Driving risk prediction system, equipment and device based on physical examination report

Abstract

The invention discloses a driving risk prediction system, equipment and device based on a physical examination report, and belongs to the technical field of medicine and driving safety. The system comprises a data acquisition module used for acquiring a driver health examination report and traffic accident data; a data processing module; the accident occurrence probability calculation module adopts a generalized linear model based on Logistic regression to calculate a p value of the model; the p value represents the conditional probability of an accident calculated under a given characteristic condition; the accident occurrence frequency calculation module is used for calculating a regression coefficient of each physical examination index by adopting a generalized linear model based on Poisson regression; and the result analysis module is used for analyzing calculation results of the accident occurrence probability calculation module and the accident occurrence frequency calculation module, and is used for guiding actual driving behaviors. According to the invention, the health examination information of the driver can be associated with the driving risk, and the probability and number of the driving risk can be predicted, thereby helping to improve the driving safety.

Description

Technical Field

[0001] The present invention belongs to the field of medicine and driving safety technology, and in particular relates to a driving risk prediction system, equipment and device based on a health examination report. Background Art

[0002] Driving safety refers to a state in which the lives and property of drivers, passengers and other road users are effectively protected in road traffic, avoiding traffic accidents or reducing the losses caused by accidents. Ensuring driving safety is one of the core goals of road traffic management. At present, most studies focus on the impact of driving behavior on driving risks, including fatigue driving (mainly determined by driving mileage and driving time), dangerous driving (speeding, sudden acceleration, sudden deceleration, sharp turns), driving emotions (road rage and aggressive driving behavior), etc. to assess driving risks. For example: based on sensors (such as cameras or on-board sensors) to monitor the driver's status, based on the recognition of the driver's status and posture movements, the driver's possible unsafe factors are judged, and reminders or interventions are made.

[0003] Although good driving habits and driving behaviors can effectively reduce driving risks, the driver's health status has a direct impact on driving behavior and driving emotions, and may even be the main cause of accidents. However, there is little research on this in the field. Summary of the invention

[0004] The present invention aims to solve one of the technical problems in the above-mentioned related art at least to a certain extent.

[0005] To this end, the purpose of the present invention is to provide a driving risk prediction system, equipment and device based on health examination reports, which can associate the driver's health examination information with driving risks, predict the probability and amount of driving risks, and thus help improve driving safety.

[0006] In order to solve the above-mentioned technical problems, the present invention is achieved as follows:

[0007] An embodiment of the present invention provides a driving risk prediction system based on a health examination report, the system comprising:

[0008] A data collection module is used to collect the driver's health examination report within a specific time and traffic accident data within a specific time;

[0009] Data processing module, used to process health examination reports and traffic accident data;

[0010] The accident probability calculation module uses the generalized linear model based on Logistic regression and the data processed by the data processing module to calculate the p value of the generalized linear model based on Logistic regression; the p value represents the conditional probability of an accident under given characteristic conditions;

[0011] The accident occurrence calculation module uses a generalized linear model based on Poisson regression and the data processed by the data processing module to calculate the regression coefficients of various physical examination indicators;

[0012] The result analysis module is used to analyze the calculation results of the accident probability calculation module and the accident frequency calculation module, and to guide actual driving behavior.

[0013] In addition, the driving risk prediction system based on health examination reports according to the present invention may also have the following additional technical features:

[0014] In some of the embodiments, the result analysis module is also used to sort the p values ​​calculated by the accident probability calculation module from large to small, select several physical examination indicators that have a great impact on the accident probability, and use the selected physical examination indicators and related data as a new health examination report for calculation by the accident probability calculation module, and at the same time obtain a new generalized linear model based on Logistic regression.

[0015] In some implementation modes, the data processing module processes the traffic accident data including:

[0016] The traffic accident data is processed into categorical data, where 1 indicates an accident and 0 indicates no accident;

[0017] The traffic accident data is processed into count data, and the number of accidents is obtained through statistical summary.

[0018] In some embodiments, the generalized linear model based on logistic regression is expressed as:

[0019]

[0020] in, is the link function and also the log odds ratio,

[0021] π is the conditional mean of the dependent variable,

[0022] is the odds ratio when the dependent variable is equal to 1.

[0023] In some of the embodiments, It follows the binomial distribution.

[0024] In some embodiments, the dependent variable of the generalized linear model based on Poisson regression obeys Poisson distribution, and the fitting formula is:

[0025]

[0026] Where λ is the mean of the dependent variable,

[0027] log e (λ) is the link function.

[0028] In some embodiments, log e The probability distribution of (λ) is Poisson distribution.

[0029] In some embodiments, the original health checkup report has 20 physical examination indicators, and the new health checkup report has 12 physical examination indicators;

[0030] The physical examination indicators of the new health examination report include: high blood pressure, blood sugar, smoking, heart murmur, heart rhythm, heart sounds, vision, ears, age, electrocardiogram, liver function combination and multiple tumor marker tests.

[0031] An embodiment of the present invention also provides a driving risk prediction device based on a health examination report, comprising a memory and a processor, wherein a program is stored on the memory, and when the processor runs the program on the memory, the content of the driving risk prediction system based on a health examination report as described in any one of the above items can be implemented.

[0032] The embodiment of the present invention further provides a driving risk prediction device based on a health examination report, comprising the driving risk prediction device based on a health examination report and a physical examination device as described above;

[0033] The physical examination device is used to detect the physical examination index of the driver and transmit it to the driving risk prediction device based on the health examination report.

[0034] Compared with the prior art, the present invention has at least the following beneficial effects:

[0035] In an embodiment of the present invention, a driving risk prediction system based on a health examination report is provided, which can predict driving risks based on the driver's health examination report information, and based on Logistic regression, explore the influence of each physical examination indicator as the independent variable of the model on driving risk (dependent variable). Based on the Poisson regression algorithm, the number of accidents that will occur in the vehicle in the future can be predicted.

[0036] The driving risk prediction device and apparatus based on a health check report of the present invention include the driving risk prediction system based on a health check report, and thus at least have all the features and advantages of the driving risk prediction system based on a health check report, which will not be described in detail here. Additional aspects and advantages of the present invention will be partially given in the following description, and part will become apparent from the following description, or will be understood through the practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 The present invention discloses a structure and flow chart of a driving risk prediction system based on a health examination report according to an embodiment of the present invention. DETAILED DESCRIPTION

[0038] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0039] The embodiments of the present invention are described in detail below through specific embodiments and application scenarios in conjunction with the accompanying drawings.

[0040] See also Figure 1 As shown, in some embodiments of the present invention, a driving risk prediction system based on a health examination report is provided, and the system includes:

[0041] 1. Data collection module: It is divided into the collection of electronic health examination reports and the collection of traffic accident data of vehicles within one year; the collected data can be used for model training after subsequent processing.

[0042] 2. Physical examination index data processing module: Based on the electronic health examination report, the physical examination indexes in the report are processed and binarized. The independent variables used as model input are all categorical data. The linguistic description of each physical examination index is shown in Table 1.

[0043] Different models are entered according to the type of accident data (dependent variable). If the accident data is categorical data, that is, 1 means there has been an accident in a certain period of time, and 0 means there has been no accident in a certain period of time, then a generalized linear model based on logistic regression is entered. If the accident is count data, that is, the number of accidents in a certain period of time in the past, which can be a year, then a generalized linear model based on Poisson regression is entered.

[0044] Table 1 Data processing of physical examination indicators

[0045]

[0046] 3. Generalized Linear Model Based on Logistic Regression

[0047] The generalized linear model based on logistic regression was used to evaluate the impact of physical examination indicators on the probability of accidents.

[0048] Assume that the dependent variable is Y and the independent variables are X1, X2, ..., X p , assuming that the dependent variable is normally distributed, then,

[0049]

[0050] Among them, μ Y refers to the conditional mean of the dependent variable, β0 refers to the expected value of the dependent variable when all independent variables are 0, and β j is the expected change in the dependent variable as the independent variable changes. Formula (1) means that a series of independent variable values, given appropriate weights, and then added together can predict the mean of the observed values ​​of the dependent variable. The distribution of the independent variable does not necessarily have to be normally distributed, and nonlinear functions can also be used for the independent variable, so the formula for the generalized linear model can be obtained as:

[0051]

[0052] Among them, g(μ Y ) is called the link function, which is a function of the conditional mean. The dependent variable Y is not necessarily normally distributed, but can obey one of the exponential distributions.

[0053] Logistic regression is applicable to the case where the dependent variable is binary (0 and 1). Assuming that the dependent variable follows a binomial distribution, the regression formula is:

[0054]

[0055] Where π = μ Y , is the conditional mean of the dependent variable, that is, the probability that the dependent variable is 1 given a series of values ​​of the independent variable, is the odds ratio when the dependent variable is equal to 1, is the log odds ratio. In this formula, the link function is The probability distribution is a binomial distribution, and the parameter values ​​can be derived through multiple iterations of maximum likelihood estimation.

[0056] 4. Generalized Linear Model Based on Poisson Regression

[0057] The present invention uses a generalized linear model based on Poisson regression to predict the number of accidents. Poisson regression is applicable to situations where the dependent variable is a specific number. Assuming that the dependent variable follows a Poisson distribution, the fitting formula of the linear model is:

[0058]

[0059] Among them, λ is the mean of the dependent variable, log e (λ) is the link function, and the probability distribution is Poisson distribution. The derivation of model parameters is based on maximum likelihood estimation.

[0060] 5. Model evaluation module

[0061] In this invention, we evaluate the applicability of the model by drawing the graph of the predicted value and residual of the dependent variable. Use R language to draw common diagnostic graphs to evaluate the applicability of the Logistic regression model. The specific R language code is as follows:

[0062] plot(predict(model, type="response"), residuals(model, type="deviance"))

[0063] Use comprehensive diagnostic plots to evaluate the applicability of the Poisson regression model. The specific R language code is as follows:

[0064] 1ibrary(car)

[0065] influencePlot(model)

[0066] In the specific implementation process, the health examination reports of 1,321 drivers in the past six months and the traffic accident data in the past year were collected. The driving risks of the vehicles were predicted respectively. The 20 physical examination indicator variables of 1,321 drivers and the accidents that occurred in the past year (1 for no accident, 0 for accident) were collected as independent variables to predict the value of the dependent variable. The dependent variable is a binary result, with a traffic accident as 0 and no traffic accident as 1. Based on the Logistic regression model, the p-value of the regression coefficient (the p-value represents the conditional probability of an accident under given characteristic conditions) is shown in Table 2.

[0067] Table 2 p-values ​​of physical examination indicators

[0068]

[0069] From the p-value of the regression coefficient, it can be seen that diabetes, breath test, abdominal ultrasound, five thyroid function tests, three renal function tests, complete blood lipid test, blood routine test, and urine routine test do not contribute significantly to the model. After removing these indicator variables and testing the goodness of fit of the new model, the results are shown in Table 3.

[0070] Table 3 p-values ​​of physical examination indicators of the new model

[0071]

[0072]

[0073] From the p-values ​​in the table above, we can conclude that each regression coefficient of the new model is very significant (p < 0.05). The two models were compared using the anova() function of the R language. The chi-square value p = 0.22 was not significant using the chi-square test method, indicating that the new model with 12 indicators as independent variables fits as well as the model with 20 indicators as independent variables. Adding 8 indicators such as diabetes as independent variables did not significantly improve the prediction accuracy of the model. Therefore, we removed the independent variables with no significant effect, used fewer risk factors to build the model, and optimized the model.

[0074] The new Logistic regression model was used for prediction, and the probability of accidents occurring when each physical examination indicator took different values ​​was calculated, and the results shown in Table 4 were obtained.

[0075] Table 4 Comparison of accident probability with different values ​​of physical examination indicators

[0076]

[0077] P1 in Table 4 is the probability of an accident when all the physical examination indicators are within the normal range, and P2 is the probability of an accident when the physical examination indicators are abnormal. It can be seen from the above table that the probability of an accident for drivers with vision and hearing problems is more than twice that of drivers with normal vision and hearing (0.66 / 0.32, 0.55 / 0.27). Similarly, the probability of an accident for drivers with heart murmurs is also more than twice that of drivers without heart murmurs. The research results of the present invention show that hypertension, heart function, and liver function have a certain influence on the probability of an accident. Having hypertension and abnormal liver function can easily affect the driver's mood, and mood such as anger and getting angry easily are important factors leading to traffic accidents. The research results of the present invention can also be seen that with the increase of age, the probability of an accident is reduced. It can be seen from the logarithmic odds ratio of age in the above table that the probability of an accident will be multiplied by 0.826 for every year of age increase, that is, the probability of an accident is reduced by 17.4% for every year of age increase.

[0078] In order to predict the number of vehicle accidents, the number of accidents is used as the dependent variable, and the Poisson regression model is used. The present invention collects the accident data of the same batch of vehicles in the previous year, and uses the Poisson regression model to predict the number of vehicle accidents in the previous year. The specific implementation process is as follows: In the Poisson regression model, the dependent variable is the logarithm form of the conditional mean log e The regression coefficients of various physical examination indicators were calculated, and the results are shown in Table 5.

[0079] Table 5 Poisson regression coefficients of physical examination indicators

[0080]

[0081] From the results in the above table, we can see that if hypertension is abnormal, the number of accidents will be multiplied by 2.014, and if the age increases by one year, the number of accidents will be multiplied by 0.783. Similarly, keeping other indicators unchanged, if the vision is abnormal, the number of accidents will be multiplied by 1.945.

[0082] Based on the evaluation results obtained by the above model, the present invention predicts the accidents that occurred in the vehicle last year. Combined with the actual data of accidents collected last year, the prediction results of the model are compared, and the accuracy rate is about 90%. The results show that it is feasible to predict driving risks based on the Logistic model and the Poisson model, as well as the driver's physical examination report, and it has a high accuracy rate. It is also concluded that although driving safety is also related to the vehicle being driven, the road being driven, the weather, etc., the driver's own physical health and psychological state account for a large proportion of the issues affecting driving safety.

[0083] Parts of the present invention that are not described in detail are well known to those skilled in the art.

[0084] The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific implementation modes, which are merely illustrative rather than restrictive. Under the guidance of the present invention, ordinary technicians in this field can also make many forms without departing from the scope of protection of the present invention and the claims, all of which are within the protection of the present invention.

Claims

1. A driving risk prediction system based on health examination reports, characterized in that: The system comprises: A data collection module is used to collect the driver's health examination report within a specific time and traffic accident data within a specific time; Data processing module, used to process health examination reports and traffic accident data; The accident probability calculation module uses the generalized linear model based on Logistic regression and the data processed by the data processing module to calculate the p value of the generalized linear model based on Logistic regression; the p value represents the conditional probability of an accident under given characteristic conditions; The accident occurrence calculation module uses a generalized linear model based on Poisson regression and the data processed by the data processing module to calculate the regression coefficients of various physical examination indicators; The result analysis module is used to analyze the calculation results of the accident probability calculation module and the accident frequency calculation module, and to guide actual driving behavior.

2. The driving risk prediction system based on health examination report according to claim 1 is characterized in that: The result analysis module is also used to sort the p values ​​calculated by the accident probability calculation module from large to small, select several physical examination indicators that have a great impact on the accident probability, use the selected physical examination indicators and related data as a new health examination report to perform calculations in the accident probability calculation module, and simultaneously obtain a new generalized linear model based on Logistic regression.

3. The driving risk prediction system based on health examination report according to claim 1 is characterized in that: The data processing module processes the traffic accident data including: The traffic accident data is processed into categorical data, where 1 indicates an accident and 0 indicates no accident; The traffic accident data is processed into count data, and the number of accidents is obtained through statistical summary.

4. The driving risk prediction system based on health examination report according to claim 1 is characterized in that: The formula of the generalized linear model based on logistic regression is: in, is the link function and also the log odds ratio, π is the conditional mean of the dependent variable, is the odds ratio when the dependent variable is equal to 1.

5. The driving risk prediction system based on health examination report according to claim 4 is characterized in that: It follows the binomial distribution.

6. The driving risk prediction system based on health examination report according to claim 1 is characterized in that: The dependent variable of the generalized linear model based on Poisson regression obeys Poisson distribution, and its fitting formula is: Where λ is the mean of the dependent variable, log e (λ) is the link function.

7. The driving risk prediction system based on health examination report according to claim 6 is characterized in that: log e The probability distribution of (λ) is Poisson distribution.

8. The driving risk prediction system based on health examination report according to claim 2 is characterized in that: The original health checkup report has 20 physical examination indicators, and the new health checkup report has 12 physical examination indicators; The physical examination indicators of the new health examination report include: high blood pressure, blood sugar, smoking, heart murmur, heart rhythm, heart sounds, vision, ears, age, electrocardiogram, liver function combination and multiple tumor marker tests.

9. A driving risk prediction device based on a health examination report, comprising a memory and a processor, wherein a program is stored in the memory, characterized in that: When the processor runs the program on the memory, it can implement the content of the driving risk prediction system based on the health examination report as described in any one of claims 1-8.

10. A driving risk prediction device based on a health examination report, characterized in that: It includes the driving risk prediction device based on the health examination report and the physical examination device as described in claim 9; The physical examination device is used to detect the physical examination index of the driver and transmit it to the driving risk prediction device based on the health examination report.

Images