can provided more guidance on clinical diagnosis and therapy.2. The model first built an auto regressive distributed lag model with EEG information monitoring in seizures and normal times, based on the above model of localizing the intracranial epilepsy lesions and the Granger Causality Test model, model of identifying the connectivity between epilepsy lesions and positioning for epilepsy primary lesions was proposed. Single case (case W) analysis results indicates that, the model of identifying the connectivity between epilepsy lesions and positioning for epilepsy primary lesions not only can identify the connectivity between brain regions of the electrodes, it can also localize the position of primary epileptic focus. Compare to traditional intracranial epileptic focus localizing methods, our model may reduce damage to people greatly, and realize epilepsy surgical accurate personalized treatment.