A method and apparatus for predicting body temperature

Abstract

The application provides a method and device for predicting body temperature, the method comprising: obtaining at least one set of body feature values at a first time and at least one body temperature at a second time; inputting the at least one set of body feature values at the first time into a first model to obtain a first body temperature at a third time; inputting the at least one body temperature at the second time into a second model to obtain a second body temperature at the third time; and determining a third body temperature at the third time according to the first body temperature and the second body temperature. According to the method, the third body temperature at the third time is determined according to the first body temperature and the second body temperature, the third body temperature is related to both the body features of the user and the body temperature change of the user, and therefore the third body temperature can reflect the real physical condition of the user, and the body temperature of the user in a future period of time is accurately predicted.

Description

Technical Field

[0001] This application relates to the field of computers, and more particularly to a method and apparatus for predicting body temperature. Background Technology

[0002] With the development of communication technology and the widespread application of artificial intelligence, people are no longer satisfied with obtaining the current body temperature of animals or humans, but are also paying more attention to body temperature in the future so as to take countermeasures in advance.

[0003] How to accurately predict body temperature over a future period remains to be solved. Summary of the Invention

[0004] This application provides a method and apparatus for predicting body temperature, which can accurately predict body temperature over a future period of time.

[0005] In a first aspect, embodiments of this application provide a method for predicting body temperature. This method can be executed by a device for predicting body temperature, which can be a terminal device or a module for a terminal device, or a server or a module for a server. This application does not limit the subject executing the method. The method includes: acquiring at least one set of body feature values ​​at a first time and at least one body temperature at a second time; inputting at least one set of body feature values ​​at the first time into a first model to predict a first body temperature at a third time; inputting at least one body temperature at the second time into a second model to predict a second body temperature at the third time; and determining a third body temperature at the third time based on the first and second body temperatures.

[0006] In the above scheme, the first body temperature is determined based on at least one set of physical characteristic values, reflecting the user's true physical condition; the second body temperature is determined based on at least one body temperature at a second time point, reflecting the user's temperature changes. A third body temperature is determined based on the first and second body temperatures. This third body temperature is related to both the user's physical characteristics and temperature changes, thus reflecting the user's true physical condition and achieving accurate prediction of the user's body temperature over a future period.

[0007] In one possible implementation, the first body temperature and the second body temperature are weighted to obtain the third body temperature.

[0008] The above scheme weights the first and second body temperatures, balancing the impact of body characteristic values ​​and body temperature on predicting body temperature over a future period, thus achieving accurate prediction of the user's body temperature over a future period.

[0009] In one possible implementation, the body characteristic values ​​include at least one of the following: systolic blood pressure, diastolic blood pressure, heart rate, or blood oxygen.

[0010] The above scheme, with its various types of body characteristic values, can comprehensively reflect the user's true physical condition and accurately predict the user's body temperature over a future period of time.

[0011] In one possible implementation, the first time is a first moment, and at least one set of body feature values ​​includes a set of body feature values ​​at the first moment; or, the first time is a first time range, and at least one set of body feature values ​​includes at least one set of body feature values ​​within the first time range or a weighted sum of multiple sets of body feature values ​​within the first time range.

[0012] The above scheme can include at least one set of body characteristic values, which may include a set of body characteristic values ​​at the first moment, or at least one set of body characteristic values ​​within the first time range, or a weighted value of multiple sets of body characteristic values ​​within the first time range. By obtaining at least one set of body characteristic values ​​that are time-related, it can reflect the user's real physical condition changes and thus achieve accurate prediction of body temperature in the future.

[0013] In one possible implementation, the second time is a second moment, and at least one body temperature includes the body temperature at the second moment; or, the second time is a second time range, and at least one body temperature includes at least one body temperature within the second time range or a weighted value of multiple body temperatures within the second time range.

[0014] The above scheme can include at least one body temperature, which may be the body temperature at a second moment; or, at least one body temperature may also include at least one body temperature within a second time range or a weighted value of multiple body temperatures within a second time range, so as to obtain at least one body temperature that is time-related, which can reflect the user's real body temperature changes and thus enable accurate prediction of body temperature in the future.

[0015] In one possible implementation, the first model uses the XGBoost model, and the second model uses the BERT model.

[0016] The aforementioned approach utilizes the XGBoost model to correlate body feature values ​​with user body temperature, reflecting trends in temperature changes and thus enabling accurate predictions of body temperature over a future period. The BERT model's word vectors take into account semantic relevance within the context, and its pre-training with massive amounts of unsupervised data further enhances the accuracy of temperature prediction.

[0017] In one possible implementation, at least one set of body feature values ​​at a fourth time point and body temperature at a fifth time point are obtained, wherein the fourth time point is earlier than the fifth time point; the at least one set of body feature values ​​at the fourth time point are used as input to a first model, and the body temperature at the fifth time point is used as output to train the first model.

[0018] The above scheme trains the first model using at least one set of time-related body feature values, which can reflect the user's real physical condition changes and thus accurately predict body temperature in the future.

[0019] In one possible implementation, at least one body temperature at a sixth time point and a body temperature at a seventh time point are obtained, wherein the sixth time point is earlier than the seventh time point; the at least one body temperature at the sixth time point is used as the input of the second model, and the body temperature at the seventh time point is used as the output of the second model, and the second model is trained.

[0020] The above approach trains the second model using at least one time-correlated body temperature reading, which can reflect the user's real physical condition changes and thus accurately predict body temperature over a future period.

[0021] In one possible implementation, an alarm instruction is issued to the user, the alarm instruction including the third body temperature and the corresponding response measures.

[0022] The above solution sends alerts to users, helping them to take timely countermeasures.

[0023] Secondly, embodiments of this application provide an apparatus for predicting body temperature, comprising: an acquisition unit, a prediction unit, and a determination unit. The acquisition unit is configured to acquire at least one set of body characteristic values ​​at a first time and at least one body temperature at a second time; the prediction unit is configured to input the at least one set of body characteristic values ​​at the first time into a first model to predict a first body temperature at a third time; and input at least one body temperature at the second time into a second model to predict a second body temperature at the third time; the determination unit is configured to determine a third body temperature at the third time based on the first and second body temperatures.

[0024] In one possible implementation, a unit is defined, specifically used to weight the first body temperature and the second body temperature to obtain the third body temperature.

[0025] In one possible implementation, the body characteristic values ​​include at least one of the following: systolic blood pressure, diastolic blood pressure, heart rate, or blood oxygen.

[0026] In one possible implementation, the first time is a first moment, and at least one set of body feature values ​​includes a set of body feature values ​​at the first moment; or, the first time is a first time range, and at least one set of body feature values ​​includes at least one set of body feature values ​​within the first time range or a weighted sum of multiple sets of body feature values ​​within the first time range.

[0027] In one possible implementation, the second time is a second moment, and at least one body temperature includes the body temperature at the second moment; or, the second time is a second time range, and at least one body temperature includes at least one body temperature within the second time range or a weighted value of multiple body temperatures within the second time range.

[0028] In one possible implementation, the first model uses the XGBoost model, and the second model uses the BERT model.

[0029] In one possible implementation, the acquisition unit is further configured to acquire at least one set of body characteristic values ​​at a fourth time and body temperature at a fifth time, wherein the fourth time is earlier than the fifth time.

[0030] In one possible implementation, the device further includes a training unit, which is used to train the first model by taking at least one set of body feature values ​​at a fourth time as input to the first model and taking the body temperature at a fifth time as output to the first model.

[0031] In one possible implementation, the acquisition unit is further configured to acquire at least one body temperature at a sixth time and a body temperature at a seventh time, wherein the sixth time is earlier than the seventh time.

[0032] In one possible implementation, a training unit is used to train the second model by taking at least one body temperature at a sixth time as input and the body temperature at a seventh time as output.

[0033] In one possible implementation, the device further includes an alarm unit for issuing an alarm indication to the user, the alarm indication including a third body temperature and corresponding measures for responding to the third body temperature.

[0034] Thirdly, embodiments of this application also provide a computing device, including:

[0035] Memory, used to store program instructions;

[0036] The processor is configured to invoke program instructions stored in the memory and execute any method for implementing the first aspect described above, according to the obtained program instructions.

[0037] Fourthly, embodiments of this application also provide a computer-readable storage medium storing computer-readable instructions, which, when read and executed by a computer, implement any of the methods described in the first aspect.

[0038] Fifthly, embodiments of this application provide a computer program product, including a computer program executable by a computer device, which, when run on the computer device, causes the computer device to perform any of the methods described in the first aspect. Attached Figure Description

[0039] Figure 1 A flowchart illustrating a method for predicting body temperature provided in an embodiment of this application;

[0040] Figure 2 A schematic diagram of an improved BERT model provided in an embodiment of this application;

[0041] Figure 3 A flowchart illustrating a method for determining a first model provided in an embodiment of this application;

[0042] Figure 4 A flowchart illustrating a method for determining a second model provided in an embodiment of this application;

[0043] Figure 5 A schematic diagram of a device for predicting body temperature provided in an embodiment of this application;

[0044] Figure 6 This is a schematic diagram of a device for predicting body temperature provided in an embodiment of this application. Detailed Implementation

[0045] Figure 1 This is a flowchart illustrating a method for predicting body temperature according to an embodiment of this application. The method can be executed by a device for predicting body temperature, which can be a terminal device or a module for a terminal device, or a server or a module for a server. This application does not limit the entity executing this method.

[0046] In one possible implementation, this application can be used to predict the highest body temperature in the future, the lowest body temperature in the future, and the time when the highest or lowest body temperature will occur.

[0047] In one possible implementation, this application can be used not only to predict the body temperature of a human body over a future period of time, but also to predict the body temperature of an animal over a future period of time.

[0048] The method includes the following steps:

[0049] Step 101: Obtain at least one set of body characteristic values ​​at a first time and at least one body temperature at a second time.

[0050] In one possible implementation, at least one set of body characteristic values ​​of the user at a given moment is acquired through devices such as a smartwatch or blood pressure monitor. These body characteristic values ​​are indicators reflecting the user's physical performance, including their body temperature characteristics. The body characteristic values ​​include one or more of the following: systolic blood pressure, diastolic blood pressure, heart rate, or blood oxygen saturation. At least one body temperature is acquired at a second moment through a smartwatch or thermometer. This application does not limit the method of collecting the user's body characteristic values ​​or body temperature.

[0051] In one possible implementation, the first time is defined as the first moment, and at least one set of body feature values ​​includes a set of body feature values ​​from the first moment. For example, a set of body feature values ​​obtained at 12:00 via a smartwatch is shown in Table 1.

[0052] Table 1

[0053]

[0054] In another possible implementation, where "first time" refers to a "first time range," then at least one set of body feature values ​​can include at least one set of body feature values ​​within the first time range, or it can be a weighted sum of multiple sets of body feature values ​​within the first time range. For example, using a smartwatch, body feature values ​​are collected every half hour to obtain multiple sets of body feature values ​​between 12:00 and 14:00, as shown in Table 2. These multiple sets of body feature values ​​can be input into a first model to predict the first body temperature at a third time. Alternatively, the multiple sets of body feature values ​​in Table 2 can be weighted by category to determine a single set of body feature values, and then used to predict the first body temperature at the third time.

[0055] Table 2

[0056]

[0057] The above scheme can include at least one set of body characteristic values, which may include a set of body characteristic values ​​at the first moment, or at least one set of body characteristic values ​​within the first time range, or a weighted value of multiple sets of body characteristic values ​​within the first time range. By obtaining at least one set of body characteristic values ​​that are time-related, it can reflect the user's real physical condition changes and thus achieve accurate prediction of body temperature in the future.

[0058] In one possible implementation, the second time is a second moment, and at least one body temperature includes the body temperature at the second moment.

[0059] In another possible implementation, where "second time" refers to a second time range, then "at least one body temperature" can include at least one body temperature within the second time range, or it can be a weighted sum of multiple body temperatures within the second time range. For example, using a smartwatch, body temperature is collected every 10 minutes to obtain multiple body temperatures between 12:00 and 13:00, as shown in Table 3. These multiple body temperatures can be input into a first model to predict the second body temperature at a third time, or the multiple body temperatures in Table 3 can be weighted, and the weighted temperature can be used to predict the second body temperature at the third time.

[0060] Table 3

[0061]

[0062] In one possible implementation, the first time and the second time can be the same time period, for example, both the first time and the second time are from 12:00 to 13:00. The first time and the second time can also partially overlap in time, for example, the first time is from 12:00 to 13:00, and the second time is from 12:30 to 13:00.

[0063] The above scheme can include at least one body temperature, which may be the body temperature at a second moment; or, at least one body temperature may also include at least one body temperature within a second time range or a weighted value of multiple body temperatures within a second time range, so as to obtain at least one body temperature that is time-related, which can reflect the user's real body temperature changes and thus enable accurate prediction of body temperature in the future.

[0064] Step 102: Input at least one set of body feature values ​​at the first time point into the first model to predict the first body temperature at the third time point.

[0065] In some embodiments of the application, the first body temperature is the first highest body temperature.

[0066] In one possible implementation, at least one set of body feature values ​​at the first moment is converted into a vector form and input into the first model. For example, the contents of Table 1 above are converted into a vector form and input into the first model, that is, [37, 110mmHg, 70mmHg, 90, 99%] are input into the first model.

[0067] In one possible implementation, the third time is a third time range. At least one set of body feature values ​​from the first time is input into the first model to predict the first body temperature within the third time range, such as predicting the body temperature between 10 o'clock and 12 o'clock.

[0068] In another possible implementation, the third time is the third moment. At least one set of body feature values ​​from the first time is input into the first model to predict the first body temperature within the range of the third moment, for example, predicting the body temperature half an hour later.

[0069] In one possible implementation, the first model employs a machine learning model, and this application does not limit the type of machine learning model used.

[0070] In some embodiments of the application, the first model employs the XGBoost model. The XGBoost model can correlate body feature values ​​with the user's body temperature, reflect the trend of changes in the user's body temperature, and thus achieve accurate prediction of body temperature over a future period.

[0071] Step 103: Input at least one body temperature at the second time point into the second model to predict the second body temperature at the third time point.

[0072] In some embodiments of the application, the second body temperature is the second highest body temperature.

[0073] In one possible implementation, at least one temperature at the second time point is converted into a vector form and input into the second model. For example, the contents of Table 3 above are converted into a vector form and input into the second model, that is, [36.5, 36.6, 36.7, 36.8, 36.9, 37, 37.1] are input into the second model.

[0074] In one possible implementation, the second model employs a deep learning model, and this application does not limit the type of deep learning model used.

[0075] In some embodiments of the application, the second model employs the BERT model. The word vectors of the BERT model take into account the semantic relevance of the context, and the BERT model is pre-trained on massive amounts of unsupervised data, which can improve the accuracy of body temperature prediction.

[0076] In another possible implementation, the second model employs an improved BERT model, the structure of which can be as follows: Figure 2 As shown, the model includes an input layer, an embedding layer, a fully connected layer, a hidden layer, and an output layer. The specific improvement method is to add at least one fully connected layer after the embedding layer of the traditional BERT model. For example, six fully connected layers are added after the embedding layer, and the output is processed through a softmax activation function to predict the second body temperature at the third time point. Adding at least one fully connected layer after the embedding layer makes the features extracted by the second model more correlated, increasing the accuracy of the second model in predicting body temperature.

[0077] In one possible implementation, the improved BERT model can be a multi-class model, and the category of the second body temperature at the third time point can be one or more of 36℃, 36.5℃, 37℃, 37.5℃, 38℃, 38.5℃, 39℃, 39.5℃ or 40℃, or other forms of categories, which are not limited in this application.

[0078] Step 104: Determine the third body temperature at the third time point based on the first and second body temperatures.

[0079] In some embodiments of the application, the third body temperature is the third highest body temperature.

[0080] In one possible implementation, the first and second body temperatures are weighted to obtain a third body temperature. The weighting formula for the first and second body temperatures is shown in formula (1). This scheme, by weighting the first and second body temperatures, can balance the influence of body characteristic values ​​and body temperature on predicting body temperature in the future, thus achieving accurate prediction of the user's body temperature in the future.

[0081] …….(1)

[0082] Where x1 is the first body temperature, x2 is the second body temperature, y is the third body temperature, a and b are constant factors, and a + b = 1. .

[0083] One possible implementation involves issuing an alarm to the user, which includes a third body temperature and corresponding measures. The alarm can be of different types, such as a high-temperature alarm, a low-temperature alarm, or a temperature rise alarm. For example, if the third body temperature is high, a high-temperature alarm will be issued to the user, reminding them that their temperature will reach a high level in the near future and requesting them to prepare fever-reducing medication and patches.

[0084] In the above scheme, the first body temperature is determined based on at least one set of physical characteristic values, reflecting the user's true physical condition; the second body temperature is determined based on at least one body temperature at a second time point, reflecting the user's temperature changes. A third body temperature is determined based on the first and second body temperatures. This third body temperature is related to both the user's physical characteristics and temperature changes, thus reflecting the user's true physical condition and achieving accurate prediction of the user's body temperature over a future period.

[0085] The following concrete example illustrates how steps 101 to 104 are implemented:

[0086] The smartwatch obtains a set of body characteristic values ​​of the user at 12 o'clock, which are: body temperature 37 degrees Celsius, systolic blood pressure 110 mmHg, diastolic blood pressure 70 mmHg, heart rate 90 beats / minute, and blood oxygen 99%. Similarly, the smartwatch measures the user's body temperature every 10 minutes to obtain a set of body temperatures between 11 o'clock and 12 o'clock, which are [36.5, 36.5, 36.6, 36.7, 36.7, 36.5, 36.5, 36.6, 36.7, 36.7].

[0087] Input a set of body feature values ​​at 12 o'clock into the first model to predict the first body temperature at the third time. For example, if the predicted body temperature within 3 hours is 37.1 degrees, then the first body temperature at the third time is 37.1 degrees.

[0088] The user's body temperature between 11:00 and 12:00 was input into the second model to predict the second body temperature at the third time, that is, the body temperature within 3 hours, and the result was 36.9 degrees.

[0089] The first and second body temperatures are weighted to obtain the third body temperature. Here, the constant factors a and b in formula (1) are both taken as 0.5, so the third body temperature is 0.5. 37.1 + 0.5 36.9 = 37.

[0090] In summary, if a user's body temperature is 37 degrees Celsius within the next 3 hours, an alert can be issued to the user, indicating that the user's body temperature and physical condition are stable within the next 3 hours, and no cooling measures are required in advance.

[0091] In one possible implementation, the first model needs to be trained and deployed before the first body temperature at the third time point is predicted using the first model.

[0092] One possible implementation involves acquiring at least one set of body feature values ​​at a fourth time point and body temperature at a fifth time point, where the fourth time point is earlier than the fifth time point. The at least one set of body feature values ​​at the fourth time point is used as input to a first model, and the body temperature at the fifth time point is used as output to train the first model. This approach, by using at least one set of time-correlated body feature values ​​to train the first model, can reflect the user's actual changes in physical condition, thereby enabling accurate prediction of body temperature over a future period.

[0093] In one possible implementation, the fourth time is the fourth moment, and at least one set of body feature values ​​includes a set of body feature values ​​at the fourth moment; or, the fourth time is a fourth time range, and at least one set of body feature values ​​includes at least one set of body feature values ​​within the fourth time range.

[0094] Figure 3A flowchart illustrating a method for determining a first model provided in this application embodiment, the method comprising the following steps:

[0095] Step 201, Data Acquisition.

[0096] In one possible implementation, at least one set of the user's physical characteristic values ​​is obtained through devices such as smartwatches and blood pressure monitors. This set of physical characteristic values ​​includes one or more of the following: systolic blood pressure, diastolic blood pressure, heart rate, or blood oxygen saturation. For example, a set of physical characteristic values ​​of the user at time t1 is obtained, along with the user's body temperature over the time range [t1, t1+m], where t1 is greater than 0 and m is a positive integer. Generally, multiple sets of physical characteristic values ​​of the user over a consecutive month, as well as the body temperature for the corresponding future time period within that month, can be obtained as training data for the first model. This application does not limit the size of the training data. The multiple sets of physical characteristic values ​​of the user over a consecutive month constitute the at least one set of physical characteristic values ​​at the fourth time point mentioned above, and the body temperature for the corresponding future time period within that month constitutes the body temperature at the fifth time point mentioned above.

[0097] Step 202, data preprocessing.

[0098] One possible implementation involves filtering the data obtained in step 201, deleting data that does not meet the requirements or is contrary to common sense, or replacing it with other data. For example, body temperature data above 50 degrees Celsius or below 33 degrees Celsius are contrary to common sense and can be deleted. Alternatively, the average of all body temperature data obtained in step 201 can be used to replace these contrary-to-common-sense data. Or, these contrary-to-common-sense body temperature data can be modified based on theory or experience to make them conform to the requirements or common sense.

[0099] In one possible implementation, if a certain body characteristic value in a set of body characteristic values ​​is missing, then that set of body characteristic values ​​can be deleted, or the missing body characteristic value can be replaced with the body characteristic value corresponding to other data. For example, a set of body characteristic values ​​may include body temperature, systolic blood pressure, diastolic blood pressure, heart rate, and blood oxygen saturation, but due to a problem with devices such as smartwatches, the body temperature data for this set of body characteristic values ​​is not measured. In this case, the set of body characteristic values ​​can be deleted, or the average body temperature of all data obtained in step 201 can be used to replace the body temperature data for this set of body characteristic values.

[0100] In one possible implementation, the training data of the first model is normalized by normalizing the training data of each feature to the range of [0,1] according to the maximum and minimum values ​​and the difference between the maximum and minimum values ​​within that feature, thereby increasing the discriminative power of the training data and improving the performance of the first model.

[0101] Step 203: Train the first model.

[0102] In one possible implementation, at least one set of body feature values ​​is used as the input to a first model, and the corresponding body temperature over a future period of time for these at least one set of body feature values ​​is used as the output of the first model. For example, the body feature value at time t1 is selected as the input to the first model, and the body temperature within the time range [t1+1, t1+m] is selected as the output of the first model to train the first model.

[0103] In one possible implementation, the first model is a multi-classification model, and the output category of the first model can be one or more of 36℃, 36.5℃, 37℃, 37.5℃, 38℃, 38.5℃, 39℃, 39.5℃ or 40℃, or other forms of categories, which are not limited in this application.

[0104] Step 204: Deploy the first model.

[0105] In one possible implementation, the first model trained in step 203 is deployed. The first model can be deployed on a terminal device or on a server. This application does not limit this.

[0106] In one possible implementation, the second model needs to be trained and deployed before it can be used to predict the second body temperature at the third time.

[0107] One possible implementation involves acquiring at least one body temperature at a sixth time point and a body temperature at a seventh time point, where the sixth time point is earlier than the seventh time point. The at least one body temperature at the sixth time point is used as input to a second model, and the body temperature at the seventh time point is used as output to train the second model. This approach, by using at least one time-correlated body temperature to train the second model, can reflect the user's actual changes in physical condition, thereby enabling accurate prediction of body temperature over a future period.

[0108] In one possible implementation, the sixth time is the sixth moment, and at least one body temperature includes the body temperature at the sixth moment; or, the sixth time is a sixth time range, and at least one body temperature includes at least one body temperature within the sixth time range.

[0109] Figure 4A flowchart illustrating a method for determining a second model provided in this application embodiment, the method comprising the following steps:

[0110] Step 301, Data Acquisition.

[0111] One possible implementation involves acquiring at least one of the user's body temperatures via a smartwatch or thermometer. For example, a smartwatch can acquire the user's body temperature in real-time over 24 hours, and all body temperatures within the time series are segmented into m+n time windows to form the training data for a second model, where m is greater than 0 and n is greater than 0. Segmenting all body temperatures within the time series into m+n time windows allows for the acquisition of a large amount of training data for the second model. Furthermore, the acquired training data is temporally correlated; that is, using body temperatures within the time range [t2-n, t2] as input samples for the second model and body temperatures within the time range [t2+1, t2+m] as output samples enables prediction of body temperatures within the [t2+1, t2+m] time range, thus allowing for prediction of body temperatures at a future time period.

[0112] The following example, Table 4, illustrates how to extract training data for the second model. In Table 4, the first row represents the body temperature values ​​measured from 1:01 to 1:10, measured once every minute for a total of 10 minutes. Each measurement showed a body temperature of 37 degrees Celsius, resulting in 10 body temperature data points.

[0113] With time windows n=3 and m=2, body temperatures within the time range [t2-n, t2] are selected as input samples for the second model, and body temperatures within the time range [t2+1, t2+m] are selected as output samples for the second model, where t2 is greater than 0.

[0114] For example, if t2 is 1:03, then the body temperature in [1:01, 1:03] is taken as the first input data of the second model, and the body temperature value in [1:04, 1:05] is taken as the output data corresponding to the first input data, as shown in the second row of Table 4.

[0115] For example, if t2 is 1:04, then the body temperature at [1:02, 1:04] is used as the second input data of the second model, and the body temperature value at [1:05, 1:06] is used as the output data corresponding to the second input data, as shown in the third row of Table 4.

[0116] Table 4

[0117]

[0118] Step 302, data preprocessing.

[0119] In one possible implementation, the data in step 301 is preprocessed, including one or more of filtering, vectorization, or normalization. This process is similar to the data preprocessing in step 202 above, and will not be described in detail here.

[0120] Step 303: Train the second model.

[0121] In one possible implementation, at least one set of body temperature values ​​is used as input to the second model, and the corresponding body temperatures within a future time period of this set of body temperature values ​​are used as output to the second model. For example, body temperatures within the time range of [t2-n, t2] are selected as input samples for the second model, and body temperatures within the time range of [t2+1, t2+m] are selected as output samples for the second model, and the second model is trained.

[0122] In one possible implementation, the second model is a multi-classification model. The output category of the second model can be one or more of 36℃, 36.5℃, 37℃, 37.5℃, 38℃, 38.5℃, 39℃, 39.5℃ or 40℃, or other types of categories. This application does not limit this.

[0123] Step 304: Deploy the second model.

[0124] In one possible implementation, the second model trained in step 303 is deployed. The second model can be deployed on a terminal device or on a server. This application does not limit this.

[0125] Based on the same technological concept Figure 5 An exemplary embodiment of this application illustrates a device 500 for predicting body temperature. For example... Figure 5 As shown, it includes: an acquisition unit 501, a prediction unit 502, and a determination unit 503. The acquisition unit 501 is used to acquire at least one set of body feature values ​​at a first time and at least one body temperature at a second time; the prediction unit 502 is used to input at least one set of body feature values ​​at the first time into a first model to predict a first body temperature at a third time; and input at least one body temperature at the second time into a second model to predict a second body temperature at the third time; the determination unit 503 is used to determine a third body temperature at the third time based on the first body temperature and the second body temperature.

[0126] In one possible implementation, the determining unit 503 is specifically used to weight the first body temperature and the second body temperature to obtain the third body temperature.

[0127] In one possible implementation, the body characteristic values ​​include at least one of the following: systolic blood pressure, diastolic blood pressure, heart rate, or blood oxygen.

[0128] In one possible implementation, the first time is a first moment, and at least one set of body feature values ​​includes a set of body feature values ​​at the first moment; or, the first time is a first time range, and at least one set of body feature values ​​includes at least one set of body feature values ​​within the first time range or a weighted sum of multiple sets of body feature values ​​within the first time range.

[0129] In one possible implementation, the second time is a second moment, and at least one body temperature includes the body temperature at the second moment; or, the second time is a second time range, and at least one body temperature includes at least one body temperature within the second time range or a weighted value of multiple body temperatures within the second time range.

[0130] In one possible implementation, the first model uses the XGBoost model, and the second model uses the BERT model.

[0131] In one possible implementation, the acquisition unit 501 is further configured to acquire at least one set of body characteristic values ​​at a fourth time and body temperature at a fifth time, wherein the fourth time is earlier than the fifth time.

[0132] In one possible implementation, the device further includes a training unit 504, which is used to train the first model by taking at least one set of body feature values ​​at a fourth time as input to the first model and taking the body temperature at a fifth time as output to the first model.

[0133] In one possible implementation, the acquisition unit 501 is further configured to acquire at least one body temperature at a sixth time and a body temperature at a seventh time, wherein the sixth time is earlier than the seventh time.

[0134] In one possible implementation, training unit 504 is used to train the second model by taking at least one body temperature at the sixth time as input to the second model and the body temperature at the seventh time as output to the second model.

[0135] In one possible implementation, the device further includes an alarm unit 505, which is used to issue an alarm indication to the user, the alarm indication including a third body temperature and corresponding measures for the third body temperature.

[0136] Based on the same technical concept, embodiments of this application provide a device 600 for predicting body temperature, which may be, for example, a computing device. Figure 6 As shown, a body temperature prediction device 600 includes at least one processor 601 and a memory 602 connected to the at least one processor. In this embodiment, the specific connection medium between the processor 601 and the memory 602 is not limited. Figure 6Taking the connection between the processor 601 and the memory 602 via a bus as an example, the bus can be divided into address bus, data bus, control bus, etc.

[0137] In this embodiment of the application, the memory 602 stores instructions that can be executed by at least one processor 601. By executing the instructions stored in the memory 602, at least one processor 601 can execute the above-described method for predicting body temperature.

[0138] The processor 601 serves as the control center of the body temperature prediction device 600. It can connect to various parts of a computer device via various interfaces and lines, and performs resource settings by running or executing instructions stored in the memory 602 and accessing data stored in the memory 602. Optionally, the processor 601 may include one or more determining units. The processor 601 may integrate an application processor and a modem processor, wherein the application processor primarily handles the operating system, user interface, and applications, while the modem processor primarily handles wireless communication. It is understood that the modem processor may not be integrated into the processor 601. In some embodiments, the processor 601 and the memory 602 may be implemented on the same chip; in some embodiments, they may also be implemented on separate chips.

[0139] Processor 601 can be a general-purpose processor, such as a central processing unit (CPU), digital signal processor, application-specific integrated circuit (ASIC), field-programmable gate array (FPGA), or other programmable logic device, discrete gate or transistor logic device, or discrete hardware component, capable of implementing or executing the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this application can be directly manifested as being executed by a hardware processor, or executed by a combination of hardware and software modules within the processor.

[0140] Memory 602, as a non-volatile computer-readable storage medium, can be used to store non-volatile software programs, non-volatile computer-executable programs, and modules. Memory 602 may include at least one type of storage medium, such as flash memory, hard disk, multimedia card, card-type memory, random access memory (RAM), static random access memory (SRAM), programmable read-only memory (PROM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), magnetic storage, magnetic disk, optical disk, etc. Memory 602 can be any other medium capable of carrying or storing desired program code in the form of instructions or data structures that can be accessed by a computer, but is not limited thereto. In the embodiments of this application, memory 602 can also be a circuit or any other device capable of implementing storage functions for storing program instructions and / or data.

[0141] This application also provides a computer-readable storage medium storing a computer-executable program for causing a computer to perform a method for predicting body temperature as listed in any of the above embodiments.

[0142] This application provides a computer program product, including a computer program executable by a computer device, which, when run on the computer device, causes the computer device to perform a method for predicting body temperature as listed in any of the above methods.

[0143] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0144] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to this application. It should be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0145] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0146] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0147] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A method for predicting body temperature, characterized in that, include: Acquire at least one set of body characteristic values ​​at a first time and at least one body temperature at a second time; wherein the body characteristic values ​​are at least one of systolic blood pressure, diastolic blood pressure, heart rate, and blood oxygen; the first time is a first time range or a first moment, the second time is a second time range or a second moment, and both the first time and the second time are earlier than a third time; there is a temporal overlap between the first time range and the second time range; If the first time is a first time range, then at least one set of body feature values ​​within the first time range is weighted by category and input into the XGBoost model; if the first time is a first moment, then a set of body feature values ​​at the first moment is directly input into the XGBoost model to predict the first body temperature at the third time. If the second time is a second time range, then at least one body temperature within the second time range is weighted by category and input into the improved BERT model; if the second time is a second moment, then a body temperature at the second moment is directly input into the improved BERT model to predict the second body temperature at the third time; wherein, the improved BERT model is a model that has at least one fully connected layer connected in series after the embedding layer of the traditional BERT model and outputs the model through the Softmax activation function; The first body temperature and the second body temperature are weighted and summed to determine the third body temperature at the third time point; wherein, the weighted formula for the first body temperature and the second body temperature is: y=a×x1+b×x2, where x1 is the first body temperature, x2 is the second body temperature, y is the third body temperature, a and b are constant factors, a+b=1, a≥0, b≥0.

2. The method as described in claim 1, characterized in that, The method further includes: Acquire at least one set of body characteristic values ​​at a fourth time point and body temperature at a fifth time point, wherein the fourth time point is earlier than the fifth time point; The XGBoost is trained by using at least one set of body feature values ​​at the fourth time point as input and the body temperature at the fifth time point as output.

3. The method as described in claim 1, characterized in that, The method further includes: Acquire at least one body temperature at a sixth time point and a body temperature at a seventh time point, wherein the sixth time point is earlier than the seventh time point; The improved BERT model is trained by using at least one body temperature at the sixth time point as input and the body temperature at the seventh time point as output.

4. The method according to any one of claims 1 to 3, characterized in that, The method further includes: An alarm instruction is issued to the user, the alarm instruction including the third body temperature and the corresponding response measures.

5. A computing device, characterized in that, include: Memory, used to store program instructions; A processor is configured to invoke program instructions stored in the memory and execute the method as described in any one of claims 1 to 4 according to the obtained program instructions.

Images