There is a close relationship between the occurrence of a variety of diseases and meteorological factors. However, the typical disease forecasting methods are based on history data and the requirement of initial data is strict. To solve these problems, we proposed a disease forecasting algorithm to adapt to real-time data. The proposed algorithm has two contributions: (1) It uses the single factor correlation analysis methods when selecting meteorological factors that affect disease (2) It introduces a new method to calculate disease prediction to build date _number _meteorological factor matrix and use JacUOD algorithm to evaluate the similarity of meteorological factors between the target dates and past ones. To find out the top-N dates are of the maximum similarity with the target one, therefore, we could forecast the number combining the similarity value and the N date's patient number. Obviously, the number of patient is obtained by calculating the similarity of different dates' meteorological factors. Experiments show that the algorithm generates a better accuracy than the traditional algorithms in disease prediction.

A disease forecasting algorithm based on single factor correlation analysis and the JacUOD algorithm

The invention provides a method and a system for detecting COVID-2019 pneumonia, which belongs to the technical field of biomedicine and data processing. Expiratory NO concentration detection is carried out on a non-new coronal pneumonia patient, statistics of basic human body characteristic information is carried out, a mathematical model between the expiratory NO concentration and the basic characteristics of the human body is established, expiratory NO data of the suspected new coronal pneumonia patient are analyzed by using the mathematical model, whether a subject is sick or not is thus judged, the problem of discomfort of a human body caused by traditional nucleic acid detection sampling is effectively solved, and the method has the advantages of low cost, rapidness, no damage and the like, causes no pain to the subject, is good in compliance, is particularly suitable for being applied to large-scale preliminary screening work of suspected COVID-19 patients, and has good practical application value.

Method and system for detecting COVID-2019 pneumonia

BP algorithm has the advantages of simple structure, easy to realize, and it has been widely applied in fault diagnosis, pattern recognition, but the shortcoming of BP algorithm has affected its further development. Based on constructing the mathematical model of the BP algorithm, this paper analyzes the shortcomings of BP algorithm. Problems exist in the slow learning speed, the large possibility of failure to learn, poor generalization ability, the presence of multiple extreme points, the network structure is not easy to determine. Based on research results, heuristic information technology and algorithms, numerical optimization methods, this paper focuses on the study of BP learning algorithm and several improved methods, including improve the algorithm on convergence rate of the variable learning rate and algorithm adding momentum item, improve the search ability and generalization ability of genetic algorithm, eliminate resilient BP algorithm of Gradient magnitude effect, overcome the simulated annealing algorithm of the local minimum. The improvements of BP learning algorithm have a very important significance, especially when combined with other intelligent algorithms, whether in academic or in the application.

The Problems and Some Improved Algorithm of BP Learning Algorithm

Objective to explore predictive method of nonlinear time series based on using BP neural network. Methods Based on dynamic learning rate BP artificial neural network with Hyperbolic Tangent function as activation function has been used. Results Build two kinds of ANN forecast models of pneumonia incidence rate. They are better than traditional method on prediction precision. Conclusion BP artificial neural network can be used to forecast for disease incidence rate.

Pneumonia Incidence Rate Predictive Model of Nonlinear Time Series Based on Dynamic Learning Rate BP Neural Network

The connotation of BP neural network algorithm and code was introduced. The construction of expert system for forecasting the incidence rate of pulmonary emphysema based on BP neural network was discussed. Methods The data of incidence rate of pulmonary emphysema and meteorological factors in plateau section from 2003 to 2009 were collected and analyzed by using Eviews for windows, version 3.1; the model of Back Propagation artificial neural network was built by Matlab, version7.0. The MER and R2 of pulmonary emphysema incidence rate forecasting model were 0.60401% and 0.95329 respectively. The effect of fitting and forecasting of the model were very well. BP neural network had a strong application value in the forecasting the incidence rate of pulmonary emphysema.

Application study on BP neural network in the incidence rate of pulmonary emphysema

The steps of data fusion for multi-sensor can be divided into data preprocessing, track correlation, and tack fusion, etc. At first, conversion of units for data has been done in this paper. Then the data has been classified by the radar number, radar type, and track height, with the methods of systematic cluster analysis, trajectory which measured by each radar has been extracted, where longitude and latitude, time, velocity are the variables. Using the time slice technology of clustering, and using the outsider push data alignment algorithm of time normalization based on the information multi-sensor data, temporal registration has been done and extracted trajectories for seven objectives; VAR model has been established based on the data which observed by radar, and judged the observation precision by virtue of goodness of fitting, AIC and SC criterion of results of model test, derived the observation precision of each radar. On this basis, trajectory of observation with larger error drew closer to the trajectory with smaller error in order to eliminate the relative deviation of the radar. After error correction for radar, a trajectory of the same goal observed by radar has been fused and relative data has also been integrated. VAR model has been established with the variables of the precision, latitude and time. Software of Matlab7.0 has been used to draw the track of the goal in the future ten minutes after the model passed the model test.

/pdf/data-fusion-and-trajectory-prediction-of-multi-sensor-4x4wyu59yq.pdf

Data Fusion and Trajectory Prediction of Multi-Sensor

The selection of the modeling data, the data's pretreatment of nephritis incidence rate forecasting and their effects on forecasting precision were discussed. Aiming at the actual difficulties in incidence rate forecasting, a new incidence rate forecasting method in which the solar term incidence rate is used as modeling data was put forward combining ARIMA model and BP network. The practice has proved that the method will improve the precision of incidence rate forecasting effectively and has excellent robusticity. The nephritis incidence rate characteristics in each time period should be duly considered in order to carry out a more focused health prevention and treatment and to effectively reduce the hazards of nephritis.

Ma Liang-liang

Papers

Pneumonia Incidence Rate Predictive Model of Nonlinear Time Series Based on Dynamic Learning Rate BP Neural Network

Application study on BP neural network in the incidence rate of pulmonary emphysema

Data Fusion and Trajectory Prediction of Multi-Sensor

Improved Nephritis Incidence Rate Forecasting Method Based on BP Network