Driver health reminding method and device, electronic equipment, storage medium and product

Abstract

The invention relates to a driver health reminding method and device, electronic equipment, a storage medium and a product. The method comprises the steps that current driving data of a vehicle, current physiological data of a driver, current behavior data of the driver and historical health data of the driver are collected; based on a preset feature selection strategy, obtaining a first state evaluation result of the driver according to the current driving data and the current physiological data, and based on a preset cloud prediction model, obtaining a second state evaluation result of the driver according to the current driving data, the current physiological data, the current behavior data and the historical health data; and generating a health state monitoring result of the driver according to the first state evaluation result and / or the second state evaluation result, and performing health reminding when the health state monitoring result meets a preset reminding condition, thereby solving the problem of relatively low real-time performance and accuracy of a vehicle-mounted health monitoring system in related technologies, and improving the user experience. And comprehensiveness and accuracy of evaluation of the health state of the driver are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent detection, in particular to a driver health reminding method and device, electronic equipment, storage medium and product. BACKGROUND

[0002] With the rapid development of intelligent transportation systems and the gradual popularization of automatic driving technology, the influence of the health status of drivers on driving safety is increasingly valued.

[0003] In the related art, when evaluating the health status of a driver, the physical data of the driver is generally collected by a millimeter wave radar and a camera, and a driver health report is obtained according to the collected physical data. However, evaluating the health status of a driver only by physical data is too single, cannot achieve comprehensive evaluation of the health status of a driver, and results in low evaluation accuracy, which needs to be solved urgently. SUMMARY

[0004] The present application provides a driver health reminding method and device, electronic equipment, storage medium and product to solve the problem of low evaluation accuracy caused by too single evaluation of the health status of a driver only by physical data in the related art, and improve the comprehensiveness and accuracy of evaluation of the health status of a driver.

[0005] The first aspect of the present application provides a driver health reminding method, comprising the following steps: collecting current driving data of a vehicle, current physiological data of a driver, current behavior data of the driver and historical health data of the driver; obtaining a first state evaluation result of the driver based on a preset feature selection strategy according to the current driving data and the current physiological data, and obtaining a second state evaluation result of the driver based on a preset cloud prediction model according to the current driving data, the current physiological data, the current behavior data and the historical health data; generating a health status monitoring result of the driver according to the first state evaluation result and / or the second state evaluation result, and performing health reminding when the health status monitoring result meets a preset reminding condition.

[0006] Further, in some embodiments, the first state evaluation result of the driver is obtained based on a preset feature selection strategy according to the current driving data and the current physiological data, comprising: determining at least one target feature based on a preset feature selection formula according to the current driving data and the current physiological data; determining a reminding threshold interval and / or a duration of each target feature, and obtaining the first state evaluation result of the driver according to the current driving data, the current physiological data, the reminding threshold interval and / or the duration of each target feature.

[0007] Further, in some embodiments, the obtaining the first state evaluation result of the driver according to the current driving data, the current physiological data, the reminding threshold interval and / or the duration of each target feature comprises: determining the current value of each target feature from the current driving data and the current physiological data; generating the reminding information corresponding to any target feature in the case that the current value of the target feature is in the corresponding reminding threshold interval, and / or the duration of the current value of the target feature in the corresponding reminding threshold interval reaches the corresponding duration; and obtaining the first state evaluation result according to the reminding information corresponding to the target feature.

[0008] Further, in some embodiments, before obtaining the second state evaluation result of the driver according to the current driving data, the current physiological data, the current behavior data and the historical health data based on the preset cloud prediction model, the method further comprises: obtaining the historical driving data of the vehicle, the historical physiological data of the driver and the behavior data of the driver; preprocessing the historical driving data, the historical physiological data, the behavior data and the historical health data respectively, and determining at least one key feature meeting a preset extraction condition from the preprocessed historical driving data, the preprocessed historical physiological data, the preprocessed behavior data and the preprocessed historical health data based on a preset feature extraction strategy; and training a preset neural network to obtain the preset cloud prediction model based on a preset cross-loss function and using the at least one key feature and the corresponding prediction classification result of the at least one feature.

[0009] Further, in some embodiments, after obtaining the second state evaluation result of the driver according to the current driving data, the current physiological data, the current behavior data and the historical health data based on the preset cloud prediction model, the method further comprises: controlling the vehicle to display the health evaluation result and the health suggestion in the second state evaluation result.

[0010] Further, in some embodiments, after collecting the current driving data, the current physiological data of the driver, the current behavior data of the driver and the historical health data of the driver, further comprising: respectively denoising the current driving data, the current physiological data, the current behavior data and the historical health data to obtain denoised driving data, denoised physiological data, denoised behavior data and denoised historical health data; respectively normalizing the denoised driving data, the denoised physiological data, the denoised behavior data and the denoised historical health data to obtain normalized driving data, normalized physiological data, normalized behavior data and normalized historical health data; respectively processing missing values of the normalized driving data, the normalized physiological data, the normalized behavior data and the normalized historical health data to obtain processed driving data, processed physiological data, processed behavior data and processed historical health data.

[0011] According to the driver health reminding method provided by the embodiment of the application, the collected vehicle driving data, the current physiological and behavior data of the driver and the historical health data are analyzed by the feature selection strategy to obtain a first state evaluation result, and a second state evaluation result is obtained by fusing the multi-source data by using the cloud prediction model, and the health state monitoring result is comprehensively generated, thereby solving the problem of low real-time performance and accuracy of the vehicle-mounted health monitoring system in the related art, and improving the comprehensiveness and accuracy of the evaluation of the health state of the driver.

[0012] The second aspect of the embodiment of the application provides a driver health reminding device, and the device comprises: a collection module configured to collect current driving data of a vehicle, current physiological data of a driver, current behavior data of the driver and historical health data of the driver; an evaluation module configured to obtain a first state evaluation result of the driver based on a preset feature selection strategy and according to the current driving data and the current physiological data, and obtain a second state evaluation result of the driver based on a preset cloud prediction model and according to the current driving data, the current physiological data, the current behavior data and the historical health data; and a warning module configured to generate a health state monitoring result of the driver according to the first state evaluation result and / or the second state evaluation result, and perform health reminding when the health state monitoring result meets a preset reminding condition.

[0013] Further, in some embodiments, the evaluation module is specifically configured to: determine at least one target feature based on a preset feature selection formula according to the current driving data and the current physiological data; determine a reminding threshold interval and / or a duration of each target feature, and obtain the first state evaluation result of the driver according to the current driving data, the current physiological data, the reminding threshold interval and / or the duration of each target feature.

[0014] Further, in some embodiments, the evaluation module is further configured to: determine a current value of each target feature from the current driving data and the current physiological data; generate corresponding reminding information of any target feature in a case that the current value of the target feature is in the corresponding reminding threshold interval, and / or the duration of the current value of the target feature in the corresponding reminding threshold interval reaches the corresponding duration, and obtain the first state evaluation result according to the reminding information of the target feature.

[0015] Further, in some embodiments, before obtaining the second state evaluation result of the driver based on the preset cloud prediction model according to the current driving data, the current physiological data, the current behavior data and the historical health data, the evaluation module is further configured to: obtain historical driving data of the vehicle, historical physiological data of the driver and behavior data of the driver; pre-process the historical driving data, the historical physiological data, the behavior data and the historical health data respectively, and determine at least one key feature meeting a preset extraction condition from the pre-processed historical driving data, the pre-processed historical physiological data, the pre-processed behavior data and the pre-processed historical health data based on a preset feature extraction strategy; train a preset neural network to obtain the preset cloud prediction model based on a preset cross-loss function by using the at least one key feature and a corresponding prediction classification result of the at least one feature.

[0016] Further, in some embodiments, after obtaining the second state evaluation result of the driver based on the preset cloud prediction model according to the current driving data, the current physiological data, the current behavior data and the historical health data, the evaluation module is further configured to: control the vehicle to display a health evaluation result and a health suggestion in the second state evaluation result.

[0017] Further, in some embodiments, after collecting the current driving data, the current physiological data of the driver, the current behavior data of the driver and the historical health data of the driver, the collection module is further configured to: perform denoising processing on the current driving data, the current physiological data, the current behavior data and the historical health data respectively to obtain denoised driving data, denoised physiological data, denoised behavior data and denoised historical health data; perform normalization processing on the denoised driving data, the denoised physiological data, the denoised behavior data and the denoised historical health data respectively to obtain normalized driving data, normalized physiological data, normalized behavior data and normalized historical health data; and perform missing value processing on the normalized driving data, the normalized physiological data, the normalized behavior data and the normalized historical health data respectively to obtain processed driving data, processed physiological data, processed behavior data and processed historical health data.

[0018] According to the driver health reminding device provided by the embodiment of the application, the collected vehicle driving data, the current physiological and behavior data of the driver and the historical health data are analyzed by the feature selection strategy to obtain a first state evaluation result, and a second state evaluation result is obtained by using a cloud prediction model to fuse the multi-source data, and a health state monitoring result is comprehensively generated, thereby solving the problem of low real-time performance and accuracy of the vehicle-mounted health monitoring system in the related art, and improving the comprehensiveness and accuracy of the evaluation of the health state of the driver.

[0019] The third aspect of the application provides an electronic device, comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the driver health reminding method of the above-mentioned embodiments.

[0020] The fourth aspect of the application provides a computer readable storage medium having a computer program stored thereon, wherein the program is executed by a processor to implement the driver health reminding method of the above-mentioned embodiments.

[0021] The fifth aspect of the application provides a computer program product comprising a computer program, wherein the computer program is executed to implement the driver health reminding method of the above-mentioned embodiments.

[0022] Therefore, the application has at least the following beneficial effects:

[0023] (1) The multi-modal data fusion of the application can integrate data from different sensors, including physiological signals, facial expressions, eye tracking, vehicle condition data, etc., to provide more comprehensive information and more accurately evaluate the overall health status of the driver.

[0024] (2) The present application can analyze physiological signals in real time at the vehicle end, identify potential health risks within a few milliseconds, and immediately trigger an alarm or take measures. At the same time, all data is uploaded to the cloud for further deep learning and multi-modal data fusion analysis, realizing the collaborative work of edge computing and cloud processing.

[0025] (3) The present application does not increase the computational load of data processing at the vehicle end without affecting the timeliness of the reminder and the accuracy of the evaluation. Moreover, the two-layer independent health evaluation mechanism can be considered as an industrial-level redundant design, which can effectively avoid the problem of untimely and inadequate reminders caused by the failure of single evaluation.

[0026] (4) The present application combines the driver's disease history database, which can customize and monitor and warn the health of each driver according to their specific circumstances, dynamically adjust the threshold and warning strategy of monitoring, and provide personalized health advice. BRIEF DESCRIPTION OF DRAWINGS

[0027] The above and / or additional aspects and advantages of the present application will become apparent and more readily appreciated from the following description, taken in conjunction with the accompanying drawings, in which:

[0028] Figure 1 A flowchart of a driver health reminder method according to an embodiment of the present application is provided.

[0029] Figure 2 A cloud algorithm processing flowchart according to one specific embodiment of the present application is provided.

[0030] Figure 3 A schematic diagram of a driver health reminder method flowchart framework according to one specific embodiment of the present application is provided.

[0031] Figure 4 An edge computing flowchart according to one specific embodiment of the present application is provided.

[0032] Figure 5 A flowchart of a driver health reminder method according to one specific embodiment of the present application is provided.

[0033] Figure 6 A block diagram of a driver health reminder device according to an embodiment of the present application is provided.

[0034] Figure 7 A structural schematic diagram of an electronic device according to an embodiment of the present application is provided. DETAILED DESCRIPTION

[0035] Embodiments of the present application are described below in detail with reference to the accompanying drawings, examples of which are shown in the drawings, wherein the same or similar notations represent the same or similar elements or elements having the same or similar functions throughout. The embodiments described below by reference to the drawings are exemplary and are intended to explain the present application, and cannot be understood as limiting the present application.

[0036] A driver health reminding method, device, electronic equipment, storage medium and product are described below with reference to the accompanying drawings. In view of the low real-time and accuracy of the vehicle-mounted health monitoring system in the related art mentioned above, the present application provides a driver health reminding method, which analyzes the collected vehicle driving data, current physiological and behavioral data of the driver and historical health data of the driver by a feature selection strategy to obtain a first state evaluation result, simultaneously uses a cloud prediction model to fuse multiple source data to obtain a second state evaluation result, and comprehensively generates a health state monitoring result, thereby solving the problem of low real-time and accuracy of the vehicle-mounted health monitoring system in the related art and improving the comprehensiveness and accuracy of the evaluation of the driver health state.

[0037] Specifically, Figure 1 A flowchart of the driver health reminding method provided according to an embodiment of the present application is shown in FIG. 1.

[0038] As shown in FIG. 1, the driver health reminding method includes the following steps: Figure 1

[0039] In step S101, the current driving data of the vehicle, the current physiological data of the driver, the current behavioral data of the driver and the historical health data of the driver are collected.

[0040] The current driving data of the vehicle refers to the data for reflecting the running state of the vehicle in the actual driving state, the current physiological data of the driver refers to the physiological signals of the driver of the vehicle in the actual driving state, such as heart rate, respiration, etc., the current behavioral data of the driver refers to the facial expression data of the driver in the actual driving state of the vehicle, such as blinking, head turning, etc., and the historical health data of the driver refers to the historical medical data and disease history data of the driver.

[0041] ​For example, the current driving data of the vehicle can be collected by sensors built in the vehicle, such as vehicle speed, vehicle acceleration, and steering angle, the current physiological data of the driver can be collected by sensors installed on the seat or steering wheel of the vehicle, such as heart rate, breathing rate, and blood pressure, the current behavior data of the driver can be captured by a camera in the cockpit, such as facial expression and eye movement, such as blink frequency, eyelid closure time, and facial expression, and the historical health data of the driver can be input or imported by the user, which records the medical history of the driver. This part of data will be stored in the local secure database.

[0042] In step S102, based on the preset feature selection strategy, the first state evaluation result of the driver is obtained according to the current driving data and the current physiological data, and based on the preset cloud prediction model, the second state evaluation result of the driver is obtained according to the current driving data, the current physiological data, the current behavior data, and the historical health data.

[0043] Among them, the feature selection strategy refers to the selection strategy of selecting physiological signal features that contribute more to the prediction ability of the model in the model prediction process, the cloud prediction model refers to a data-driven algorithm model deployed in the cloud server, which realizes the prediction of the state or trend of the target object through the learning and analysis of multidimensional data, and the evaluation result refers to the quantitative or qualitative conclusion calculated by the cloud prediction model, which is used to reflect the specific attribute of the evaluation object or the degree of meeting the evaluation target.

[0044] Further, in some embodiments, based on the preset feature selection strategy, the first state evaluation result of the driver is obtained according to the current driving data and the current physiological data, including: determining at least one target feature based on the preset feature selection formula according to the current driving data and the current physiological data; determining the warning threshold interval and / or the duration of each target feature, and obtaining the first state evaluation result of the driver according to the current driving data, the current physiological data, the warning threshold interval and / or the duration of each target feature.

[0045] It should be noted that not all features in the physiological data contribute equally to the prediction ability of the model, and a feature selection strategy needs to be used to identify the most relevant features directly related to the target variable to simplify the model, reduce the risk of overfitting, and improve computational efficiency.

[0046] Specifically, the feature selection strategy can use a tree-based method to identify the most relevant features directly related to the target variable, for example:

[0047]

[0048] Among them, f is a feature, q iis the weight of feature f in the model.

[0049] Further, in some embodiments, the first state evaluation result of the driver is obtained according to the current driving data, the current physiological data, the reminding threshold interval and / or the duration of each target feature, including: determining the current value of each target feature from the current driving data and the current physiological data; generating the corresponding reminding information of any target feature in the case that the current value of any target feature is in the corresponding reminding threshold interval, and / or the duration of the current value of any target feature in the corresponding reminding threshold interval reaches the corresponding duration; and obtaining the first state evaluation result according to the reminding information of any target feature.

[0050] For example, assuming that two layers of decision layers are set, the weight of heart rate in the first layer decision tree is q1=0.1, and the weight of heart rate in the second layer decision tree is q2=0.05, then the target feature Im(f) of heart rate is 0.1+0.05=0.15, and by analogy, the weight of respiratory rate in the first layer decision tree is q1=0.05, and the weight of respiratory rate in the second layer decision tree is q2=0.05, then the target feature Im(f) of respiratory rate is 0.1+0.1=0.2, and by analogy, the weight of blood pressure in the first layer decision tree is q1=0.05, and the weight of blood pressure in the second layer decision tree is q2=0.05, then the target feature Im(f) of blood pressure is 0.05+0.05=0.1, if the driver fatigue is to be reminded, then the threshold interval is Im(f)>0.15, thereby determining the target features as heart rate and respiratory rate, when the Im(f) of respiratory rate is high, the current value is in the corresponding reminding threshold interval, the reminding information is generated, for example, when the respiratory rate is less than 8 times / minute or greater than 25 times / minute, the respiratory abnormality evaluation result is obtained, and at the same time, when the Im(f) of heart rate is high, the duration of the current value in the corresponding reminding threshold interval reaches the corresponding duration, the corresponding reminding information is generated, for example, when the heart rate is more than 120 times / minute and the duration is greater than 30 seconds, the cardiac arrest evaluation result is obtained.

[0051] Further, in order to ensure accuracy, the embodiments of the present application can also assist in judging and reminding the driver safety through the cloud prediction model, and based on the result of multi-modal data fusion, the cloud processing layer will generate a detailed health evaluation report, including the analysis results of the driver's physical state, emotional state, driving behavior and the like.

[0052] Specifically, the embodiments of the present application can also analyze and process the current driving data, the current physiological data, the current behavior data and the historical health data through the cloud to obtain the driver health condition analysis result (i.e. the second state evaluation result of the driver). As shown in Figure 2 Figure 2 ​A cloud algorithm processing flowchart is provided according to one embodiment of the present application.

[0053] The embodiments of the present application can be used in the context of monitoring physiological characteristics of a driver. Convolutional Neural Network (CNN) can be used to analyze image data, such as facial expressions of a driver, or extract features from sensor data, such as electrocardiogram. On the other hand, Long Short-Term Memory (LSTM) is good at processing dynamic signals by capturing long-term dependencies, and can also process data such as vehicle speed and acceleration, and finally obtain a situation report through a Softmax function.

[0054] Specifically, feature extraction is a key step in the learning process, in which a convolutional neural network is used to extract effective and discriminative features from pre-processed physiological signals. The convolution operation uses a convolution kernel to extract local features, and the mathematical expression is:

[0055] (f*g)(i,j) = ∑ m ∑ n f(m,n)g(i-m,j-n);

[0056] where f is the input feature map, representing the signal data after preprocessing, such as physiological signals (electrocardiogram, heart rate), image data (driver's face image) or sensor data (such as vehicle speed, acceleration), and g is the convolution kernel.

[0057] In order to introduce nonlinearity in the model and enable it to learn more complex patterns, a ReLU activation function is applied:

[0058] R(x) = max(0,(f*g)(i,j));

[0059] where (f*g)(i,j) is the output result of the convolution layer.

[0060] The pooling layer is used to reduce the dimension of the feature map and only keep the most significant features. The max-pooling operation selects the maximum value from a given region of the feature map, and the max-pooling operation is represented as:

[0061]

[0062] where r is the output after the activation function, and the max-pooling process not only reduces the size of the feature map, but also improves the computational efficiency of the model, thereby increasing its generalization ability. Through the pooling operation, the model can reduce the computational load while retaining the most important feature information, enhancing robustness.

[0063] After the convolution and pooling layers, the output of the CNN is a multi-dimensional feature map. In order to input these features into the LSTM network, a multi-dimensional output layer must be flattened into a one-dimensional vector, mathematically represented as:

[0064] Flatten(X) = [x1, x2, …, x n ];

[0065] Thus, the effective and discriminative features are extracted from the pre-processed physiological signals.

[0066] Further, after the features are extracted, the physiological state changes can be analyzed by the LSTM. The long short-term memory (LSTM) network is a powerful tool for processing time series, which can capture the temporal changes of sequence signals and effectively capture the long-term dependencies in time series. The input gate controls the preservation of the current input information, and the expression is:

[0067] i t = σ(W i · [h t-1 , x t ] + b i );

[0068] where W i is the weight of the input gate, σ is the sigmoid activation function, b i is the bias term, and x t is the input at the current time, which is the flattened feature vector, such as the value of heart rate or respiratory rate at time t.

[0069] The forgetting gate determines the degree of preservation of the previous state information, and the calculation formula is:

[0070] o t = σ(W o · [h t-1 , x t ] + b o );

[0071] where W o is the weight of the forgetting gate, and b o is the bias term. Through this forgetting gate, the LSTM can selectively forget unnecessary information and improve learning efficiency.

[0072] The update process of the cell state combines the influence of the input gate and the forgetting gate, and the formula is:

[0073] C t = f t C t-1 +i t tanh(W C · [h t-1 , xt ]+b C ;

[0074] This formula ensures efficient storage and updating of information, allowing the cell state to flexibly capture temporal information. Based on the current cell state, the hidden state update formula is:

[0075] h t =o t ·tanh(C t );

[0076] The hidden state update mechanism ensures that LSTM can retain important context information in time series. These formulas work together to enable the model to selectively store and forget information, allowing it to dynamically adjust its memory content to adapt to changing driving environments. The design of LSTM enables it to handle long and short dependencies in sequential data, ensuring timely capture of changes in driver physiological state.

[0077] Further, after the LSTM, a fully connected (dense) layer is added to the network for classification or regression tasks. This layer generates the final output by applying a linear transformation to the LSTM output:

[0078]

[0079] where, represents the predicted output, W is the weight matrix, h t is the final hidden state of LSTM, and b is the bias term.

[0080] Further, the commonly used cross-entropy loss function can be expressed as:

[0081]

[0082] where N is the number of samples, C is the number of categories, y i,c represents the true label, is the predicted value.

[0083] The fully connected layer uses the Softmax activation function to classify the final hidden state output of LSTM:

[0084]

[0085] This layer is responsible for generating the final prediction result, such as the probability of driver fatigue, providing important real-time feedback to ensure driving safety. Through this process, the model can convert the input physiological information into a specific assessment of the driver's state.

[0086] It should be noted that the preset cloud prediction model can be pre-trained.

[0087] In some embodiments, before obtaining the second state evaluation result of the driver based on the preset cloud prediction model according to the current driving data, the current physiological data, the current behavior data and the historical health data, the method further comprises: obtaining historical driving data of the vehicle, historical physiological data of the driver and behavior data of the driver; preprocessing the historical driving data, the historical physiological data, the behavior data and the historical health data respectively, and determining at least one key feature meeting a preset extraction condition from the preprocessed historical driving data, the preprocessed historical physiological data, the preprocessed behavior data and the preprocessed historical health data based on a preset feature extraction strategy; and training the preset neural network based on a preset cross-entropy loss function to obtain the preset cloud prediction model using the at least one key feature and a prediction classification result corresponding to the at least one key feature.

[0088] Specifically, when training the cloud prediction model, the embodiment of the application collects historical driving data of the vehicle, historical physiological data of the driver and behavior data of the driver, and trains the model using the Adam optimization algorithm and updates the parameters thereof through back propagation. The Adam optimizer combines the advantages of adaptive learning rate and momentum, making it efficient when training deep networks. The update rule of the Adam optimizer is as follows:

[0089]

[0090]

[0091]

[0092] wherein β1 and β2 are momentum decay rates, η is a learning rate, and ε is a constant.

[0093] Further, based on the cross-entropy loss function, the preset neural network is trained using the at least one key feature and a prediction classification result corresponding to the at least one key feature to obtain the preset cloud prediction model. Through these update rules, the model can adaptively adjust the learning rate to improve the training efficiency. Meanwhile, a Dropout layer is introduced to prevent overfitting. The CNN-LSTM combined method is very effective in monitoring the features of the driver, because it extracts spatial features from physiological signals through convolutional layers and captures time-dependent relationships through LSTM.

[0094] It should be noted that the training process can adopt the training method in related technologies, and details are not described herein to avoid redundancy.

[0095] In step S103, a health state monitoring result of the driver is generated according to the first state evaluation result and / or the second state evaluation result, and a health reminder is performed when the health state monitoring result meets a preset reminder condition.

[0096] The preset reminding condition is that there is a result in the first state evaluation result and / or the second state evaluation result that needs to remind the driver, for example, the heart rate is too fast, the driver is tired, and the face has a pathological pain symptom, etc.

[0097] Specifically, after obtaining the first state evaluation result, or the second state evaluation result, or the first state evaluation result and the second state evaluation result, the embodiments of the present application can perform corresponding processing. For example, after obtaining the first state evaluation result and the second state evaluation result, the embodiments of the present application can integrate the first state evaluation result and the second state evaluation result to obtain a health state monitoring result of the driver, and then determine whether there is a related calculation in the health state monitoring result that needs to be reminded of health, if there is, remind through the display screen or the loudspeaker of the vehicle, or provide personalized health suggestions, such as rest reminders, medical treatment prompts, etc.

[0098] Therefore, by analyzing the collected vehicle driving data, the current physiological and behavioral data of the driver and the historical health data through the feature selection strategy, the first state evaluation result is obtained, and the second state evaluation result is obtained by using the cloud prediction model to fuse the multi-source data, and the health state monitoring result is generated comprehensively, which solves the problem of low real-time and accuracy of the vehicle health monitoring system in the related art, and improves the comprehensiveness and accuracy of the evaluation of the health state of the driver.

[0099] Further, in some embodiments, after obtaining the second state evaluation result of the driver based on the preset cloud prediction model according to the current driving data, the current physiological data, the current behavioral data and the historical health data, the method further comprises: controlling the vehicle to display the health evaluation result and the health suggestion in the second state evaluation result.

[0100] For example, the health evaluation result will be transmitted back to the vehicle end safely through an encrypted communication protocol, update the user interface of the vehicle-mounted system, and trigger the corresponding alarm or feedback mechanism. For example, if it is detected that the driver is in an extreme fatigue state, the system will suggest the driver to stop and rest immediately; if it is detected that the heart rate of the driver is abnormal, the system will remind the driver to go to the hospital as soon as possible.

[0101] Further, in some embodiments, after collecting the current driving data, the current physiological data of the driver, the current behavior data of the driver, and the historical health data of the driver, the method further comprises: performing denoising processing on the current driving data, the current physiological data, the current behavior data, and the historical health data respectively to obtain denoised driving data, denoised physiological data, denoised behavior data, and denoised historical health data; performing normalization processing on the denoised driving data, the denoised physiological data, the denoised behavior data, and the denoised historical health data respectively to obtain normalized driving data, normalized physiological data, normalized behavior data, and normalized historical health data; and performing missing value processing on the normalized driving data, the normalized physiological data, the normalized behavior data, and the normalized historical health data respectively to obtain processed driving data, processed physiological data, processed behavior data, and processed historical health data.

[0102] Specifically, data preprocessing is crucial to ensure the best performance of the model, including denoising, normalization, and missing value processing. For example, due to the external interference (such as vehicle vibration, noise, etc.) that the collected physiological signals and data may be subjected to, the original data needs to be filtered and denoised first. Due to the differences in dimensions and amplitudes of physiological signals, normalization is necessary to ensure that all features are equally important in model performance. Finally, missing values in physiological signal data can cause model performance to decline. To fill in missing values, linear interpolation can be used.

[0103] For example, the embodiment of the present application uses a low-pass filter to suppress high-frequency noise, and the frequency response formula of the low-pass filter is:

[0104]

[0105] where h(f) is the frequency response, f is the frequency, f c is the cutoff frequency, and n is the order of the filter.

[0106] By setting an appropriate cutoff frequency, the low-frequency components of the signal can be effectively preserved, while unnecessary high-frequency noise is removed, thereby improving the signal-to-noise ratio.

[0107] The normalization process is:

[0108]

[0109] By converting the data to the range [0, 1], the problem of inconsistent dimensions during model training can be reduced, and the stability and convergence speed of the training process can be improved.

[0110] The linear interpolation formula is:

[0111]

[0112] wherein t is a time variable, x filled is an interpolated missing value, representing an estimated value of a certain signal at that time, for example, a missing data point of heart rate or respiratory rate, x prev is an observation value before the interpolation of the missing value, x next is an observation value after the interpolation of the missing value.

[0113] Thus, the data preprocessing of the historical driving data, the historical physiological data, the behavior data and the historical health data is completed, and the processed driving data, the processed physiological data, the processed behavior data and the processed historical health data are obtained.

[0114] To enable those skilled in the art to better understand the driver health reminding method of the embodiments of the present application, specific embodiments will be described below.

[0115] Figure 3 FIG. 1 shows a schematic diagram of a process framework of a driver health reminding method according to an embodiment of the present application, Figure 4 FIG. 2 shows a schematic diagram of an edge computing process according to an embodiment of the present application, Figure 5 FIG. 3 shows a flowchart of a driver health reminding method according to an embodiment of the present application.

[0116] In combination with Figure 3 and Figure 4 shown, the driver health reminding method includes sensor acquisition, edge computing, cloud computing and user interaction. The physiological signals, facial expressions, eye movement conditions and vehicle condition data of the driver are acquired in real time by various sensors. Since the acquired signals can be disturbed by external interference (such as vehicle vibration, noise, etc.), all raw data need to be preprocessed (such as filtering and noise reduction processing) first. Then, the signals acquired by different sensors are normalized to ensure that the data ranges of different data sources are consistent, facilitating subsequent feature extraction and analysis. At the same time, deep learning and multi-modal data fusion analysis are performed in the cloud to generate a health assessment report. Finally, according to the analysis results, the system provides intelligent feedback and auxiliary decision support to help the driver better manage his own health status. The user can view the health report through the vehicle display screen or mobile application, accept personalized suggestions, and provide feedback and score to the system.

[0117] As Figure 5 shown, the driver health reminding method first acquires data through sensors, and the user inputs medical history. After the two types of data are preprocessed by filtering and normalization, they are respectively input into vehicle-side edge computing and cloud-side deep learning. If the vehicle-side edge computing result reaches the alarm threshold interval and / or the continuous duration, an alarm reminder is triggered. If not, cloud-side edge computing is performed. Finally, regardless of the situation, the data will be archived and updated.

[0118] It should be noted that the vehicle end and the cloud end cooperate to evaluate, which can not increase the calculation load of the vehicle end data processing under the premise of ensuring the timeliness of the reminder and the accuracy of the evaluation, and the two-layer independent health evaluation mechanism can be regarded as an industrial-level redundant design, which can effectively avoid the problem of not timely and not in place due to single evaluation failure.

[0119] According to the driver health reminding method provided by the embodiment of the present application, the collected vehicle driving data, the current physiological and behavior data of the driver and the historical health data are analyzed by the feature selection strategy to obtain the first state evaluation result, and the second state evaluation result is obtained by using the cloud prediction model to fuse the multi-source data, and the health state monitoring result is generated comprehensively, which solves the problem of low real-time and accuracy of the vehicle-mounted health monitoring system in the related art, and improves the comprehensiveness and accuracy of the evaluation of the driver health state.

[0120] Secondly, the driver health reminding device according to the embodiment of the present application is described with reference to the accompanying drawings.

[0121] Figure 6 The block diagram of the driver health reminding device provided by the embodiment of the present application is shown.

[0122] As shown in the figure, Figure 6 The driver health reminding device includes a collection module 100, an evaluation module 200 and a warning module 300.

[0123] The collection module 100 is configured to collect the current driving data of the vehicle, the current physiological data of the driver, the current behavior data of the driver and the historical health data of the driver; the evaluation module 200 is configured to obtain the first state evaluation result of the driver based on the preset feature selection strategy according to the current driving data and the current physiological data, and obtain the second state evaluation result of the driver based on the preset cloud prediction model according to the current driving data, the current physiological data, the current behavior data and the historical health data; the warning module 300 is configured to generate the health state monitoring result of the driver according to the first state evaluation result and / or the second state evaluation result, and perform health reminding when the health state monitoring result meets the preset reminding condition.

[0124] Further, in some embodiments, the evaluation module 20 is specifically configured to determine at least one target feature based on a preset feature selection formula according to the current driving data and the current physiological data; determine the reminding threshold interval and / or the duration of each target feature, and obtain the first state evaluation result of the driver according to the current driving data, the current physiological data, the reminding threshold interval and / or the duration of each target feature.

[0125] Further, in some embodiments, the evaluation module 200 is further configured to: determine a current value of each target feature from the current driving data and the current physiological data; generate the reminding information corresponding to any target feature in a case that the current value of any target feature is in the corresponding reminding threshold interval, and / or the current value of any target feature is in the corresponding reminding threshold interval for a duration reaching the corresponding duration; and obtain the first state evaluation result according to the reminding information corresponding to any target feature.

[0126] Further, in some embodiments, before obtaining the second state evaluation result of the driver based on the preset cloud prediction model according to the current driving data, the current physiological data, the current behavior data and the historical health data, the evaluation module 200 is further configured to: obtain the historical driving data of the vehicle, the historical physiological data of the driver and the behavior data of the driver; pre-process the historical driving data, the historical physiological data, the behavior data and the historical health data respectively, and determine at least one key feature meeting a preset extraction condition from the pre-processed historical driving data, the pre-processed historical physiological data, the pre-processed behavior data and the pre-processed historical health data based on a preset feature extraction strategy; and train the preset neural network to obtain the preset cloud prediction model based on a preset cross-entropy loss function using the at least one key feature and the prediction classification result corresponding to the at least one key feature.

[0127] Further, in some embodiments, after obtaining the second state evaluation result of the driver based on the preset cloud prediction model according to the current driving data, the current physiological data, the current behavior data and the historical health data, the evaluation module 200 is further configured to: control the vehicle to display the health evaluation result and the health suggestion in the second state evaluation result.

[0128] Further, in some embodiments, after collecting the current driving data, the current physiological data of the driver, the current behavior data of the driver and the historical health data of the driver, the collection module 100 is further configured to: pre-process the current driving data, the current physiological data, the current behavior data and the historical health data respectively to obtain denoised driving data, denoised physiological data, denoised behavior data and denoised historical health data; normalize the denoised driving data, the denoised physiological data, the denoised behavior data and the denoised historical health data respectively to obtain normalized driving data, normalized physiological data, normalized behavior data and normalized historical health data; and process the normalized driving data, the normalized physiological data, the normalized behavior data and the normalized historical health data respectively to obtain processed driving data, processed physiological data, processed behavior data and processed historical health data.

[0129] It should be noted that the above explanation of the driver health reminding method embodiment is also applicable to the driver health reminding device, which will not be repeated here.

[0130] The driver health reminding device provided by the embodiment of the present application analyzes the collected vehicle driving data, the current physiological and behavioral data of the driver and the historical health data through the feature selection strategy, obtains a first state evaluation result, and simultaneously uses a cloud prediction model to fuse multiple source data to obtain a second state evaluation result, thereby comprehensively generating a health state monitoring result, and solving the problem of low real-time performance and accuracy of the vehicle-mounted health monitoring system in the related art, and improving the comprehensiveness and accuracy of the evaluation of the health state of the driver.

[0131] Figure 7 A structural schematic diagram of an electronic device according to an embodiment of the present application is provided. The electronic device can include:

[0132] The memory 701, the processor 702 and the computer program stored in the memory 701 and executable on the processor 702.

[0133] The processor 702 implements the driver health reminding method provided in the above embodiments when executing the program.

[0134] Further, the electronic device further includes:

[0135] The communication interface 703 is used for communication between the memory 701 and the processor 702.

[0136] The memory 701 is used to store the computer program executable on the processor 702.

[0137] The memory 701 can include a high-speed RAM memory, and can also include a non-volatile memory, for example, at least one disk memory.

[0138] If the memory 701, the processor 702 and the communication interface 703 are independently implemented, the communication interface 703, the memory 701 and the processor 702 can be connected to each other through a bus and complete communication between each other. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. For the convenience of representation, Figure 7Only one bus or only one type of bus can exist, however.

[0139] Optionally, in a specific implementation, if the memory 701, the processor 702 and the communication interface 703 are integrated on a chip, the memory 701, the processor 702 and the communication interface 703 can complete the communication among each other through an internal interface.

[0140] The processor 702 can be a central processing unit (CPU) or an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement embodiments of the present application.

[0141] In addition, the embodiment of the present application further provides a computer readable storage medium, which stores a computer program, and the program is executed by a processor to implement the driver health reminding method.

[0142] In addition, the embodiment of the present application further provides a computer program product, which includes a computer program, and the computer program is executed to implement the driver health reminding method.

[0143] In the description of the present application, the description of the terms "one embodiment", "some embodiments", "an example", "a specific example", or "some examples" means that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. In the present application, the illustrative description of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any appropriate manner in any one or N embodiments or examples. In addition, the person skilled in the art can combine and combine the different embodiments or examples described in the present application and the features of the different embodiments or examples without contradiction.

[0144] In addition, the terms "first", "second" are only for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined with "first", "second" can explicitly or implicitly include at least one of the features. In the description of the present application, the meaning of "N" is at least two, for example, two, three, etc., unless otherwise specifically limited.

[0145] Any process or method described in a flowchart or otherwise described herein can be understood as representing code modules, segments, or portions of code that include one or more executable instructions for implementing the specified logical functions or steps, and the various embodiments of the application include alternative implementations of the application in which the steps are performed in an order different from the order shown or discussed, including substantially concurrently or in reverse order, depending upon the functionality involved, as will be understood by those having ordinary skill in the art.

[0146] It should be understood that aspects of the application can be implemented in hardware, software, firmware or combinations thereof. In the above embodiments, the steps or methods can be implemented in software or firmware that is stored in memory and executed by a suitable instruction execution system. As well as in hardware, as in another embodiment, any of the following technologies can be used to implement the above described embodiments: a combination of logic gates, discrete logic circuits having logic gates for implementing logic functions upon an application of data signals, application specific integrated circuits having appropriate combinational logic gates, programmable gate arrays (PGA), field programmable gate arrays (FPGA), and the like.

[0147] It will be appreciated by those skilled in the art that the steps of the above-described embodiments can be carried out by program instructions stored in a computer readable storage medium and executed by a corresponding instruction execution system. The program instructions can be stored in a computer readable storage medium, which in execution, includes one or more of the steps of the embodiments.

Claims

1. A driver health reminding method, characterized in that, The method comprises the following steps: collecting current driving data of a vehicle, current physiological data of a driver, current behavior data of the driver, and historical health data of the driver; obtaining a first state evaluation result of the driver based on preset feature selection strategies and the current driving data and the current physiological data, and obtaining a second state evaluation result of the driver based on a preset cloud prediction model and the current driving data, the current physiological data, the current behavior data, and the historical health data; generating a health state monitoring result of the driver according to the first state evaluation result and / or the second state evaluation result, and performing health prompting when the health state monitoring result meets a preset prompting condition.

2. The method of claim 1, wherein, The method of obtaining the first state evaluation result of the driver based on the preset feature selection strategies and the current driving data and the current physiological data comprises: determining at least one target feature based on a preset feature selection formula and the current driving data and the current physiological data; determining a prompting threshold interval and / or a duration of each target feature, and obtaining the first state evaluation result of the driver based on the current driving data, the current physiological data, the prompting threshold interval and / or the duration of each target feature.

3. The method of claim 2, wherein, The method of obtaining the first state evaluation result of the driver based on the current driving data, the current physiological data, the prompting threshold interval and / or the duration of each target feature comprises: determining a current value of each target feature from the current driving data and the current physiological data; generating prompting information corresponding to any target feature in a case that the current value of the target feature is in a corresponding prompting threshold interval, and / or the duration of the current value of the target feature in the corresponding prompting threshold interval reaches a corresponding duration, and obtaining the first state evaluation result according to the prompting information corresponding to the target feature.

4. The method of claim 1, wherein, Before obtaining the second state evaluation result of the driver based on the preset cloud prediction model and the current driving data, the current physiological data, the current behavior data, and the historical health data, the method further comprises: obtaining historical driving data of the vehicle, historical physiological data of the driver, and behavior data of the driver; preprocessing the historical driving data, the historical physiological data, the behavior data, and the historical health data respectively, and determining at least one key feature meeting a preset extraction condition from the preprocessed historical driving data, the preprocessed historical physiological data, the preprocessed behavior data, and the preprocessed historical health data based on a preset feature extraction strategy; training a preset neural network based on a preset cross-loss function and the at least one key feature and a prediction classification result corresponding to the at least one feature to obtain the preset cloud prediction model.

5. The method of claim 1, wherein, After obtaining the second state evaluation result of the driver based on the preset cloud prediction model according to the current driving data, the current physiological data, the current behavior data and the historical health data, the method further comprises: controlling the vehicle to display the health evaluation result and the health suggestion in the second state evaluation result.

6. The method of claim 1, wherein, After collecting the current driving data, the current physiological data of the driver, the current behavior data of the driver and the historical health data of the driver, the method further comprises: performing denoising processing on the current driving data, the current physiological data, the current behavior data and the historical health data respectively to obtain denoised driving data, denoised physiological data, denoised behavior data and denoised historical health data; performing normalization processing on the denoised driving data, the denoised physiological data, the denoised behavior data and the denoised historical health data respectively to obtain normalized driving data, normalized physiological data, normalized behavior data and normalized historical health data; performing missing value processing on the normalized driving data, the normalized physiological data, the normalized behavior data and the normalized historical health data respectively to obtain processed driving data, processed physiological data, processed behavior data and processed historical health data.

7. A driver health reminding device, characterized by, The device comprises: a collection module configured to collect current driving data of a vehicle, current physiological data of a driver, current behavior data of the driver and historical health data of the driver; an evaluation module configured to obtain a first state evaluation result of the driver based on a preset feature selection strategy according to the current driving data and the current physiological data, and obtain a second state evaluation result of the driver based on a preset cloud prediction model according to the current driving data, the current physiological data, the current behavior data and the historical health data; a warning module configured to generate a health state monitoring result of the driver according to the first state evaluation result and / or the second state evaluation result, and perform health prompting when the health state monitoring result meets a preset prompting condition.

8. An electronic device, comprising: comprise: a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor executes the program to implement the driver health prompting method according to any one of claims 1-6.

9. A computer-readable storage medium having stored thereon a computer program, characterized in that, The program is executed by the processor to implement the driver health prompting method according to any one of claims 1-6.

10. A computer program product comprising a computer program, characterized in that, The computer program is executed by the processor to implement the driver health prompting method according to any one of claims 1-6.

Images