Treatment Strategies and Prognostic Factors in Patients with Stage T3 and T4 Laryngeal Carcinoma / 肿瘤防治研究

Objective To investigate the selection of treatment strategies and prognostic factors for patients with stage T3 and T4 laryngeal carcinoma. Methods A total of 132 patients with stage T3 and T4 laryngeal cancer admitted to our hospital from March 2010 to March 2019 were retrospectively selected as research objects. According to the different treatment strategies, the patients were divided into simple surgery group (group A, 57 cases), simple chemoradiotherapy group (group B, 32 cases), and surgery combined with chemoradiotherapy group (group C, 43 cases). The general data and clinicopathological features of the three groups were compared, and a survival curve was drawn by the Kaplan–Meier method. The 3-year survival rates of the three groups were compared. Then, the same 132 patients were divided into survival and death groups. The clinical data of the two groups were compared, and the prognostic factors were analyzed by multivariate logistic regression. A back propagation (BP) neural network model was constructed, and its differentiation and accuracy were evaluated. Results The proportions and 3 year survival rates of patients with poor differentiation, lymphatic vascular invasion, and involvement of lymph nodes outside the capsule in group C were significantly higher than those in groups A and B (P < 0.05). The 3 year survival rate of 132 patients was 68.94%(41/132). Poor differentiation, N2-N3 stage, lymphatic vascular invasion, and involvement of lymph nodes outside the capsule were risk factors for death (P < 0.05), whereas surgery combined with radiotherapy and chemotherapy were protective factors (P < 0.05). The BP neural network model exhibited good discrimination and high accuracy. Conclusion Surgery combined with radiotherapy and chemotherapy can significantly improve survival rate in patients with poor differentiation, lymphatic vascular invasion, and involvement of lymph nodes outside the capsule. Close attention should be paid to patients with stage N2-N3 in the formulation of reasonable treatment strategies.

Postmortem Interval Estimation Using Protein Chip Technology Combined with Multivariate Analysis Methods / 法医学杂志

OBJECTIVES@#To estimate postmortem interval (PMI) by analyzing the protein changes in skeletal muscle tissues with the protein chip technology combined with multivariate analysis methods.@*METHODS@#Rats were sacrificed for cervical dislocation and placed at 16 ℃. Water-soluble proteins in skeletal muscles were extracted at 10 time points (0 d, 1 d, 2 d, 3 d, 4 d, 5 d, 6 d, 7 d, 8 d and 9 d) after death. Protein expression profile data with relative molecular mass of 14 000-230 000 were obtained. Principal component analysis (PCA) and orthogonal partial least squares (OPLS) were used for data analysis. Fisher discriminant model and back propagation (BP) neural network model were constructed to classify and preliminarily estimate the PMI. In addition, the protein expression profiles data of human skeletal muscles at different time points after death were collected, and the relationship between them and PMI was analyzed by heat map and cluster analysis.@*RESULTS@#The protein peak of rat skeletal muscle changed with PMI. The result of PCA combined with OPLS discriminant analysis showed statistical significance in groups with different time points (P<0.05) except 6 d, 7 d and 8 d after death. By Fisher discriminant analysis, the accuracy of internal cross-validation was 71.4% and the accuracy of external validation was 66.7%. The BP neural network model classification and preliminary estimation results showed the accuracy of internal cross-validation was 98.2%, and the accuracy of external validation was 95.8%. There was a significant difference in protein expression between 4 d and 25 h after death by the cluster analysis of the human skeletal muscle samples.@*CONCLUSIONS@#The protein chip technology can quickly, accurately and repeatedly obtain water-soluble protein expression profiles in rats' and human skeletal muscles with the relative molecular mass of 14 000-230 000 at different time points postmortem. The establishment of multiple PMI estimation models based on multivariate analysis can provide a new idea and method for PMI estimation.

Construction and application of inadvertent perioperative hypothermia prediction model for patients under general anesthesia based on deep learning / 中华急诊医学杂志

Objective:To construct a prediction model of inadvertent perioperative hypothermia in patients under general anesthesia , and to apply to clinic to verify its performance.Methods:The data of 19 068 surgical patients in a Grade Ⅲ Class A hospital in Zhejiang Province from January 2016 to September 2020 were included. The model was constructed by using artificial intelligence technology based on deep learning, and the prediction effect of the model was tested by using the area under the subject operating characteristic curve and decision curve. Totally 2 157 surgical patients were included from October 2020 to March 2021 to test the prediction accuracy of the model.Results:The incidence of hypothermia was 13.89% (2 649/19 068) in the modeling group and 14.18% (306/2 157) in the validation group. The area under the subject operating characteristic curve of the prediction model was 0.724 (95% CI: 0.707-0.741), the sensitivity was 0.516, the specificity was 0.823, the cut-off value was 0.175, and the accuracy of practical application was 79.54%. Conclusions:This model can stably predict the incidence of perioperative inadvertent hypothermia in patients under general anesthesia, and provide reference for clinical prevention of inadvertent perioperative hypothermia.

Strike Velocity Prediction of Stick Blunt Instruments Based on Backpropagation Neural Network / 法医学杂志

OBJECTIVES@#To analyze and predict the striking velocity range of stick blunt instruments in different populations, and to provide basic data for the biomechanical analysis of blunt force injuries in forensic identification.@*METHODS@#Based on the Photron FASTCAM SA3 high-speed camera, Photron FASTCAM Viewer 4.0 and SPSS 26.0 software, the tester's maximum striking velocity of stick blunt instruments and related factors were calculated and analyzed, and inputed to the backpropagation (BP) neural network for training. The trained and verified BP neural network was used as the prediction model.@*RESULTS@#A total of 180 cases were tested and 470 pieces of data were measured. The maximum striking velocity range was 11.30-35.99 m/s. Among them, there were 122 female data, the maximum striking velocity range was 11.63-29.14 m/s; there were 348 male data, the maximum striking velocity range was 20.11-35.99 m/s. The maximum striking velocity of stick blunt instruments increased with the increase of weight and height, but there was no obvious increase trend in the male group; the maximum striking velocity decreased with age, but there was no obvious downward trend in the female group. The maximum striking velocity of stick blunt instruments has no significant correlation with the material and strike posture. The root mean square error (RMSE), the mean absolute error (MAE) and the coefficient of determination (R2) of the prediction results by using BP neural network were 2.16, 1.63 and 0.92, respectively.@*CONCLUSIONS@#The prediction model of BP neural network can meet the demand of predicting the maximum striking velocity of different populations.

Establishment and evaluation of a method for identifying the random error in the quantitative measurement procedure based on back propagation neural network / 中华检验医学杂志

Objective:To establish and evaluate a new real-time quality control method that can identify the random errors by using the backpropagation neural network (BPNN) algorithm and taking blood glucose test as an example.Methods:A total of 219 000 blood glucose results measured by Siemens advia 2 400 analytical system from January 2019 to July 2020 and derived from Laboratory Information System of Beijing Chaoyang Hospital Laboratory Department was regarded as the unbiased data of our study. Six deviations with different sizes were introduced to generate the corresponding biased data. With each biased data, BPNN and MovSD algorithms were used and tested, and then evaluated by traceability method and clinical method.Results:For BPNN algorithm, the block size was pre-set to 10 and the false-positive rate in all biases was within 0.1%. For MovSD, however, the optimal block size and exclusive limit were 150 and 10% separately and its false-positive rate in all biases was 0.38%, which was 0.28% higher than BPNN. Especially, for the least two error factors of 0.5 and 1, all the random errors were not detected by MovSD; for the error factor larger than 1, random errors could be detected by MovSD but the MNPed was higher than that of BPNN under all deviations. The difference was up to 91.67 times. 460 000 reference data were produced by traceability procedure. The uncertainty of BPNN algorithm evaluated by these reference data was only 0.078%.Conclusion:A real-time quality control method based on BPNN algorithm was successfully established to identify random errors in analytical phase, which was more efficient than MovSD method and provided a new idea and method for the identification of random errors in clinical practice.

Study on noninvasive blood glucose detection method using the near-infrared light based on particle swarm optimization and back propagation neural network / 生物医学工程学杂志

Most of the existing near-infrared noninvasive blood glucose detection models focus on the relationship between near-infrared absorbance and blood glucose concentration, but do not consider the impact of human physiological state on blood glucose concentration. In order to improve the performance of prediction model, particle swarm optimization (PSO) algorithm was used to train the structure paramters of back propagation (BP) neural network. Moreover, systolic blood pressure, pulse rate, body temperature and 1 550 nm absorbance were introduced as input variables of blood glucose concentration prediction model, and BP neural network was used as prediction model. In order to solve the problem that traditional BP neural network is easy to fall into local optimization, a hybrid model based on PSO-BP was introduced in this paper. The results showed that the prediction effect of PSO-BP model was better than that of traditional BP neural network. The prediction root mean square error and correlation coefficient of ten-fold cross-validation were 0.95 mmol/L and 0.74, respectively. The Clarke error grid analysis results showed that the proportion of model prediction results falling into region A was 84.39%, and the proportion falling into region B was 15.61%, which met the clinical requirements. The model can quickly measure the blood glucose concentration of the subject, and has relatively high accuracy.

Multi-index optimization of extraction process of Fengyin Decoction based on BAS-GA-BP neural network combined with entropy weight method / 中国中药杂志

To optimize the ethanol extraction technology parameters of Fengyin Decoction by orthogonal experiment combined with beetle antennae search(BAS)-genetic algorithm(GA)-back propagation neural network(BPNN). Based on single factor investigation, the extraction temperature, ethanol volume, extraction time, and ethanol concentration were used as orthogonal experiment factors, and entropy weight method was used to calculate the comprehensive scores of aloe-emodin, glycyrrhizic acid ammonium salt, rhein, emodin, chrysophanol, physcion, cinnamaldehyde, 6-gingerol, extraction ratio and fingerprint similarity. BAS-BPNN model was established, and then, GA was used to predict the optimal extraction process. The results showed that BAS-BPNN was optimized to obtain the optimal ethanol extraction process of Fengyin Decoction as follows: extraction temperature of 87 ℃, adding 9 times of 75 % ethanol, and extracting for 47 minutes, with a comprehensive score of 1.052 9. Meanwhile, the optimal process parameters obtained by orthogonal design were as follows: the extraction temperature of 80 ℃, adding 10 times of 75% ethanol, extracting for 30 minutes, with a comprehensive score of 1.003 7. The comprehensive score of the process obtained from the BAS-BPNN model was slightly better than that from the orthogonal test, indicating that the optimized process from BAS-BPNN model was more ideal, so it was finally determined as the best extraction process for Fengyin Decoction. The process of Fengyin Decoction obtained from BAS-GA-BPNN has high extraction efficiency and good stability, which provides reference for the subsequent development and quality control.

Analysis of factors influencing hospitalization cost based on back propagation neural network model / 中华医院管理杂志

Objective@#To analyze the influencing factors of hospitalization cost by back propagation(BP)neural network.@*Methods@#Inpatients with a total of 50 611 were collected from the hospital information system in a tertiary hospital. BP Neural network modeling was developed by SPSS 22.0 software to study the factors which influence the hospitalization cost.@*Results@#The hospitalization expenses prediction accuracies of training sample and test sample were 81.1% and 79.0% respectively. The top three factors were pharmaceutical cost(100.0%), general medical operation cost(88.8%)and disposable medical materials cost(60.7%).@*Conclusions@#The results show that BP model is suitable for dependent variable with the hospitalization expenses. The pharmaceutical cost, disposable medical materials cost and general medical operation cost were greater influencing factors than others. We should take actions by improving the treatment level and measures, so as to reduce the economic burden of patients and society

Multi class disorder detection of magnetic resonance brain images using composite features and neural network

Brain disorder recognition has becoming a promising area of study. In reality, some disorders share similar features and signs, making the task of diagnosis and treatment challenging. This paper presents a rigorous and robust computer aided diagnosis system for the detection of multiple brain abnormalities which can assist physicians in the diagnosis and treatment of brain diseases. In this system, we used energy of wavelet sub bands, textural features of gray level co-occurrence matrix and intensity feature of MR brain images. These features are ranked using Wilcoxon test. The composite features are classifi ed using back propagation neural network. Bayesian regulation is adopted to fi nd the optimal weights of neural network. The experimentation is carried out on datasets DS-90 and DS-310 of Harvard Medical School. To enhance the generalization capability of the network, fi vefold stratifi ed cross validation technique is used. The proposed system yields multi class disease classifi cation accuracy of 100% in diff erentiating 90 MR brain images into 18 classes and 97.81% in diff erentiating 310 MR brain images into 6 classes. The experimental results reveal that the composite features along with BPNN classifi er create a competent and reliable system for the identifi cation of multiple brain disorders which can be used in clinical applications. The Wilcoxon test outcome demonstrates that standard deviation feature along with energies of approximate and vertical sub bands of level 7 contribute the most in achieving enhanced multi class classifi cation performance results.

The application of auxiliary diagnosis expert system of biochemical analyzer / 中国医学装备

The combination of biochemistry analyzer and medical expert system was proposed in this report. Biochemistry analyzer is one of the most important analytical instrument used in clinical detection. It could take immunological examination and biochemical analysis for blood, urine, pleural effusion and cerebrospinal fluid and other body fluids. Medical expert system is an intelligent program system with knowledge and experience of a large number of medical specialists. It could use the knowledge and method of medical experts to solve and deal with problems in the field. This system mainly includes human interface, inference engine, interpreter, knowledge acquisition procedures, integrated database and knowledge base. Some parts of system design, such as the expression and design, and interpretation mechanism of the knowledge base, have been interpreted in details. It adopts production as an expression of knowledge. Generally, knowledge was expressed as if the conditions, then the conclusion form. Interpretation mechanism use the error counter propagation of neural network to train the algorithm (BP algorithm for short).The combination could automatically conduct comprehensive analysis of various data generated by the instruments, and then obtain the science theoretical foundation and the most reasonable specialist conclusions. This report provides an overview of the system design of medical expert system.

Application and MATLAB realization of drugs' classification based on the combination of NIRS detection and BP ANN algorithm / 国际生物医学工程杂志

Objective To realize rapid and non-destructive drug classification and improve the accuracy of drug classification.Methods A model for drug classification based on the combination of principal components analysis and artificial neural network (PCA-ANN) method was introduced.The software for drugs classification was then developed with the utility of MATLAB language.The near infra-red spectrum (NIRS) detection technique was executed on five kinds of drugs (a total of 120 batch samples) and the detection data was collected within the range of 1 350-1 800 nm of excitation wavelength and 0.5 nm of wavelength interval.Results The network training mean square error (MSE) was 5.91e-03,and the prediction error (β) was 2.469％ when the number of the interfering drugs number was less than 5.Conclusions The classification of drugs by NIRS combined with PCA-ANN is feasible and the classification accuracy can be increased.

Non-destructive prediction of tablet hardness by near infrared diffuse reflection spectroscopy / 中国药学杂志

OBJECTIVE: To establish a method for predicting tablet hardness by near infrared diffuse reflection spectroscopy. METHODS: Tablet hardness value was obtained by hardness meter. Calibration model between NIR spectra and the hardness was establish by partial least squares regression (PLSR) method and BP-ANN method. RESULTS: Correlation coefficients (r), root mean squares error of cross-validation (RMSECV), and root mean square error of prediction (RMSEP) obtained by PLSR model were 0.977 8, 0.742 and 0.815 kg respectively. And the correlation coefficients of training set, monitor set and testing set by BP-ANN were 0.987 3, 0.985 6, and 0.986 8, with RSE% of 6.83%, 8.77%, and 6.69%, respectively. CONCLUSION: The prediction accuracy of BP-ANN nonlinear model is superior to the PLSR model.

Optimization of Preparation Technology of Stachydrine in Fructus Leonuri Based on Back-propagation Neural Network and Genetic Algorithm / 广州中医药大学学报

Objective To optimize the preparative procedure for stachydrine in Fructus Leonuri. Methods The preparation was screened by orthogonal experiment, and a mathematical model of relationship of extraction time, methanol concentration, and solid-liquid ratio with the content of stachydrine hydrochloride was established by using back-propagation (BP) neural network. And the process parameters were optimized with genetic algorithm (GA) . Results The optimum process parameters were as follows: extraction with 69% of methanol concentration and with solid-liquid ratio being 11 times for 62 min. The content of stachydrine obtained by BP neural network modeling and GA was higher than that achieved by orthogonal experiment. Conclusion The optimum preparative procedure could be achieved by combining BP modeling with GA. The model developed in this study was proved to be predictable and feasible for the optimization of process parameters of multi-dimension nonlinear system.

Predictive effect of back propagation neural network model on hematoma enlargement in patients with cerebral hemorrhage / 中国脑血管病杂志

Objective To study predicting results of the back propagation (BP)neural network model for hematoma enlargement (HE)in patients with intracerebral hemorrhage. Methods The clinical data of 128 patients with cerebral hemorrhage admitted to the 309th hospital of People′s Liberation Army from January 2011 to December 2014 were analyzed retrospectively. The Matlab 7. 14 software was used to achieve BP neural network model for predicting hematoma enlargement within 24 hours in patients with intracerebral hemorrhage (HE ≥6. 0 ml and HE ≥12. 5 ml). The mean square error (MSE)of the model and the accuracy of the overall prediction were calculated. The receiver operation characteristic (ROC) curve was drawn for predicting HE. Results When the BP neural network predicted HE ≥6. 0 ml and HE ≥12. 5 ml,the mean square deviations of the training set,validation set,and test set were 0. 061, 0. 143,0. 052 and 0. 023,0. 057,and 0. 065,respectively. The best fitting performance verification of hematoma enlargement was as follows:≥ 6. 0 ml for network training 11 times and the error value 0. 224;≥12. 5 ml for network training 20 times,and the error value 0. 057. The overall accuracies of predicting HE ≥6. 0 ml and HE ≥12. 5 ml were 92. 2% (118/ 128)and 96. 9% (124/ 128)respectively. Conclusion The BP neural network model have no special limitation for data. It can accurately fit the hematoma expansion model of cerebral hemorrhage.

Effect of meteorological factors on incidence of tuberculosis: A 15-year retrospective study based on Chinese medicine theory of five circuits and six qi / 中国结合医学杂志

<p><b>OBJECTIVE</b>To analyze the correlations between the incidence of tuberculosis and meteorological factors over the same period and previous periods including 1, 2 and 3 years ago, defined according to the Chinese medicine theory of five circuits (Wu Yun) and six qi, to establish medical-meteorological forecast models for the Beijing area of China.</p><p><b>METHODS</b>Data regarding the incidence of tuberculosis between 1990 and 2004 were obtained from the Beijing Center for Disease Control and Prevention, and the data regarding the meteorological factors (including daily average temperatures, wind speeds, precipitations, relative humidities, vapor pressures and low cloud covers) between 1987 and 2004 were collected from the Beijing Meteorological Observatory and analyzed. Descriptive statistics and a back-propagation artificial neural network were adopted to analyze the data.</p><p><b>RESULTS</b>There were significant correlations between the incidence of tuberculosis and the meteorological factors in the corresponding year and previous years. Among these correlations, wind speed was the factor with the strongest influence on tuberculosis (the standardized significance was 100%). Additionally, all prediction models would successfully established, suggesting the use of a collection of meteorological factors spanning from three years ago to the present is superior to the use of single data.</p><p><b>CONCLUSIONS</b>The incidence of tuberculosis in Beijing area is correlated to meteorological factors in the current year and previous years, which verifies the practicality of the theory of five circuits and six qi.</p>

Dispersive Liquid-Liquid Microextraction Based on Solidification of Floating Organic Drop with High Performance Liquid Chromatography for Determination of Decabrominated Diphenyl Ether in Surficial Sediments / 分析化学

A method for the determination of decabrominated diphenyl ether(decaBDE) in sediment samples at trace level using dispersive liquid-liquid microextraction based on the solidification of floating organic drop (DLLME-SFO) and high performance liquid chromatography-ultraviolet detector (HPLC-UV) has developed.Based on the data of interactive orthogonal array design, the optimization experimental conditions were obtained with BP artificial neural network model: 1.00 mL methanol as dispersive solvent, 35.0 μL dodecanol as extractive solvent, 10.00% NaCl, pH 5, and extraction in 10 min.The extraction recovery (ER) was 62.22% at the extraction conditions.The proposed method exhibited a wide linear range(3.5-1400 ng/g) with R~2 =0.9921.The limit of detection (LOD) and the limit of quantification (LOQ) of this method were 2.3 pg/g(S/N =2) and 5.6 pg/g(S/N = 5), respectively.The recoveries of real samples at different spiking levels of decaBDE were 104.2%, 98.4% and 97.7%, respectively.Extraction, concentration and separation procedures for decaBDE from the sediment sample were carried out by one step, and hence, the process of DLLME-SFO for decaBDE was shortened.

Nonlinear inverse modeling of sensor based on back-propagation fuzzy logical system / 药物分析学报

Objective To correct the nonlinear error of sensor output, a new approach to sensor inverse modeling based on Back-Propagation Fuzzy Logical System (BP FS) is presented. Methods The BP FS is a computationally efficient nonlinear universal approximator, which is capable of implementing complex nonlinear mapping from its input pattern space to the output with fast convergence speed. Results The neuro-fuzzy hybrid system, i.e. BP FS, is then applied to construct nonlinear inverse model of pressure sensor. The experimental results show that the proposed inverse modeling method automatically compensates the associated nonlinear error in pressure estimation, and thus the performance of pressure sensor is significantly improved. Conclusion The proposed method can be widely used in nonlinearity correction of various kinds of sensors to compensate the effects of nonlinearity and temperature on sensor output.

Nonlinear inverse modeling of sensor based on back-propagation fuzzy logical system / 西安交通大学学报·英文版

Objective: To correct the nonlinear error of sensor output, a new approach to sensor inverse modeling based on Back-Propagation Fuzzy Logical System (BP FS) is presented. Methods: The BP FS is a computationally efficient nonlinear universal approximator, which is capable of implementing complex nonlinear mapping from its input pattern space to the output with fast convergence speed. Results: The neuro-fuzzy hybrid system, i. e. BP FS, is then applied to construct nonlinear inverse model of pressure sensor. The experimental results show that the proposed inverse modeling method automatically compensates the associated nonlinear error in pressure estimation, and thus the performance of pressure sensor is significantly improved. Conclusion: The proposed method can be widely used in nonlinearity correction of various kinds of sensors to compensate the effects of nonlinearity and temperature on sensor output.

Development of Nursing Diagnosis Program Using Back-propagating Neural Network Model for Self-learning / 대한의료정보학회지

With the rapid growth of research and recognition about usefulness and importnace of the Nursing Diagnosis, the demand for application of Nursing Diagnosis has never been stronger. But in clinical field, not many nurses has used Nursing Diagnosis. Especially, nursing student have a difficulty to use Nursing Diagnosis because it demands for high level of capability of analyzing collected data and combining with relevant references. Therefore. this research has developed Nursing Diagnosis Self-learning Program using Back-propagating Neutral Network Model which is based on 98 surgery patients' data for nursing student. The twenty-six nursing diagnoses based on NANDA Taxonomy with 189 cases' reports and aid of 8 nursing experts wee determined to develop the program. To verify the usefulness of Nursing Diagnosis Self-learning Program constructed with the fully trained neural nets, the Program was tested with 70 real patients' data. The simulated output of program was compared with the judgement of the researcher and of two experts of nursing. The misdiagnosis rate of this program was eleven percent. This Program needs input of Signs and Symptoms, risk factors and 'related to' factors and also input the nursing diagnoses which a student selects. And than prints out two types of diagnoses. One is from the system and the other is what the student inputed. And the student makes the final diagnosis by refering the two types of diagnoses. Finally, the program prints out the completed diagnosis which problem combines with etiology in the diagnosis producing module. The program helps students to improve her capacity related to use Nursing Diagnosis.

Development of Breast Cancer Predication Model Using Neural Networks / 대한의료정보학회지

This paper describes the developing a neural network breast cancer prediction model using neural network. Neural Networks are nonparametric, pattern recognition techniques that can be used to complex biological relationships. The applicability of multilayer perceptron (MLP) was accessed using the data of 1143 patients who had breast cancer or visited hospital for the treatment of other disease. The MLP prediction model consists of one-hidden layer and 4 to 10 hidden nodes and it is trained by back propagation algorithm. The overall prediction performance of the model was 76% as evaluated with test data sets. Principal Component Analysis(PCA) was done for featuring the input data. The performance of PCA neural networks which had 7 transformed input nodes was 72% as it was tested on the unknown data.