WO2023087672A1 - 基于光学信号特征及代谢热特征的血糖预测方法和装置 - Google Patents

基于光学信号特征及代谢热特征的血糖预测方法和装置 Download PDF

Info

Publication number
WO2023087672A1
WO2023087672A1 PCT/CN2022/097242 CN2022097242W WO2023087672A1 WO 2023087672 A1 WO2023087672 A1 WO 2023087672A1 CN 2022097242 W CN2022097242 W CN 2022097242W WO 2023087672 A1 WO2023087672 A1 WO 2023087672A1
Authority
WO
WIPO (PCT)
Prior art keywords
signal
data
generate
metabolic heat
optical signal
Prior art date
Application number
PCT/CN2022/097242
Other languages
English (en)
French (fr)
Inventor
曹君
张碧莹
王思翰
Original Assignee
乐普（北京）医疗器械股份有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 乐普（北京）医疗器械股份有限公司 filed Critical 乐普（北京）医疗器械股份有限公司
Priority to EP22894215.7A priority Critical patent/EP4437954A1/en
Publication of WO2023087672A1 publication Critical patent/WO2023087672A1/zh

Links

Images

Classifications

    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/145—Measuring characteristics of blood in vivo, e.g. gas concentration, pH value; Measuring characteristics of body fluids or tissues, e.g. interstitial fluid, cerebral tissue
    • A61B5/14532—Measuring characteristics of blood in vivo, e.g. gas concentration, pH value; Measuring characteristics of body fluids or tissues, e.g. interstitial fluid, cerebral tissue for measuring glucose, e.g. by tissue impedance measurement
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/01—Measuring temperature of body parts ; Diagnostic temperature sensing, e.g. for malignant or inflamed tissue
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/145—Measuring characteristics of blood in vivo, e.g. gas concentration, pH value; Measuring characteristics of body fluids or tissues, e.g. interstitial fluid, cerebral tissue
    • A61B5/1455—Measuring characteristics of blood in vivo, e.g. gas concentration, pH value; Measuring characteristics of body fluids or tissues, e.g. interstitial fluid, cerebral tissue using optical sensors, e.g. spectral photometrical oximeters
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
    • A61B5/7235—Details of waveform analysis
    • A61B5/7264—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems
    • A61B5/7267—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems involving training the classification device
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
    • A61B5/7271—Specific aspects of physiological measurement analysis
    • A61B5/7275—Determining trends in physiological measurement data; Predicting development of a medical condition based on physiological measurements, e.g. determining a risk factor

Definitions

  • the invention relates to the technical field of data processing, in particular to a blood sugar prediction method and device based on optical signal features and metabolic heat features.
  • Glucose in the blood is called blood sugar (Glu). Maintaining blood sugar at a certain level can maintain the normal work of various organs and tissues in the body. Short-term blood sugar instability will cause various degrees of discomfort in various organs and tissues, and long-term blood sugar instability may lead to various degrees of lesions in various organs and tissues. Therefore, the observed value of blood glucose is a very important health evaluation parameter in the field of health monitoring.
  • the current common blood glucose monitoring method is mainly based on the extraction of human blood in an invasive way, and chemical analysis is performed on the extracted blood to obtain the blood glucose observation value. Because this observation method will cause trauma to the human body, if this observation method is used to monitor the blood sugar of the human body for a long time, it will undoubtedly bring a lot of inconvenience to the user, such as wound pain, wound infection and other adverse experiences.
  • MHC Metabolic Heat Conformation
  • the purpose of the present invention is to address the defects of the prior art, to provide a blood sugar prediction method, device, electronic equipment and computer-readable storage medium based on optical signal characteristics and metabolic heat characteristics.
  • long-term blood sugar observation can provide A non-invasive observation mechanism that reduces the inconvenience caused to users, reduces the difficulty of observation, and improves user experience.
  • the first aspect of the embodiment of the present invention provides a blood sugar prediction method based on optical signal characteristics and metabolic heat characteristics, the method comprising:
  • the type of human optical signal and the three human optical signal bands continuously collect the human optical signal to generate corresponding first optical signal, second optical signal and third optical signal; and according to the optical signal
  • the acquisition time is long, and the ambient light signal is continuously collected to generate the fourth optical signal;
  • the preset metabolic heat signal acquisition time continuously collect the human body contact heat signal to generate the first metabolic heat signal, and continuously collect the radiant heat signal of the proximal end of the human body to generate the second metabolic heat signal, and continuously collect the temperature change information of the proximal end of the human body Constitute the third metabolic heat signal, and continuously collect the humidity change information near the human body to form the fourth metabolic heat signal, and continuously collect the calibration output information of the proximal radiation sensor used to collect the radiation heat signal to form the fifth metabolic heat signal , and continuously collect the temperature change information at the far end of the human body to form the sixth metabolic heat signal, and continuously collect the humidity change information at the far end of the human body to form the seventh metabolic heat signal;
  • For the first eigenvector perform a floating blood glucose value prediction process based on a blood glucose floating prediction model to generate first floating blood glucose data; and generate First predicted blood glucose data.
  • the optical signal type includes a photoplethysmography signal type
  • the three optical signal bands include 650nm infrared band, 940nm near infrared band and 1050nm near infrared band.
  • the first, second and third optical signals are subjected to human body optical signal feature extraction and normalization processing to generate corresponding first, second and third optical feature data sets, specifically including:
  • a corresponding first optical feature data set, a second optical feature data set or a third optical feature data set is formed from the obtained AC feature data and the DC feature data.
  • performing ambient light signal feature extraction and normalization processing on the fourth optical signal to generate corresponding fourth optical feature data specifically includes:
  • the first, second, third, fourth, fifth, sixth and seventh metabolic heat signals are subjected to metabolic heat feature extraction and normalization processing to generate corresponding first and seventh metabolic heat signals.
  • Second, third, fourth, fifth, sixth and seventh metabolic heat characteristic data specifically including:
  • the blood sugar fluctuation prediction model is implemented based on a network structure of a multi-layer neural network;
  • the blood sugar fluctuation prediction model includes an input layer, a hidden layer and an output layer;
  • the input layer includes a first number M of input nodes;
  • the hidden layer includes a second number N of hidden layer input nodes and a third number S of hidden layer output nodes;
  • the output layer includes an output node; each of the hidden layer input nodes of the hidden layer is connected to the input layer All the input nodes of the hidden layer are respectively connected to form a corresponding first fully connected network;
  • each of the hidden layer output nodes of the hidden layer is respectively connected to all the hidden layer input nodes to form a corresponding second fully connected network;
  • the output node of the output layer is connected to all the hidden layer input nodes;
  • the first number M is associated with the total number of feature classifications of the first feature vector, and the second number N>the third number S>0;
  • the blood sugar floating prediction model When the blood sugar floating prediction model performs the floating blood sugar value prediction processing, it calls the input layer to extract each feature data of the first feature vector in order and input them to the corresponding input node to generate the corresponding input node data;
  • the second aspect of the embodiment of the present invention provides a device for implementing the method described in the first aspect above, including: an acquisition module, a first data collection module, a second data collection module, a first feature data processing module, a second feature Data processing module, feature data fusion processing module and blood sugar prediction processing module;
  • the acquiring module is used to acquire the calibrated blood glucose value to generate the first calibrated blood glucose data
  • the first data acquisition module is used to continuously collect the human body optical signal to generate corresponding first optical signal, second optical signal and third an optical signal; and according to the optical signal collection duration, continuously collect the ambient light signal to generate a fourth optical signal;
  • the second data acquisition module is used to continuously collect the human body contact heat signal to generate the first metabolic heat signal according to the preset metabolic heat signal acquisition time, and continuously collect the radiant heat signal near the human body to generate the second metabolic heat signal, and Continuously collect the temperature change information of the proximal end of the human body to form the third metabolic heat signal, and continuously collect the humidity change information of the proximal end of the human body to form the fourth metabolic heat signal, and output information for the calibration of the proximal radiation sensor used to collect the radiant heat signal Continuously collect the fifth metabolic heat signal, and continuously collect the temperature change information of the remote body to form the sixth metabolic heat signal, and continuously collect the humidity change information of the remote human body to form the seventh metabolic heat signal;
  • the first characteristic data processing module is used to perform human body optical signal feature extraction and normalization processing on the first, second and third optical signals, and generate corresponding first, second and third optical characteristic data sets ; and performing ambient light signal feature extraction and normalization processing on the fourth optical signal to generate corresponding fourth optical feature data;
  • the second feature data processing module is used to perform metabolic heat feature extraction and normalization processing on the first, second, third, fourth, fifth, sixth and seventh metabolic heat signals to generate corresponding The first, second, third, fourth, fifth, sixth and seventh metabolic heat characteristic data;
  • the feature data fusion processing module is used to perform feature fusion processing on all the obtained feature data to generate a first feature vector
  • the blood glucose prediction processing module is used to perform a floating blood glucose value prediction process on the first feature vector based on a blood glucose floating prediction model to generate first floating blood glucose data; and according to the first calibrated blood glucose data and the first floating The sum of the additions of the blood glucose data generates the first predicted blood glucose data.
  • the third aspect of the embodiment of the present invention provides an electronic device, including: a memory, a processor, and a transceiver;
  • the processor is configured to be coupled with the memory, read and execute instructions in the memory, so as to implement the method steps described in the first aspect above;
  • the transceiver is coupled to the processor, and the processor controls the transceiver to send and receive messages.
  • the fourth aspect of the embodiments of the present invention provides a computer-readable storage medium, the computer-readable storage medium stores computer instructions, and when the computer instructions are executed by a computer, the computer executes the above-mentioned first aspect. method directive.
  • Embodiments of the present invention provide a blood sugar prediction method, device, electronic equipment, and computer-readable storage medium based on optical signal characteristics and metabolic heat characteristics, which collect photoplethysmographic signals of human body and environmental optical signals, and Collect contact heat, radiant heat, temperature, humidity and radiation sensor calibration signals related to human metabolic heat, collect environmental temperature and humidity signals related to human metabolic heat; and collect optical signals and metabolic heat related signals.
  • Feature extraction and normalization processing form the feature vector; and use the blood sugar fluctuation prediction model that reflects the relationship between optics, metabolic heat and blood sugar change to predict the feature vector to obtain the corresponding blood sugar fluctuation data, so that based on the pre-acquired blood sugar calibration value +Blood sugar floating data can get blood sugar prediction data.
  • a non-invasive observation mechanism is provided for long-term blood glucose observation, which reduces, for example, the inconvenience caused to users, reduces, for example, the difficulty of observation, and improves user experience.
  • FIG. 1 is a schematic diagram of a blood sugar prediction method based on optical signal characteristics and metabolic heat characteristics provided by Embodiment 1 of the present invention
  • FIG. 2 is a schematic diagram of the neural network structure of the blood sugar fluctuation prediction model provided by Embodiment 1 of the present invention.
  • Fig. 3 is a module structure diagram of a blood sugar prediction device based on optical signal characteristics and metabolic heat characteristics provided by Embodiment 2 of the present invention
  • FIG. 4 is a schematic structural diagram of an electronic device provided by Embodiment 3 of the present invention.
  • Embodiment 1 of the present invention provides a blood sugar prediction method based on optical signal features and metabolic heat features, as shown in Figure 1, a schematic diagram of a blood sugar prediction method based on optical signal features and metabolic heat features provided in Embodiment 1 of the present invention , this method mainly includes the following steps:
  • Step 1 obtain the calibrated blood glucose value to generate the first calibrated blood glucose data.
  • the method of the embodiment of the present invention will perform a blood sugar calibration process in advance to generate a calibrated blood sugar value, specifically including:
  • Step A1 Perform one or more routine blood sugar tests on the current user through conventional blood sugar testing methods to obtain one or more measured blood sugar data; record the time information of the current blood sugar test and the basic physiological information of the user, collect and record the current blood sugar test
  • the optical correlation signal and metabolic heat related signal at the time; and the corresponding calibration reference data set is composed of each measurement blood glucose data, time information, basic physiological information of the user, optical correlation signal and metabolic heat related signal;
  • the user's basic physiological information includes age, gender, height, weight, body mass index (Body Mass Index, BMI), etc.
  • optical related signals include three optical signal bands (650nm infrared band, 940nm near-infrared band and 1050nm near-infrared band ) human body photoplethysmography signal and environmental optical signal
  • metabolic heat related signals include human body contact heat signal, human body proximal radiation heat signal, human body proximal temperature change signal, human body proximal humidity change signal, used for The calibration output signal of the near-end radiation sensor that collects radiant heat signals, the temperature change signal at the far end of the human body, and the humidity change signal at the far end of the human body;
  • Step A2 if only one routine blood glucose test is performed on the user, the measured blood glucose data obtained from this test is used as the calibration blood glucose value and stored, and the data other than the measured blood glucose data in the corresponding calibration reference data group are used as the calibration condition data Group and store;
  • Step A3 if the user has only performed multiple routine blood glucose tests, then calculate the mean value of the multiple measured blood glucose data obtained to generate corresponding average measured blood glucose data; and use the difference with the average measured blood glucose data as the minimum value
  • the blood glucose data is stored as a calibrated blood glucose value, and other data in the calibration reference data set corresponding to the calibrated blood glucose value except the measured blood glucose data are stored as a calibration condition data set.
  • the calibrated blood glucose value and calibrated condition data set is created, it will be saved as the reference data of the current user.
  • the calibration condition data group of the benchmark data will be used as a prediction model reference to predict the blood glucose fluctuation of the current user to obtain the corresponding floating blood glucose data, and then the The calibrated blood glucose value of the reference data + the floating blood glucose data can obtain the blood glucose prediction data of the current user. That is to say, after the baseline data of the current user is created, each subsequent blood glucose observation process of the user is a non-invasive observation process.
  • Step 2 according to the preset optical signal acquisition time, human body optical signal type and three human body optical signal bands, continuously collect human body optical signals to generate corresponding first optical signal, second optical signal and third optical signal; and press optical The signal acquisition time is long, and the ambient light signal is continuously collected to generate the fourth optical signal;
  • the optical signal type includes a photoplethysmography signal type; the three optical signal wavebands include a 650nm infrared waveband, a 940nm near-infrared waveband and a 1050nm near-infrared waveband.
  • the optical signal acquisition time is usually set to 30 seconds or 1 minute, and it can also be set separately based on the acquisition quality of the signal acquisition equipment;
  • the light sources do not cause mutual interference to each other.
  • the preset sampling frequency normally 125Hz
  • three light-emitting diodes (light- The emitting diode (LED) light is irradiated sequentially close to the fingertips of the human body, and three corresponding photosensitive diodes are used to collect the transmitted light signals of the three bands in sequence on the finger back at the position vertically corresponding to each LED light.
  • the first optical signal specifically for transmitted light in the 650nm infrared band, the second optical signal specifically for transmitted light in the 940nm infrared band, and the third optical signal specifically for transmitted light in the 1050nm infrared band are obtained; when collecting ambient light signals, To ensure that there is no interference from the above-mentioned LED light source, the natural light signal of the ambient background light will be continuously collected according to the preset sampling frequency (normally 125Hz) after the above-mentioned first, second and third optical signal collection is completed, so as to obtain
  • the fourth optical signal: the above-mentioned first, second, third and fourth optical signals have the same collection frequency and the same signal duration.
  • Step 3 According to the preset metabolic heat signal acquisition time, continuously collect the human body contact heat signal to generate the first metabolic heat signal, and continuously collect the radiant heat signal of the proximal end of the human body to generate the second metabolic heat signal, and continuously collect the proximal human body heat signal
  • the temperature change information constitutes the third metabolic heat signal
  • the continuous collection of the humidity change information near the human body constitutes the fourth metabolic heat signal
  • the continuous collection of the calibration output information of the proximal radiation sensor used to collect the radiation heat signal constitutes the fifth Metabolic heat signal, and continuously collect the temperature change information of the far end of the human body to form the sixth metabolic heat signal, and continuously collect the humidity change information of the far end of the human body to form the seventh metabolic heat signal.
  • the acquisition of the metabolic heat signal can be synchronized with the acquisition process of any one of the first, second and third optical signals.
  • the corresponding metabolic heat The signal acquisition duration is consistent with the optical signal acquisition duration; it can also be synchronized with the overall acquisition process of the first, second and third optical signals, and at this time the corresponding metabolic heat signal acquisition duration is greater than or equal to three times the optical signal acquisition duration; if
  • the sampling frequency of all metabolic heat signals can be set to be consistent with the optical signal sampling frequency (normally 125Hz);
  • the purpose of collecting the calibration output information of the near-end radiation sensor is to provide a calibration signal for the collected near-end radiant heat signal
  • Step 4 performing human body optical signal feature extraction and normalization processing on the first, second and third optical signals, and generating corresponding first, second and third optical feature data sets;
  • the first, second and third optical characteristic data sets are all composed of AC characteristic data and DC characteristic data;
  • the feature extraction of the human body optical signal is actually to extract the AC and DC features of the three-way human body photoplethysmography signal
  • step 41 extracting an optical signal segment of a first specified duration from an intermediate signal period of the first optical signal, the second optical signal or the third optical signal, and generating a corresponding first segment signal;
  • the first specified duration is usually 20 seconds
  • Step 42 Perform band-pass filtering on the first segment signal according to the preset band-pass filtering frequency band to generate a corresponding first filtered signal; and perform signal peak recognition processing on the first filtered signal to obtain a plurality of corresponding first filter signals.
  • signal peak data and performing mean value calculation on a plurality of first signal peak data to generate corresponding first average peak data; and performing signal up-down flip processing on the first filtered signal to generate a corresponding first flip signal; and first Inverting the signal to perform signal peak identification processing to obtain a plurality of corresponding second signal peak data; and performing mean value calculation on the plurality of second signal peak data, and inverting the calculation results to generate corresponding first average valley data; and according to The difference between the first average peak value data and the first average valley value data is subtracted to generate corresponding first feature data; and the first feature data is normalized to generate AC feature data;
  • the frequency band of the bandpass filter is usually 0.5-10Hz;
  • the method of the embodiment of the present invention regards the signal lower than 0.5Hz as a DC characteristic signal , the signal higher than 10Hz is regarded as an interference or noise signal, and the AC characteristic signal can be obtained by filtering out the DC characteristic signal and the interference or noise signal;
  • the characteristic of the AC characteristic signal is the amplitude difference of the waveform, that is, the peak-valley difference,
  • the AC characteristic data calculated here is actually the average amplitude data of the AC characteristic signal;
  • the embodiment of the present invention when performing normalization processing on the first feature data to generate communication feature data, provides a preset normalization mode on the basis of a large amount of existing training data for training subsequent prediction models.
  • At least two normalization processing procedures including:
  • Step B1 when the normalization mode is the first mode, normalize the first feature data using the maximum normalization function
  • the most valued normalization function is X norm is the normalized data, X is the first feature data input, X min and X max are respectively the minimum and maximum values of the training data consistent with the first feature data type in the data set composed of the above-mentioned large amount of training data;
  • Step B2 when the normalization mode is the second mode, normalize the first feature data using a mean-variance normalization function
  • the mean variance normalization function is X norm is the normalized data, X is the input first feature data, ⁇ and ⁇ are respectively the average value and standard deviation of the training data consistent with the first feature data type in the data set composed of the above-mentioned large amount of training data;
  • Step 43 Perform low-pass filtering on the first segment signal according to the preset low-pass filtering frequency threshold to generate a corresponding second filtered signal; and perform mean value calculation on the second filtered signal to generate corresponding second feature data; and performing normalization processing on the second characteristic data to generate DC characteristic data;
  • the low-pass filter frequency threshold is generally 0.5Hz
  • the DC characteristic signal is obtained by retaining the signal lower than 0.5Hz
  • the AC characteristic signal The characteristic is that the waveform is basically flat and has no obvious peak-to-valley changes.
  • the DC characteristic data calculated here is actually the average amplitude data of the DC characteristic signal;
  • the normalization processing process is similar to the normalization processing process on the first characteristic data in step 41, and no further details are given here;
  • Step 44 compose the corresponding first optical characteristic data set, second optical characteristic data set or third optical characteristic data set from the obtained AC characteristic data and DC characteristic data.
  • Step 5 performing ambient light signal feature extraction and normalization processing on the fourth optical signal to generate corresponding fourth optical feature data
  • the third feature data can reflect the average light intensity of the ambient background light; when the third feature data is normalized to generate the fourth optical feature data, its normalization process is the same as that of the first feature data in step 41.
  • the normalization process of is similar and will not be described further here.
  • Step 6 perform metabolic heat feature extraction and normalization processing on the first, second, third, fourth, fifth, sixth and seventh metabolic heat signals, and generate corresponding first, second, third , fourth, fifth, sixth and seventh metabolic heat characteristic data;
  • step 61 extracting a metabolic heat signal segment of a second specified duration from the last signal period of the first, second, third, fourth, fifth, sixth or seventh metabolic heat signal, and generating a corresponding third fragment signal;
  • the fragment interception method of the metabolic heat signal is the last specified
  • Step 62 performing mean value calculation on the third segment signal to generate corresponding fourth characteristic data
  • the fourth characteristic data is the average value of human body contact heat; if the third segment signal is the segment data of the second metabolic heat signal, then the fourth characteristic data It is the average value of the proximal radiant heat of the human body; if the third segment signal is the segment data of the third metabolic heat signal, then the fourth characteristic data is the average temperature of the proximal end of the human body; if the third segment signal is the segment data of the fourth metabolic heat signal Fragment data, then the fourth feature data is the average value of the proximal humidity of the human body; if the third segment signal is the segment data of the fifth metabolic heat signal, then the fourth feature data is the calibration of the proximal radiation sensor used to collect the radiant heat signal Average value; if the third fragment signal is the fragment data of the sixth metabolic heat signal, then the fourth characteristic data is the average temperature of the remote body temperature, which is actually the average ambient temperature; if the third fragment signal is the fragment of the seventh metabolic heat signal data, then
  • Step 63 Perform normalization processing on the fourth characteristic data to generate corresponding first metabolic heat characteristic data, second metabolic heat characteristic data, third metabolic heat characteristic data, fourth metabolic heat characteristic data, and fifth metabolic heat characteristic data data, sixth metabolic heat characteristic data or seventh metabolic heat characteristic data.
  • Step 7 performing feature fusion processing on all the obtained feature data to generate a first feature vector.
  • the first, second and third optical characteristic data sets, the fourth optical characteristic data and the first, second, third, fourth, fifth, sixth and the 14 feature data of the seventh metabolic heat feature data are sorted, and the corresponding first feature vector is constructed according to the sorting result.
  • the predicted time and personal basic physiological information can also be collected in real time, and the relevant collected data can be included in the second In a eigenvector; specific cases include taking collection time as one of the blood sugar change factors, taking personal basic physiological information changes as one of the blood sugar change factors, and so on.
  • Step 8 For the first eigenvector, perform a floating blood sugar value prediction process based on a blood sugar floating prediction model to generate first floating blood sugar data;
  • the blood sugar fluctuation prediction model is realized based on the network structure of a multi-layer neural network;
  • the blood sugar fluctuation prediction model includes an input layer, a hidden layer and an output layer;
  • the input layer includes a first number M of input nodes;
  • the hidden layer includes a second number N of Hidden layer input node and the hidden layer output node of the third quantity S;
  • Output layer comprises an output node;
  • Each hidden layer input node of hidden layer is connected with all input nodes of input layer respectively, forms the corresponding first fully connected network;
  • Hidden Each hidden layer output node of the layer is connected with all hidden layer input nodes respectively to form a corresponding second fully connected network;
  • the output node of the output layer is connected with all hidden layer input nodes;
  • the feature classification of the first quantity M and the first feature vector The total number is associated, the second number N> the third number S>0.
  • Figure 2 is a schematic diagram of the neural network structure of the blood sugar fluctuation prediction model provided in Embodiment 1 of the present invention; the model is constructed based on Beer-Lambert law and metabolic heat conformation theory The network operation model of the first and second fully connected networks and the parameter matrix structure required for operation; the prediction model will set the number of input nodes of the input layer with the data type of the first feature vector at each prediction, and Create the corresponding calibration feature vector based on the corresponding relationship between the calibration condition data group of the current user reference data and the data type of the first feature vector, and then activate the parameter matrix and output layer preset of the first and second fully connected networks according to the calibration feature vector Fine-tune the parameters of the function; then predict the blood sugar fluctuation value of the current user at the current moment according to the first feature vector to obtain the corresponding first floating blood sugar data.
  • Figure 2 is a schematic diagram of the neural network structure of the blood sugar fluctuation prediction model provided in Embodiment 1 of the present invention; the model is constructed based on Beer-Lambert law and metabolic
  • the data category of the first feature vector must include 14 feature data (the first, second and third optical feature data sets, the fourth optical feature data and the first, second, The third, fourth, fifth, sixth and seventh metabolic heat characteristic data), can also optionally include characteristic data such as time, age, sex, height, weight, BMI of this prediction;
  • characteristic data such as time, age, sex, height, weight, BMI of this prediction
  • the method of extracting human optical signal features and normalizing processing is to intersect and process the human body photoplethysmography signals of the three optical signal bands (650nm infrared band, 940nm near infrared band and 1050nm near infrared band) of the calibration condition data group.
  • DC feature data extraction and normalization processing to obtain the calibration feature data of 6 human optical signals; perform ambient light signal feature extraction and normalization processing on the fourth optical signal in a manner similar to step 5 to perform ambient optical signal calibration of the calibration condition data group.
  • the signal is subjected to feature data extraction and normalization processing to obtain 1 ambient optical calibration feature data; perform metabolism on the first, second, third, fourth, fifth, sixth and seventh metabolic heat signals in a similar manner to step 6
  • the method of thermal feature extraction and normalization processing is used to collect the human body contact heat signal, the near-end radiation heat signal of the human body, the temperature change signal of the near-end human body, and the humidity change signal of the near-end human body in the calibration condition data group.
  • the calibration output signal of the radiation sensor at the near end of the signal, the temperature change signal at the far end of the human body, and the humidity change signal at the far end of the human body are subjected to feature data extraction and normalization processing to obtain 7 metabolic heat signal calibration feature data; if the first feature
  • the data types of the vector also include characteristic data types such as time, age, gender, height, weight or BMI, and are further generated according to the time information of the calibration condition data group and the basic physiological information of the user (age, sex, height, weight, BMI) Corresponding time, age, gender, height, weight or BMI calibration feature data, and the aforementioned 6 human optical signal calibration feature data, 1 environmental optical calibration feature data and 7 metabolic heat signal calibration feature data according to the first
  • a data sorting structure with the same feature vectors is used for fusion processing of calibration feature data, thereby obtaining calibration feature vectors.
  • the blood sugar fluctuation prediction model in the embodiment of the present invention is pre-formed through the training of a large number of different individual data collected, so the prediction accuracy of the model is related to the characteristics of the population to which the training data belongs; Further related, it is necessary to fine-tune the parameter matrix of the first and second fully connected network and the parameters of the preset activation function of the output layer according to the calibration feature vector to achieve the purpose of personalized calibration; after completing the personalized calibration, the model can be The prediction results are fine-tuned according to the personalized benchmark characteristics of the current individual, so that the prediction results can be closer to the real data of the current individual, and the personalized prediction accuracy of the model is improved.
  • the blood sugar floating prediction model performs floating blood sugar value prediction processing, it specifically includes:
  • Step C1 call the input layer to extract each feature data of the first feature vector in order and input it to the corresponding input node to generate the corresponding input node data;
  • Step C2 call the hidden layer based on the parameter matrix of each first fully connected network, perform the corresponding first fully connected network operation on all the input node data connected to it, and input the operation result to the corresponding hidden layer input node as hidden layer input node data;
  • Step C3 call the hidden layer based on the parameter matrix of each second fully connected network, perform the corresponding second fully connected network operation on the data of all hidden layer input nodes connected to it, and input the operation result to the corresponding hidden layer output node Output node data as a hidden layer;
  • Step C4 call the output layer based on the preset activation function, perform the corresponding activation function operation on all hidden layer output node data, and output the operation result as the first floating blood sugar data.
  • Step 9 generating first predicted blood glucose data according to the sum of the first calibrated blood glucose data and the first floating blood glucose data.
  • the blood glucose prediction of the current user at the current moment can be obtained
  • the value is the first predicted blood glucose data.
  • Fig. 3 is a module structure diagram of a blood sugar prediction device based on optical signal characteristics and metabolic heat characteristics provided by Embodiment 2 of the present invention.
  • the device can be a terminal device or a server implementing the method of the embodiment of the present invention, or it can be the same as the above-mentioned
  • the device includes: an acquisition module 201, a first data collection module 202, a second data collection module 203, a first feature data processing module 204, a second feature data processing module 205, and a feature data fusion processing module 206 and a blood glucose prediction processing module 207.
  • the acquiring module 201 is used to acquire a calibrated blood glucose value to generate first calibrated blood glucose data.
  • the first data acquisition module 202 is used to continuously collect the human body optical signal to generate corresponding first optical signal, second optical signal and third optical signal according to the preset optical signal acquisition duration, human optical signal type and three human optical signal bands. signal; and according to the optical signal acquisition time, continuously collect the ambient light signal to generate the fourth optical signal.
  • the second data acquisition module 203 is used to continuously collect the human body contact heat signal to generate the first metabolic heat signal according to the preset metabolic heat signal collection time, and continuously collect the radiant heat signal of the proximal end of the human body to generate the second metabolic heat signal, and continuously
  • the temperature change information of the proximal end of the human body is collected to form the third metabolic heat signal
  • the humidity change information of the proximal end of the human body is continuously collected to form the fourth metabolic heat signal
  • the calibration output information of the proximal radiation sensor used to collect the radiant heat signal is calculated.
  • the fifth metabolic heat signal is continuously collected, and the sixth metabolic heat signal is formed by continuously collecting the temperature change information of the remote end of the human body, and the seventh metabolic heat signal is formed by continuously collecting the humidity change information of the remote end of the human body.
  • the first feature data processing module 204 is used to perform human body light signal feature extraction and normalization processing on the first, second and third optical signals, and generate corresponding first, second and third optical feature data sets; and
  • the fourth optical signal is subjected to ambient light signal feature extraction and normalization processing to generate corresponding fourth optical feature data.
  • the second feature data processing module 205 is used to perform metabolic heat feature extraction and normalization processing on the first, second, third, fourth, fifth, sixth and seventh metabolic heat signals to generate corresponding first , second, third, fourth, fifth, sixth and seventh metabolic heat characteristic data.
  • the feature data fusion processing module 206 is used to perform feature fusion processing on all the feature data obtained to generate the first feature vector.
  • the blood sugar prediction processing module 207 is used to perform floating blood sugar value prediction processing on the first eigenvector based on the blood sugar floating prediction model to generate first floating blood sugar data; and according to the sum of the first calibration blood sugar data and the first floating blood sugar data, First predicted blood glucose data is generated.
  • a blood sugar prediction device based on optical signal features and metabolic heat features provided by the embodiments of the present invention can execute the method steps in the above method embodiments, and its implementation principles and technical effects are similar, and will not be repeated here.
  • each module of the above device is only a division of logical functions, and may be fully or partially integrated into one physical entity or physically separated during actual implementation.
  • these modules can all be implemented in the form of calling software through processing elements; they can also be implemented in the form of hardware; some modules can also be implemented in the form of calling software through processing elements, and some modules can be implemented in the form of hardware.
  • the acquisition module can be a separate processing element, or it can be integrated into a chip of the above-mentioned device.
  • it can also be stored in the memory of the above-mentioned device in the form of program code, and a certain processing element of the above-mentioned device can Call and execute the functions of the modules identified above.
  • each step of the above method or each module above can be completed by an integrated logic circuit of hardware in the processor element or an instruction in the form of software.
  • the above modules may be one or more integrated circuits configured to implement the above method, for example: one or more specific integrated circuits (Application Specific Integrated Circuit, ASIC), or, one or more digital signal processors ( Digital Signal Processor, DSP), or, one or more Field Programmable Gate Arrays (Field Programmable Gate Array, FPGA), etc.
  • ASIC Application Specific Integrated Circuit
  • DSP Digital Signal Processor
  • FPGA Field Programmable Gate Array
  • the processing element may be a general-purpose processor, such as a central processing unit (Central Processing Unit, CPU) or other processors that can call program codes.
  • these modules can be integrated together and implemented in the form of a System-on-a-chip (SOC).
  • SOC System-on-a-chip
  • all or part of them may be implemented by software, hardware, firmware or any combination thereof.
  • software When implemented using software, it may be implemented in whole or in part in the form of a computer program product.
  • the computer program product includes one or more computer instructions.
  • the above-mentioned computers may be general-purpose computers, special-purpose computers, computer networks, or other programmable devices.
  • the above-mentioned computer instructions may be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium.
  • the above-mentioned computer-readable storage medium may be any available medium that can be accessed by a computer, or a data storage device such as a server or a data center integrated with one or more available media.
  • the above-mentioned usable medium may be a magnetic medium (for example, a floppy disk, a hard disk, or a magnetic tape), an optical medium (for example, DVD), or a semiconductor medium (for example, a solid state disk (solid state disk, SSD)) and the like.
  • FIG. 4 is a schematic structural diagram of an electronic device provided by Embodiment 3 of the present invention.
  • the electronic device may be the aforementioned terminal device or server, or may be a terminal device or server connected to the aforementioned terminal device or server to implement the method of the embodiment of the present invention.
  • the electronic device may include: a processor 301 (such as a CPU), a memory 302 , and a transceiver 303 ;
  • Various instructions may be stored in the memory 302 for completing various processing functions and realizing the methods and processing procedures provided in the above-mentioned embodiments of the present invention.
  • the electronic device involved in this embodiment of the present invention further includes: a power supply 304 , a system bus 305 and a communication port 306 .
  • the system bus 305 is used to realize the communication connection among the components.
  • the above-mentioned communication port 306 is used for connection and communication between the electronic device and other peripheral devices.
  • the system bus mentioned in FIG. 4 may be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (Extended Industry Standard Architecture, EISA) bus or the like.
  • PCI Peripheral Component Interconnect
  • EISA Extended Industry Standard Architecture
  • the system bus can be divided into address bus, data bus, control bus and so on. For ease of representation, only one thick line is used in the figure, but it does not mean that there is only one bus or one type of bus.
  • the communication interface is used to realize the communication between the database access device and other devices (such as client, read-write library and read-only library).
  • the memory may include random access memory (Random Access Memory, RAM), and may also include non-volatile memory (Non-Volatile Memory), such as at least one disk memory.
  • processor can be general-purpose processor, comprises central processing unit CPU, network processor (Network Processor, NP) etc.; Can also be digital signal processor DSP, application-specific integrated circuit ASIC, field programmable gate array FPGA or other available Program logic devices, discrete gate or transistor logic devices, discrete hardware components.
  • CPU central processing unit
  • NP Network Processor
  • DSP digital signal processor
  • ASIC application-specific integrated circuit
  • FPGA field programmable gate array
  • embodiments of the present invention also provide a computer-readable storage medium, and instructions are stored in the storage medium, and when the storage medium is run on a computer, the computer executes the methods and processing procedures provided in the above-mentioned embodiments.
  • the embodiment of the present invention also provides a chip for running instructions, and the chip is used for executing the method and the processing procedure provided in the foregoing embodiments.
  • Embodiments of the present invention provide a blood sugar prediction method, device, electronic equipment, and computer-readable storage medium based on optical signal characteristics and metabolic heat characteristics, which collect photoplethysmographic signals of human body and environmental optical signals, and Collect contact heat, radiant heat, temperature, humidity and radiation sensor calibration signals related to human metabolic heat, collect environmental temperature and humidity signals related to human metabolic heat; and collect optical signals and metabolic heat related signals.
  • Feature extraction and normalization processing form the feature vector; and use the blood sugar fluctuation prediction model that reflects the relationship between optics, metabolic heat and blood sugar change to predict the feature vector to obtain the corresponding blood sugar fluctuation data, so that based on the pre-acquired blood sugar calibration value +Blood sugar floating data can get blood sugar prediction data.
  • a non-invasive observation mechanism is provided for long-term blood glucose observation, which reduces, for example, the inconvenience caused to users, reduces, for example, the difficulty of observation, and improves user experience.
  • RAM random access memory
  • ROM read-only memory
  • EEPROM electrically programmable ROM
  • EEPROM electrically erasable programmable ROM
  • registers hard disk, removable disk, CD-ROM, or any other Any other known storage medium.

Landscapes

  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • General Health & Medical Sciences (AREA)
  • Pathology (AREA)
  • Biomedical Technology (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Medical Informatics (AREA)
  • Molecular Biology (AREA)
  • Surgery (AREA)
  • Animal Behavior & Ethology (AREA)
  • Biophysics (AREA)
  • Public Health (AREA)
  • Veterinary Medicine (AREA)
  • Artificial Intelligence (AREA)
  • Optics & Photonics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Physiology (AREA)
  • Psychiatry (AREA)
  • Signal Processing (AREA)
  • Spectroscopy & Molecular Physics (AREA)
  • Emergency Medicine (AREA)
  • Evolutionary Computation (AREA)
  • Fuzzy Systems (AREA)
  • Mathematical Physics (AREA)
  • Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)

Abstract

一种基于光学信号特征及代谢热特征的血糖预测方法和装置，方法包括：获取第一标定血糖数据(1)；采集生成第一、二、三、四光学信号(2)；采集生成第一、二、三、四、五、六、七代谢热信号(3)；进行人体光信号特征提取及归一化处理生成第一、二、三光学特征数据组(4)；进行环境光信号特征提取及归一化处理生成第四光学特征数据(5)；进行代谢热特征提取及归一化处理生成第一、二、三、四、五、六、七代谢热特征数据(6)；进行特征融合处理生成第一特征向量(7)；基于血糖浮动预测模型进行浮动血糖值预测生成第一浮动血糖数据(8)；根据第一标定血糖数据和第一浮动血糖数据相加的和生成第一预测血糖数据(9)。基于光学信号特征及代谢热特征的血糖预测方法可提供一种无创血糖观测机制。

Description

基于光学信号特征及代谢热特征的血糖预测方法和装置
本申请要求于2021年11月22日提交中国专利局、申请号为202111386652.X、发明名称为“基于光学信号特征及代谢热特征的血糖预测方法和装置”的中国专利申请的优先权。
技术领域
本发明涉及数据处理技术领域，特别涉及一种基于光学信号特征及代谢热特征的血糖预测方法和装置。
背景技术
血中的葡萄糖称为血糖(Glu)。血糖保持在一定水平才能维持体内各器官和组织的正常工作，短期血糖不稳定会导致各器官和组织出现不同程度的不适，长期血糖不稳定则可能会导致各器官和组织出现不同程度的病变。因此，血糖观测值是健康监护领域一项非常重要的健康评判参数。当前常见的血糖观测手段主要是基于有创方式对人体血液进行提取，并根据提取血液进行化学分析从而得到血糖观测值。因为这种观测手段是会给人体造成创伤的，若使用这种观测手段对人体进行长期血糖观测，无疑会给用户带来很多不便，例如创口疼痛、创口感染等不良体验。
基于比尔-朗伯定律我们可知溶液对透射光的吸收度与溶质浓度有关，进而可知血液中葡萄糖浓度越高则透过人体组织的光强就越小，那么通过采集反映人体血液透射光强变化的光体积描记图法(Photoplethysmography,PPG)信号就能对血糖变化进行观测。
基于代谢热构象(Metabolic Heat Conformation,MHC)理论我们可知人 体不同时段的能量节律与人体代谢之后释放的能量有一定相关性，而葡萄糖氧化与产生的能量可以热能形式从人体散发到环境中，因此人体代谢热与葡萄糖水平和供氧量相关。也即是说在供氧保证的情况下，基于代谢热构象理论可以人体代谢热为参考对血糖变化进行观测。
发明内容
本发明的目的，就是针对现有技术的缺陷，提供一种基于光学信号特征及代谢热特征的血糖预测方法、装置、电子设备及计算机可读存储介质，通过本发明，可为长期血糖观测提供一种无创观测机制，减少对用户造成的不便，降低观测难度，提高用户体验。
为实现上述目的，本发明实施例第一方面提供了一种基于光学信号特征及代谢热特征的血糖预测方法，所述方法包括：
获取标定血糖值生成第一标定血糖数据；
按预设的光学信号采集时长、人体光学信号类型和三个人体光学信号波段，持续采集人体光信号生成对应的第一光学信号、第二光学信号和第三光学信号；并按所述光学信号采集时长，持续采集环境光信号生成第四光学信号；
按预设的代谢热信号采集时长，持续采集人体接触热信号生成第一代谢热信号，并持续采集人体近端的辐射热信号生成第二代谢热信号，并持续采集人体近端的温度变化信息构成第三代谢热信号，并持续采集人体近端的湿度变化信息构成第四代谢热信号，并对用于采集辐射热信号的近端辐射传感器的校准输出信息进行持续采集构成第五代谢热信号，并持续采集人体远端的温度变化信息构成第六代谢热信号，并持续采集人体远端的湿度变化信息构成第七代谢热信号；
对所述第一、第二和第三光学信号进行人体光信号特征提取及归一化处理，生成对应的第一、第二和第三光学特征数据组；
对所述第四光学信号进行环境光信号特征提取及归一化处理，生成对应 的第四光学特征数据；
对所述第一、第二、第三、第四、第五、第六和第七代谢热信号，进行代谢热特征提取及归一化处理，生成对应的第一、第二、第三、第四、第五、第六和第七代谢热特征数据；
对得到的所有特征数据进行特征融合处理，生成第一特征向量；
对所述第一特征向量，基于血糖浮动预测模型进行浮动血糖值预测处理，生成第一浮动血糖数据；并根据所述第一标定血糖数据和所述第一浮动血糖数据相加的和，生成第一预测血糖数据。
优选的，所述光学信号类型包括光体积描记图法信号类型；
所述三个光学信号波段包括650nm红外波段、940nm近红外波段和1050nm近红外波段。
优选的，所述对所述第一、第二和第三光学信号进行人体光信号特征提取及归一化处理，生成对应的第一、第二和第三光学特征数据组，具体包括：
从所述第一光学信号、所述第二光学信号或所述第三光学信号的中间信号时段提取第一指定时长的光学信号片段，生成对应的第一片段信号；
按预设的带通滤波频段，对所述第一片段信号进行带通滤波处理，生成对应的第一滤波信号；并对所述第一滤波信号进行信号峰值识别处理，得到对应的多个第一信号峰值数据；并对所述多个第一信号峰值数据进行均值计算，生成对应的第一平均峰值数据；并对所述第一滤波信号进行信号上下翻转处理，生成对应的第一翻转信号；并对所述第一翻转信号进行信号峰值识别处理，得到对应的多个第二信号峰值数据；并对所述多个第二信号峰值数据进行均值计算，并对计算结果取反生成对应的第一平均谷值数据；并根据所述第一平均峰值数据减去所述第一平均谷值数据的差，生成对应的第一特征数据；并对所述第一特征数据进行归一化处理生成交流特征数据；
按预设的低通滤波频率阈值，对所述第一片段信号进行低通滤波处理，生成对应的第二滤波信号；并对所述第二滤波信号进行均值计算，生成对应的第 二特征数据；并对所述第二特征数据进行归一化处理生成直流特征数据；
由得到的所述交流特征数据和所述直流特征数据，组成对应的第一光学特征数据组、第二光学特征数据组或第三光学特征数据组。
优选的，所述对所述第四光学信号进行环境光信号特征提取及归一化处理，生成对应的第四光学特征数据，具体包括：
从所述第四光学信号的中间信号时段提取第一指定时长的光学信号片段，生成第二片段信号；
对所述第二片段信号进行均值计算，生成第三特征数据；
对所述第三特征数据进行归一化处理生成所述第四光学特征数据。
优选的，所述对所述第一、第二、第三、第四、第五、第六和第七代谢热信号，进行代谢热特征提取及归一化处理，生成对应的第一、第二、第三、第四、第五、第六和第七代谢热特征数据，具体包括：
从所述第一、第二、第三、第四、第五、第六或第七代谢热信号的最后信号时段提取第二指定时长的代谢热信号片段，生成对应的第三片段信号；
对所述第三片段信号进行均值计算，生成对应的第四特征数据；
对所述第四特征数据进行归一化处理生成对应的第一代谢热特征数据、第二代谢热特征数据、第三代谢热特征数据、第四代谢热特征数据、第五代谢热特征数据、第六代谢热特征数据或第七代谢热特征数据。
优选的，所述血糖浮动预测模型基于多层神经网络的网络结构予以实现；所述血糖浮动预测模型包括输入层、隐藏层和输出层；所述输入层包括第一数量M的输入节点；所述隐藏层包括第二数量N的隐藏层输入节点和第三数量S的隐藏层输出节点；所述输出层包括一个输出节点；所述隐藏层的各个所述隐藏层输入节点与所述输入层的所有输入节点分别连接，形成对应的第一全连接网络；所述隐藏层的各个所述隐藏层输出节点与所有所述隐藏层输入节点分别连接，形成对应的第二全连接网络；所述输出层的所述输出节点与所有所述隐藏层输入节点连接；所述第一数量M与所述第一特征向量的特征分类 总数关联，第二数量N>第三数量S>0；
所述血糖浮动预测模型在进行所述浮动血糖值预测处理时，调用所述输入层将所述第一特征向量的各个特征数据按顺序提取出来输入到对应的所述输入节点生成对应的输入节点数据；
并调用所述隐藏层基于各个所述第一全连接网络的参数矩阵，对与之连接的所有所述输入节点数据进行对应的第一全连接网络运算，并将运算结果输入到对应的所述隐藏层输入节点作为隐藏层输入节点数据；
并调用所述隐藏层基于各个所述第二全连接网络的参数矩阵，对与之连接的所有所述隐藏层输入节点数据进行对应的第二全连接网络运算，并将运算结果输入到对应的所述隐藏层输出节点作为隐藏层输出节点数据；
并调用所述输出层基于预设的激活函数，对所有所述隐藏层输出节点数据进行对应的激活函数运算，并将运算结果作为所述第一浮动血糖数据进行输出。
本发明实施例第二方面提供了一种实现上述第一方面所述的方法的装置，包括：获取模块、第一数据采集模块、第二数据采集模块、第一特征数据处理模块、第二特征数据处理模块、特征数据融合处理模块和血糖预测处理模块；
所述获取模块用于获取标定血糖值生成第一标定血糖数据；
所述第一数据采集模块用于按预设的光学信号采集时长、人体光学信号类型和三个人体光学信号波段，持续采集人体光信号生成对应的第一光学信号、第二光学信号和第三光学信号；并按所述光学信号采集时长，持续采集环境光信号生成第四光学信号；
所述第二数据采集模块用于按预设的代谢热信号采集时长，持续采集人体接触热信号生成第一代谢热信号，并持续采集人体近端的辐射热信号生成第二代谢热信号，并持续采集人体近端的温度变化信息构成第三代谢热信号，并持续采集人体近端的湿度变化信息构成第四代谢热信号，并对用于采集辐射热信号的近端辐射传感器的校准输出信息进行持续采集构成第五代谢热信 号，并持续采集人体远端的温度变化信息构成第六代谢热信号，并持续采集人体远端的湿度变化信息构成第七代谢热信号；
所述第一特征数据处理模块用于对所述第一、第二和第三光学信号进行人体光信号特征提取及归一化处理，生成对应的第一、第二和第三光学特征数据组；并对所述第四光学信号进行环境光信号特征提取及归一化处理，生成对应的第四光学特征数据；
所述第二特征数据处理模块用于对所述第一、第二、第三、第四、第五、第六和第七代谢热信号，进行代谢热特征提取及归一化处理，生成对应的第一、第二、第三、第四、第五、第六和第七代谢热特征数据；
所述特征数据融合处理模块用于对得到的所有特征数据进行特征融合处理，生成第一特征向量；
所述血糖预测处理模块用于对所述第一特征向量，基于血糖浮动预测模型进行浮动血糖值预测处理，生成第一浮动血糖数据；并根据所述第一标定血糖数据和所述第一浮动血糖数据相加的和，生成第一预测血糖数据。
本发明实施例第三方面提供了一种电子设备，包括：存储器、处理器和收发器；
所述处理器用于与所述存储器耦合，读取并执行所述存储器中的指令，以实现上述第一方面所述的方法步骤；
所述收发器与所述处理器耦合，由所述处理器控制所述收发器进行消息收发。
本发明实施例第四方面提供了一种计算机可读存储介质，所述计算机可读存储介质存储有计算机指令，当所述计算机指令被计算机执行时，使得所述计算机执行上述第一方面所述的方法的指令。
本发明实施例提供了一种基于光学信号特征及代谢热特征的血糖预测方法、装置、电子设备及计算机可读存储介质，对人体的光体积描记图法信号及环境光学信号进行采集，对与人体代谢热相关的接触热、辐射热、温度、湿度 及辐射传感器校准信号进行采集，对与人体代谢热相关的环境温度、湿度信号进行采集；并对采集到的光学信号、代谢热相关信号进行特征提取与归一化处理构成特征向量；并使用体现光学、代谢热与血糖变化三者关联关系的血糖浮动预测模型对特征向量进行预测得到对应的血糖浮动数据，从而基于预先获得的血糖标定值+血糖浮动数据就可得到血糖预测数据。通过本发明，为长期血糖观测提供例如一种无创观测机制，减少例如对用户造成的不便，降低例如观测难度，提高了用户体验。
附图说明
图1为本发明实施例一提供的一种基于光学信号特征及代谢热特征的血糖预测方法示意图；
图2为本发明实施例一提供的血糖浮动预测模型的神经网络结构示意图；
图3为本发明实施例二提供的一种基于光学信号特征及代谢热特征的血糖预测装置的模块结构图；
图4为本发明实施例三提供的一种电子设备的结构示意图。
具体实施方式
为了使本发明的目的、技术方案和优点更加清楚，下面将结合附图对本发明作进一步地详细描述，显然，所描述的实施例仅仅是本发明一部份实施例，而不是全部的实施例。基于本发明中的实施例，本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其它实施例，都属于本发明保护的范围。
本发明实施例一提供的一种基于光学信号特征及代谢热特征的血糖预测方法，如图1为本发明实施例一提供的一种基于光学信号特征及代谢热特征的血糖预测方法示意图所示，本方法主要包括如下步骤：
步骤1，获取标定血糖值生成第一标定血糖数据。
需要说明的是，本发明实施例方法在当前步骤之前会预先进行血糖标定 处理生成标定血糖值，具体包括：
步骤A1，通过常规血糖检测手段对当前用户进行一次或多次常规血糖检测得到一个或多个测量血糖数据；并记录当次血糖检测的时间信息和用户基本生理信息，采集和记录当次血糖检测时的光学相关信号和代谢热相关信号；并由每次的测量血糖数据、时间信息、用户基本生理信息、光学相关信号和代谢热相关信号组成对应的校准参考数据组；
其中，用户基本生理信息包括年龄、性别、身高、体重、身体质量指数(Body Mass Index，BMI)等；光学相关信号包括三个光学信号波段(650nm红外波段、940nm近红外波段和1050nm近红外波段)的人体光体积描记图法信号和环境光学信号；代谢热相关信号包括人体接触热信号、人体近端的辐射热信号、人体近端的温度变化信号、人体近端的湿度变化信号、用于采集辐射热信号的近端辐射传感器的校准输出信号、人体远端的温度变化信号以及人体远端的湿度变化信号；
步骤A2，若只对用户做一次常规血糖检测，则将这次检测得到测量血糖数据作为标定血糖值并存储，将对应的校准参考数据组中除测量血糖数据之外的其他数据作为标定条件数据组并存储；
步骤A3，若只对用户做了多次常规血糖检测，则对得到的多个测量血糖数据进行均值计算生成对应的平均测量血糖数据；并将与平均测量血糖数据的差值为最小值的测量血糖数据作为标定血糖值并存储，并将与标定血糖值对应的校准参考数据组中除测量血糖数据之外的其他数据作为标定条件数据组并存储。
这里，标定血糖值和标定条件数据组一旦创建完成就会保存作为当前用户的基准数据。在后续每次对当前用户基于本发明实施例进行血糖观测时，都会以该基准数据的标定条件数据组作为预测模型参考对当前用户的血糖波动进行预测从而得到对应的浮动血糖数据，再由该基准数据的标定血糖值+浮动血糖数据就可得到当次该用户的血糖预测数据。也就是说，在创建了当前用户 的基准数据之后，后续每次对该用户的血糖观测过程都是无创观测处理过程。
步骤2，按预设的光学信号采集时长、人体光学信号类型和三个人体光学信号波段，持续采集人体光信号生成对应的第一光学信号、第二光学信号和第三光学信号；并按光学信号采集时长，持续采集环境光信号生成第四光学信号；
其中，光学信号类型包括光体积描记图法信号类型；三个光学信号波段包括650nm红外波段、940nm近红外波段和1050nm近红外波段。
这里，为保证光学信号的采集质量，光学信号采集时长常规被设为30秒或1分钟，也可基于信号采集设备的采集质量对其另行设置；在进行人体光信号采集时，为保证不同波段光源不对彼此造成相互干扰，常规会根据预设的采样频率(常规为125Hz)按时间先后，分别使用光波段为650nm红外波段、940nm近红外波段和1050nm近红外波段的三个发光二极管(light-emitting diode，LED)灯贴近人体手指指腹处依次进行照射，并使用三个对应的光敏二极管在该手指指背上与各个LED灯垂直对应的位置对三个波段的透射光信号依次进行采集，从而得到具体为650nm红外波段透射光的第一光学信号、具体为940nm红外波段透射光的第二光学信号以及具体为1050nm红外波段透射光的第三光学信号；在进行环境光信号采集时，为保证不受上述LED灯光源干扰，常规会在上述第一、第二和第三光学信号采集结束之后根据预设的采样频率(常规为125Hz)对环境背景光的自然光信号进行持续采集，从而得到第四光学信号；上述第一、第二、第三和第四光学信号的采集频率一致，且信号时长相同。
步骤3，按预设的代谢热信号采集时长，持续采集人体接触热信号生成第一代谢热信号，并持续采集人体近端的辐射热信号生成第二代谢热信号，并持续采集人体近端的温度变化信息构成第三代谢热信号，并持续采集人体近端的湿度变化信息构成第四代谢热信号，并对用于采集辐射热信号的近端辐射传感器的校准输出信息进行持续采集构成第五代谢热信号，并持续采集人体远端的温度变化信息构成第六代谢热信号，并持续采集人体远端的湿度变化 信息构成第七代谢热信号。
这里，为保证代谢热信号与光学信号的数据同步性，对代谢热信号的采集可与第一、第二和第三光学信号中任一光学信号的采集过程同步，此时其对应的代谢热信号采集时长与光学信号采集时长一致；也可与第一、第二和第三光学信号的总体采集过程同步，此时其对应的代谢热信号采集时长大于或等于三倍光学信号采集时长；若要进一步保证代谢热信号与光学信号的数据精度一致性，可将所有代谢热信号的采样频率设为与光学信号采样频率(常规为125Hz)一致；
在对人体接触热信号进行持续采集时，使用接触式传感器贴着上述手指指腹进行热信号采集；
在对人体近端的辐射热信号、温度变化信息和湿度变化信息进行持续采集时，使用距离上述手指指腹一定微小距离的近端辐射传感器、温度传感器和湿度传感器进行信号采集；
采集近端辐射传感器的校准输出信息是为了给采集到的近端辐射热信号提供一个校准信号；
在对人体远端的温度变化信息和湿度变化信息进行持续采集时，使用距离上述手指一定距离的温度传感器和湿度传感器进行信号采集，该距离大于前述微小距离以确保采集到的环境温度与湿度不受人体的温、湿度参数影响。
步骤4，对第一、第二和第三光学信号进行人体光信号特征提取及归一化处理，生成对应的第一、第二和第三光学特征数据组；
其中，第一、第二和第三光学特征数据组都由交流特征数据和直流特征数据组成；
这里，人体光信号特征提取实际就是对三路人体光体积描记图法信号的交、直流特征进行提取；
具体包括：步骤41，从第一光学信号、第二光学信号或第三光学信号的中间信号时段提取第一指定时长的光学信号片段，生成对应的第一片段信号；
这里，第一指定时长常规为20秒；
步骤42，按预设的带通滤波频段，对第一片段信号进行带通滤波处理，生成对应的第一滤波信号；并对第一滤波信号进行信号峰值识别处理，得到对应的多个第一信号峰值数据；并对多个第一信号峰值数据进行均值计算，生成对应的第一平均峰值数据；并对第一滤波信号进行信号上下翻转处理，生成对应的第一翻转信号；并对第一翻转信号进行信号峰值识别处理，得到对应的多个第二信号峰值数据；并对多个第二信号峰值数据进行均值计算，并对计算结果取反生成对应的第一平均谷值数据；并根据第一平均峰值数据减去第一平均谷值数据的差，生成对应的第一特征数据；并对第一特征数据进行归一化处理生成交流特征数据；
这里，实际就是在对第一、第二和第三光学信号的交流特征信息进行提取；带通滤波频段常规为0.5-10Hz；本发明实施例方法将低于0.5Hz的信号视为直流特征信号，高于10Hz的信号视为干扰或噪声信号，通过滤除直流特征信号和干扰或噪声信号就可得到交流特征信号；交流特征信号的特点就是存在波形的幅差也就是峰-谷差值，这里计算的交流特征数据实际就是交流特征信号的平均幅值数据；
进一步的，在对第一特征数据进行归一化处理生成交流特征数据时，本发明实施例在已有的用于训练后续预测模型的大量训练数据的基础上根据预设的归一化模式提供至少两种归一化处理流程，具体包括：
步骤B1，当归一化模式为第一模式时，使用最值归一化函数对第一特征数据进行归一化处理；
其中，最值归一化函数为
Figure PCTCN2022097242-appb-000001
X norm为归一化数据，X为输入的第一特征数据，X min、X max分别为由上述大量训练数据组成的数据集合中与第一特征数据类型一致的训练数据的最小、最大值；
步骤B2，当归一化模式为第二模式时，使用均值方差归一化函数对第一特征数据进行归一化处理；
其中，均值方差归一化函数为
Figure PCTCN2022097242-appb-000002
X norm为归一化数据，X为输入的第一特征数据，μ、σ分别为由上述大量训练数据组成的数据集合中与第一特征数据类型一致的训练数据的平均值和标准差；
步骤43，按预设的低通滤波频率阈值，对第一片段信号进行低通滤波处理，生成对应的第二滤波信号；并对第二滤波信号进行均值计算，生成对应的第二特征数据；并对第二特征数据进行归一化处理生成直流特征数据；
这里，实际就是在对第一、第二和第三光学信号的直流特征信息进行提取；低通滤波频率阈值常规为0.5Hz；通过保留低于0.5Hz的信号得到直流特征信号，交流特征信号的特点就是波形基本平整无明显峰-谷变化，这里计算的直流特征数据实际就是直流特征信号的平均幅值数据；
在对第二特征数据进行归一化处理生成直流特征数据时，其归一化处理过程与步骤41中对第一特征数据的归一化处理过程类似，在此不做进一步赘述；
步骤44，由得到的交流特征数据和直流特征数据，组成对应的第一光学特征数据组、第二光学特征数据组或第三光学特征数据组。
步骤5，对第四光学信号进行环境光信号特征提取及归一化处理，生成对应的第四光学特征数据；
具体包括：从第四光学信号的中间信号时段提取第一指定时长的光学信号片段，生成第二片段信号；并对第二片段信号进行均值计算，生成第三特征数据；并对第三特征数据进行归一化处理生成第四光学特征数据。
这里，第三特征数据可以反映环境背景光的平均光强；在对第三特征数据进行归一化处理生成第四光学特征数据时，其归一化处理过程与步骤41中对第一特征数据的归一化处理过程类似，在此不做进一步赘述。
步骤6，对第一、第二、第三、第四、第五、第六和第七代谢热信号，进行代谢热特征提取及归一化处理，生成对应的第一、第二、第三、第四、第五、第六和第七代谢热特征数据；
具体包括：步骤61，从第一、第二、第三、第四、第五、第六或第七代谢热信号的最后信号时段提取第二指定时长的代谢热信号片段，生成对应的第三片段信号；
这里，因为与代谢热相关的这些信号大多为缓慢变化信号，对应的传感器采集数据会呈现缓慢变化的趋势，需要一定时间才能再次到达一个稳定值，所以代谢热信号的片段截取方式是对最后指定长度的片段进行截取；第二指定时长常规为10秒；
步骤62，对第三片段信号进行均值计算，生成对应的第四特征数据；
这里，若第三片段信号为第一代谢热信号的片段数据，那么第四特征数据就是人体接触热的平均值；若第三片段信号为第二代谢热信号的片段数据，那么第四特征数据就是人体近端辐射热的平均值；若第三片段信号为第三代谢热信号的片段数据，那么第四特征数据就是人体近端温度平均值；若第三片段信号为第四代谢热信号的片段数据，那么第四特征数据就是人体近端湿度平均值；若第三片段信号为第五代谢热信号的片段数据，那么第四特征数据就是用于采集辐射热信号的近端辐射传感器的校准平均值；若第三片段信号为第六代谢热信号的片段数据，那么第四特征数据就是人体远端温度平均值实际就是环境温度平均值；若第三片段信号为第七代谢热信号的片段数据，那么第四特征数据就是人体远端湿度平均值实际就是环境湿度平均值；
步骤63，对第四特征数据进行归一化处理，生成对应的第一代谢热特征数据、第二代谢热特征数据、第三代谢热特征数据、第四代谢热特征数据、第五代谢热特征数据、第六代谢热特征数据或第七代谢热特征数据。
这里，在对第四特征数据进行归一化处理时，其归一化处理过程与步骤41中对第一特征数据的归一化处理过程类似，在此不做进一步赘述；在第三片段信号分别为第一、第二、第三、第四、第五、第六和第七代谢热信号的片段数据时，归一化处理结果即是对应的第一、第二、第三、第四、第五、第六和第七代谢热特征数据。
步骤7，对得到的所有特征数据进行特征融合处理，生成第一特征向量。
这里，按后续预测模型对输入数据的顺序要求，对第一、第二和第三光学特征数据组、第四光学特征数据以及第一、第二、第三、第四、第五、第六和第七代谢热特征数据的14个特征数据进行排序，并根据排序结果构建对应的第一特征向量。
需要说明的是，本发明实施例在特定情况下，还可对本次预测的时间及个人基本生理信息(年龄、性别、身高、体重、BMI)进行实时采集，并将相关采集数据纳入到第一特征向量中；特定情况包括以采集时间作为血糖变化因子之一，以个人基本生理信息变化作为血糖变化因子之一等。
步骤8，对第一特征向量，基于血糖浮动预测模型进行浮动血糖值预测处理，生成第一浮动血糖数据；
其中，血糖浮动预测模型基于多层神经网络的网络结构予以实现；血糖浮动预测模型包括输入层、隐藏层和输出层；输入层包括第一数量M的输入节点；隐藏层包括第二数量N的隐藏层输入节点和第三数量S的隐藏层输出节点；输出层包括一个输出节点；隐藏层的各个隐藏层输入节点与输入层的所有输入节点分别连接，形成对应的第一全连接网络；隐藏层的各个隐藏层输出节点与所有隐藏层输入节点分别连接，形成对应的第二全连接网络；输出层的输出节点与所有隐藏层输入节点连接；第一数量M与第一特征向量的特征分类总数关联，第二数量N>第三数量S>0。
这里，本发明实施例的血糖浮动预测模型的结构如图2为本发明实施例一提供的血糖浮动预测模型的神经网络结构示意图所示；该模型基于比尔-朗伯定律和代谢热构象理论构建第一、第二全连接网络的网络运算模型和运算所需的参数矩阵结构；该预测模型在每次预测时，会以第一特征向量的数据种类对输入层的输入节点数量进行设置，并以当前用户基准数据的标定条件数据组与第一特征向量的数据种类对应关系创建对应的标定特征向量，再根据标定特征向量对第一、第二全连接网络的参数矩阵以及输出层预设激活函数 的参数进行微调；再根据第一特征向量对当前用户在当前时刻的血糖波动值进行预测从而得到对应的第一浮动血糖数据。
需要说明的是，第一特征向量的数据种类中必然包括上述处理步骤产生的14个特征数据(第一、第二和第三光学特征数据组、第四光学特征数据以及第一、第二、第三、第四、第五、第六和第七代谢热特征数据)，还可以可选地包括本次预测的时间、年龄、性别、身高、体重、BMI等特征数据；在以第一特征向量的数据种类对输入层的输入节点数量进行设置时，需要根据第一特征向量的实际数据种类增减对应的输入节点。
需要说明的是，在以当前用户基准数据的标定条件数据组与第一特征向量的数据种类对应关系创建对应的标定特征向量时，需按类似步骤4对第一、第二、第三光学信号进行人体光信号特征提取及归一化处理的方式对标定条件数据组的三个光学信号波段(650nm红外波段、940nm近红外波段和1050nm近红外波段)的人体光体积描记图法信号进行交、直流特征数据提取和归一化处理从而得到6个人体光学信号标定特征数据；按类似步骤5对第四光学信号进行环境光信号特征提取及归一化处理的方式对标定条件数据组的环境光学信号进行特征数据提取和归一化处理从而得到1个环境光学标定特征数据；按类似步骤6对第一、第二、第三、第四、第五、第六和第七代谢热信号进行代谢热特征提取及归一化处理的方式对标定条件数据组的人体接触热信号、人体近端的辐射热信号、人体近端的温度变化信号、人体近端的湿度变化信号、用于采集辐射热信号的近端辐射传感器的校准输出信号、人体远端的温度变化信号以及人体远端的湿度变化信号进行特征数据提取和归一化处理从而得到7个代谢热信号标定特征数据；若第一特征向量的数据种类中还包括时间、年龄、性别、身高、体重或BMI等特征数据种类，则根据标定条件数据组的时间信息、用户基本生理信息(年龄、性别、身高、体重、BMI)进一步生成对应的时间、年龄、性别、身高、体重或BMI标定特征数据，并将之与前述6个人体光学信号标定特征数据、1个环境光学标定特征数据和7个代谢热信号标 定特征数据按与第一特征向量相同的数据排序结构进行标定特征数据融合处理，从而得到标定特征向量。
需要说明的是，本发明实施例的血糖浮动预测模型是预先通过采集的大量不同个体数据训练形成的，所以模型的预测精度与训练数据所属人群特征相关；若要使得模型预测结果与个性化特征进一步相关，就需要在使用时根据标定特征向量对第一、第二全连接网络的参数矩阵以及输出层预设激活函数的参数进行微调从而达到个性化校准目的；完成个性化校准之后，模型可根据当前个体的个性化基准特征对预测结果进行微调，从而使得预测结果能更加贴近当前个体的真实数据，提高了模型的个性化预测精度。
进一步的，在血糖浮动预测模型在进行浮动血糖值预测处理时，具体包括：
步骤C1，调用输入层将第一特征向量的各个特征数据按顺序提取出来输入到对应的输入节点生成对应的输入节点数据；
步骤C2，调用隐藏层基于各个第一全连接网络的参数矩阵，对与之连接的所有输入节点数据进行对应的第一全连接网络运算，并将运算结果输入到对应的隐藏层输入节点作为隐藏层输入节点数据；
步骤C3，调用隐藏层基于各个第二全连接网络的参数矩阵，对与之连接的所有隐藏层输入节点数据进行对应的第二全连接网络运算，并将运算结果输入到对应的隐藏层输出节点作为隐藏层输出节点数据；
步骤C4，调用输出层基于预设的激活函数，对所有隐藏层输出节点数据进行对应的激活函数运算，并将运算结果作为第一浮动血糖数据进行输出。
步骤9，根据第一标定血糖数据和第一浮动血糖数据相加的和，生成第一预测血糖数据。
这里，以当前用户基准数据的标定血糖值也就是第一标定血糖数据，加上血糖浮动预测模型输出的血糖波动值也就是第一浮动血糖数据，就可以得到对当前用户在当前时刻的血糖预测数值也就是第一预测血糖数据。
图3为本发明实施例二提供的一种基于光学信号特征及代谢热特征的血 糖预测装置的模块结构图，该装置可以为实现本发明实施例方法的终端设备或者服务器，也可以为与上述终端设备或者服务器连接的实现本发明实施例方法的装置，例如该装置可以是上述终端设备或者服务器的装置或芯片系统。如图3所示，该装置包括：获取模块201、第一数据采集模块202、第二数据采集模块203、第一特征数据处理模块204、第二特征数据处理模块205、特征数据融合处理模块206和血糖预测处理模块207。
获取模块201用于获取标定血糖值生成第一标定血糖数据。
第一数据采集模块202用于按预设的光学信号采集时长、人体光学信号类型和三个人体光学信号波段，持续采集人体光信号生成对应的第一光学信号、第二光学信号和第三光学信号；并按光学信号采集时长，持续采集环境光信号生成第四光学信号。
第二数据采集模块203用于按预设的代谢热信号采集时长，持续采集人体接触热信号生成第一代谢热信号，并持续采集人体近端的辐射热信号生成第二代谢热信号，并持续采集人体近端的温度变化信息构成第三代谢热信号，并持续采集人体近端的湿度变化信息构成第四代谢热信号，并对用于采集辐射热信号的近端辐射传感器的校准输出信息进行持续采集构成第五代谢热信号，并持续采集人体远端的温度变化信息构成第六代谢热信号，并持续采集人体远端的湿度变化信息构成第七代谢热信号。
第一特征数据处理模块204用于对第一、第二和第三光学信号进行人体光信号特征提取及归一化处理，生成对应的第一、第二和第三光学特征数据组；并对第四光学信号进行环境光信号特征提取及归一化处理，生成对应的第四光学特征数据。
第二特征数据处理模块205用于对第一、第二、第三、第四、第五、第六和第七代谢热信号，进行代谢热特征提取及归一化处理，生成对应的第一、第二、第三、第四、第五、第六和第七代谢热特征数据。
特征数据融合处理模块206用于对得到的所有特征数据进行特征融合处 理，生成第一特征向量。
血糖预测处理模块207用于对第一特征向量，基于血糖浮动预测模型进行浮动血糖值预测处理，生成第一浮动血糖数据；并根据第一标定血糖数据和第一浮动血糖数据相加的和，生成第一预测血糖数据。
本发明实施例提供的一种基于光学信号特征及代谢热特征的血糖预测装置，可以执行上述方法实施例中的方法步骤，其实现原理和技术效果类似，在此不再赘述。
需要说明的是，应理解以上装置的各个模块的划分仅仅是一种逻辑功能的划分，实际实现时可以全部或部分集成到一个物理实体上，也可以物理上分开。且这些模块可以全部以软件通过处理元件调用的形式实现；也可以全部以硬件的形式实现；还可以部分模块通过处理元件调用软件的形式实现，部分模块通过硬件的形式实现。例如，获取模块可以为单独设立的处理元件，也可以集成在上述装置的某一个芯片中实现，此外，也可以以程序代码的形式存储于上述装置的存储器中，由上述装置的某一个处理元件调用并执行以上确定模块的功能。其它模块的实现与之类似。此外这些模块全部或部分可以集成在一起，也可以独立实现。这里所描述的处理元件可以是一种集成电路，具有信号的处理能力。在实现过程中，上述方法的各步骤或以上各个模块可以通过处理器元件中的硬件的集成逻辑电路或者软件形式的指令完成。
例如，以上这些模块可以是被配置成实施以上方法的一个或多个集成电路，例如：一个或多个特定集成电路(Application Specific Integrated Circuit，ASIC)，或，一个或多个数字信号处理器(Digital Signal Processor，DSP)，或，一个或者多个现场可编程门阵列(Field Programmable Gate Array，FPGA)等。再如，当以上某个模块通过处理元件调度程序代码的形式实现时，该处理元件可以是通用处理器，例如中央处理器(Central Processing Unit，CPU)或其它可以调用程序代码的处理器。再如，这些模块可以集成在一起，以片上系统(System-on-a-chip，SOC)的形式实现。
在上述实施例中，可以全部或部分地通过软件、硬件、固件或者其任意组合来实现。当使用软件实现时，可以全部或部分地以计算机程序产品的形式实现。该计算机程序产品包括一个或多个计算机指令。在计算机上加载和执行该计算机程序指令时，全部或部分地产生按照本发明实施例所描述的流程或功能。上述计算机可以是通用计算机、专用计算机、计算机网络、或者其他可编程装置。上述计算机指令可以存储在计算机可读存储介质中，或者从一个计算机可读存储介质向另一个计算机可读存储介质传输，例如，上述计算机指令可以从一个网站站点、计算机、服务器或数据中心通过有线(例如同轴电缆、光纤、数字用户线路(Digital Subscriber Line，DSL))或无线(例如红外、无线、蓝牙、微波等)方式向另一个网站站点、计算机、服务器或数据中心进行传输。上述计算机可读存储介质可以是计算机能够存取的任何可用介质或者是包含一个或多个可用介质集成的服务器、数据中心等数据存储设备。上述可用介质可以是磁性介质，(例如，软盘、硬盘、磁带)、光介质(例如，DVD)、或者半导体介质(例如固态硬盘(solid state disk，SSD))等。
图4为本发明实施例三提供的一种电子设备的结构示意图。该电子设备可以为前述的终端设备或者服务器，也可以为与前述终端设备或者服务器连接的实现本发明实施例方法的终端设备或服务器。如图4所示，该电子设备可以包括：处理器301(例如CPU)、存储器302、收发器303；收发器303耦合至处理器301，处理器301控制收发器303的收发动作。存储器302中可以存储各种指令，以用于完成各种处理功能以及实现本发明上述实施例中提供的方法和处理过程。优选的，本发明实施例涉及的电子设备还包括：电源304、系统总线305以及通信端口306。系统总线305用于实现元件之间的通信连接。上述通信端口306用于电子设备与其他外设之间进行连接通信。
在图4中提到的系统总线可以是外设部件互连标准(Peripheral Component Interconnect，PCI)总线或扩展工业标准结构(Extended Industry Standard Architecture，EISA)总线等。该系统总线可以分为地址 总线、数据总线、控制总线等。为便于表示，图中仅用一条粗线表示，但并不表示仅有一根总线或一种类型的总线。通信接口用于实现数据库访问装置与其他设备(例如客户端、读写库和只读库)之间的通信。存储器可能包含随机存取存储器(Random Access Memory，RAM)，也可能还包括非易失性存储器(Non-Volatile Memory)，例如至少一个磁盘存储器。
上述的处理器可以是通用处理器，包括中央处理器CPU、网络处理器(Network Processor，NP)等；还可以是数字信号处理器DSP、专用集成电路ASIC、现场可编程门阵列FPGA或者其他可编程逻辑器件、分立门或者晶体管逻辑器件、分立硬件组件。
需要说明的是，本发明实施例还提供一种计算机可读存储介质，该存储介质中存储有指令，当其在计算机上运行时，使得计算机执行上述实施例中提供的方法和处理过程。
本发明实施例还提供一种运行指令的芯片，该芯片用于执行上述实施例中提供的方法和处理过程。
本发明实施例提供了一种基于光学信号特征及代谢热特征的血糖预测方法、装置、电子设备及计算机可读存储介质，对人体的光体积描记图法信号及环境光学信号进行采集，对与人体代谢热相关的接触热、辐射热、温度、湿度及辐射传感器校准信号进行采集，对与人体代谢热相关的环境温度、湿度信号进行采集；并对采集到的光学信号、代谢热相关信号进行特征提取与归一化处理构成特征向量；并使用体现光学、代谢热与血糖变化三者关联关系的血糖浮动预测模型对特征向量进行预测得到对应的血糖浮动数据，从而基于预先获得的血糖标定值+血糖浮动数据就可得到血糖预测数据。通过本发明，为长期血糖观测提供例如一种无创观测机制，减少例如对用户造成的不便，降低例如观测难度，提高了用户体验。
专业人员应该还可以进一步意识到，结合本文中所公开的实施例描述的各示例的单元及算法步骤，能够以电子硬件、计算机软件或者二者的结合来实 现，为了清楚地说明硬件和软件的可互换性，在上述说明中已经按照功能一般性地描述了各示例的组成及步骤。这些功能究竟以硬件还是软件方式来执行，取决于技术方案的特定应用和设计约束条件。专业技术人员可以对每个特定的应用来使用不同方法来实现所描述的功能，但是这种实现不应认为超出本发明的范围。
结合本文中所公开的实施例描述的方法或算法的步骤可以用硬件、处理器执行的软件模块，或者二者的结合来实施。软件模块可以置于随机存储器(RAM)、内存、只读存储器(ROM)、电可编程ROM、电可擦除可编程ROM、寄存器、硬盘、可移动磁盘、CD-ROM、或技术领域内所公知的任意其它形式的存储介质中。
以上所述的具体实施方式，对本发明的目的、技术方案和有益效果进行了进一步详细说明，所应理解的是，以上所述仅为本发明的具体实施方式而已，并不用于限定本发明的保护范围，凡在本发明的精神和原则之内，所做的任何修改、等同替换、改进等，均应包含在本发明的保护范围之内。

Claims (9)

  1. 一种基于光学信号特征及代谢热特征的血糖预测方法，其特征在于，所述方法包括：
    获取标定血糖值生成第一标定血糖数据；
    按预设的光学信号采集时长、人体光学信号类型和三个人体光学信号波段，持续采集人体光信号生成对应的第一光学信号、第二光学信号和第三光学信号；并按所述光学信号采集时长，持续采集环境光信号生成第四光学信号；
    按预设的代谢热信号采集时长，持续采集人体接触热信号生成第一代谢热信号，并持续采集人体近端的辐射热信号生成第二代谢热信号，并持续采集人体近端的温度变化信息构成第三代谢热信号，并持续采集人体近端的湿度变化信息构成第四代谢热信号，并对用于采集辐射热信号的近端辐射传感器的校准输出信息进行持续采集构成第五代谢热信号，并持续采集人体远端的温度变化信息构成第六代谢热信号，并持续采集人体远端的湿度变化信息构成第七代谢热信号；
    对所述第一、第二和第三光学信号进行人体光信号特征提取及归一化处理，生成对应的第一、第二和第三光学特征数据组；
    对所述第四光学信号进行环境光信号特征提取及归一化处理，生成对应的第四光学特征数据；
    对所述第一、第二、第三、第四、第五、第六和第七代谢热信号，进行代谢热特征提取及归一化处理，生成对应的第一、第二、第三、第四、第五、第六和第七代谢热特征数据；
    对得到的所有特征数据进行特征融合处理，生成第一特征向量；
    对所述第一特征向量，基于血糖浮动预测模型进行浮动血糖值预测处理，生成第一浮动血糖数据；并根据所述第一标定血糖数据和所述第一浮动血糖数据相加的和，生成第一预测血糖数据。
  2. 根据权利要求1所述的基于光学信号特征及代谢热特征的血糖预测方 法，其特征在于，
    所述光学信号类型包括光体积描记图法信号类型；
    所述三个光学信号波段包括650nm红外波段、940nm近红外波段和1050nm近红外波段。
  3. 根据权利要求1所述的基于光学信号特征及代谢热特征的血糖预测方法，其特征在于，所述对所述第一、第二和第三光学信号进行人体光信号特征提取及归一化处理，生成对应的第一、第二和第三光学特征数据组，具体包括：
    从所述第一光学信号、所述第二光学信号或所述第三光学信号的中间信号时段提取第一指定时长的光学信号片段，生成对应的第一片段信号；
    按预设的带通滤波频段，对所述第一片段信号进行带通滤波处理，生成对应的第一滤波信号；并对所述第一滤波信号进行信号峰值识别处理，得到对应的多个第一信号峰值数据；并对所述多个第一信号峰值数据进行均值计算，生成对应的第一平均峰值数据；并对所述第一滤波信号进行信号上下翻转处理，生成对应的第一翻转信号；并对所述第一翻转信号进行信号峰值识别处理，得到对应的多个第二信号峰值数据；并对所述多个第二信号峰值数据进行均值计算，并对计算结果取反生成对应的第一平均谷值数据；并根据所述第一平均峰值数据减去所述第一平均谷值数据的差，生成对应的第一特征数据；并对所述第一特征数据进行归一化处理生成交流特征数据；
    按预设的低通滤波频率阈值，对所述第一片段信号进行低通滤波处理，生成对应的第二滤波信号；并对所述第二滤波信号进行均值计算，生成对应的第二特征数据；并对所述第二特征数据进行归一化处理生成直流特征数据；
    由得到的所述交流特征数据和所述直流特征数据，组成对应的第一光学特征数据组、第二光学特征数据组或第三光学特征数据组。
  4. 根据权利要求1所述的基于光学信号特征及代谢热特征的血糖预测方法，其特征在于，所述对所述第四光学信号进行环境光信号特征提取及归一化处理，生成对应的第四光学特征数据，具体包括：
    从所述第四光学信号的中间信号时段提取第一指定时长的光学信号片段，生成第二片段信号；
    对所述第二片段信号进行均值计算，生成第三特征数据；
    对所述第三特征数据进行归一化处理生成所述第四光学特征数据。
  5. 根据权利要求1所述的基于光学信号特征及代谢热特征的血糖预测方法，其特征在于，所述对所述第一、第二、第三、第四、第五、第六和第七代谢热信号，进行代谢热特征提取及归一化处理，生成对应的第一、第二、第三、第四、第五、第六和第七代谢热特征数据，具体包括：
    从所述第一、第二、第三、第四、第五、第六或第七代谢热信号的最后信号时段提取第二指定时长的代谢热信号片段，生成对应的第三片段信号；
    对所述第三片段信号进行均值计算，生成对应的第四特征数据；
    对所述第四特征数据进行归一化处理生成对应的第一代谢热特征数据、第二代谢热特征数据、第三代谢热特征数据、第四代谢热特征数据、第五代谢热特征数据、第六代谢热特征数据或第七代谢热特征数据。
  6. 根据权利要求1所述的基于光学信号特征及代谢热特征的血糖预测方法，其特征在于，
    所述血糖浮动预测模型基于多层神经网络的网络结构予以实现；所述血糖浮动预测模型包括输入层、隐藏层和输出层；所述输入层包括第一数量M的输入节点；所述隐藏层包括第二数量N的隐藏层输入节点和第三数量S的隐藏层输出节点；所述输出层包括一个输出节点；所述隐藏层的各个所述隐藏层输入节点与所述输入层的所有输入节点分别连接，形成对应的第一全连接网络；所述隐藏层的各个所述隐藏层输出节点与所有所述隐藏层输入节点分别连接，形成对应的第二全连接网络；所述输出层的所述输出节点与所有所述隐藏层输入节点连接；所述第一数量M与所述第一特征向量的特征分类总数关联，第二数量N>第三数量S>0；
    所述血糖浮动预测模型在进行所述浮动血糖值预测处理时，调用所述输 入层将所述第一特征向量的各个特征数据按顺序提取出来输入到对应的所述输入节点生成对应的输入节点数据；
    并调用所述隐藏层基于各个所述第一全连接网络的参数矩阵，对与之连接的所有所述输入节点数据进行对应的第一全连接网络运算，并将运算结果输入到对应的所述隐藏层输入节点作为隐藏层输入节点数据；
    并调用所述隐藏层基于各个所述第二全连接网络的参数矩阵，对与之连接的所有所述隐藏层输入节点数据进行对应的第二全连接网络运算，并将运算结果输入到对应的所述隐藏层输出节点作为隐藏层输出节点数据；
    并调用所述输出层基于预设的激活函数，对所有所述隐藏层输出节点数据进行对应的激活函数运算，并将运算结果作为所述第一浮动血糖数据进行输出。
  7. 一种用于实现权利要求1-6任一项所述的基于光学信号特征及代谢热特征的血糖预测方法步骤的装置，其特征在于，所述装置包括：获取模块、第一数据采集模块、第二数据采集模块、第一特征数据处理模块、第二特征数据处理模块、特征数据融合处理模块和血糖预测处理模块；
    所述获取模块用于获取标定血糖值生成第一标定血糖数据；
    所述第一数据采集模块用于按预设的光学信号采集时长、人体光学信号类型和三个人体光学信号波段，持续采集人体光信号生成对应的第一光学信号、第二光学信号和第三光学信号；并按所述光学信号采集时长，持续采集环境光信号生成第四光学信号；
    所述第二数据采集模块用于按预设的代谢热信号采集时长，持续采集人体接触热信号生成第一代谢热信号，并持续采集人体近端的辐射热信号生成第二代谢热信号，并持续采集人体近端的温度变化信息构成第三代谢热信号，并持续采集人体近端的湿度变化信息构成第四代谢热信号，并对用于采集辐射热信号的近端辐射传感器的校准输出信息进行持续采集构成第五代谢热信号，并持续采集人体远端的温度变化信息构成第六代谢热信号，并持续采集人 体远端的湿度变化信息构成第七代谢热信号；
    所述第一特征数据处理模块用于对所述第一、第二和第三光学信号进行人体光信号特征提取及归一化处理，生成对应的第一、第二和第三光学特征数据组；并对所述第四光学信号进行环境光信号特征提取及归一化处理，生成对应的第四光学特征数据；
    所述第二特征数据处理模块用于对所述第一、第二、第三、第四、第五、第六和第七代谢热信号，进行代谢热特征提取及归一化处理，生成对应的第一、第二、第三、第四、第五、第六和第七代谢热特征数据；
    所述特征数据融合处理模块用于对得到的所有特征数据进行特征融合处理，生成第一特征向量；
    所述血糖预测处理模块用于对所述第一特征向量，基于血糖浮动预测模型进行浮动血糖值预测处理，生成第一浮动血糖数据；并根据所述第一标定血糖数据和所述第一浮动血糖数据相加的和，生成第一预测血糖数据。
  8. 一种电子设备，其特征在于，包括：存储器、处理器和收发器；
    所述处理器用于与所述存储器耦合，读取并执行所述存储器中的指令，以实现权利要求1-6任一项所述的方法步骤；
    所述收发器与所述处理器耦合，由所述处理器控制所述收发器进行消息收发。
  9. 一种计算机可读存储介质，其特征在于，所述计算机可读存储介质存储有计算机指令，当所述计算机指令被计算机执行时，使得所述计算机执行权利要求1-6任一项所述的方法的指令。
PCT/CN2022/097242 2021-11-22 2022-06-07 基于光学信号特征及代谢热特征的血糖预测方法和装置 WO2023087672A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
EP22894215.7A EP4437954A1 (en) 2021-11-22 2022-06-07 Blood glucose prediction method and device based on optical signal features and metabolic thermal features

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202111386652.X 2021-11-22
CN202111386652.XA CN114098724B (zh) 2021-11-22 2021-11-22 基于光学信号特征及代谢热特征的血糖预测方法和装置

Publications (1)

Publication Number Publication Date
WO2023087672A1 true WO2023087672A1 (zh) 2023-05-25

Family

ID=80439497

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2022/097242 WO2023087672A1 (zh) 2021-11-22 2022-06-07 基于光学信号特征及代谢热特征的血糖预测方法和装置

Country Status (3)

Country Link
EP (1) EP4437954A1 (zh)
CN (1) CN114098724B (zh)
WO (1) WO2023087672A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117373586A (zh) * 2023-08-28 2024-01-09 北京华益精点生物技术有限公司 血糖数据比对方法及相关设备

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114098724B (zh) * 2021-11-22 2024-03-26 乐普（北京）医疗器械股份有限公司 基于光学信号特征及代谢热特征的血糖预测方法和装置
CN115299887B (zh) * 2022-10-10 2023-01-03 安徽星辰智跃科技有限责任公司 一种动态代谢功能的检测量化方法和系统

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20050043630A1 (en) * 2003-08-21 2005-02-24 Buchert Janusz Michal Thermal Emission Non-Invasive Analyte Monitor
CN102293654A (zh) * 2011-06-17 2011-12-28 清华大学 基于代谢热-光学方法的无创血糖检测仪
CN104665840A (zh) * 2015-03-02 2015-06-03 桂林麦迪胜电子科技有限公司 无创血糖测量方法及指端测量探头
CN107802255A (zh) * 2017-11-30 2018-03-16 杭州电子科技大学 一种基于代谢法的血糖数据处理方法及装置
JP2019198363A (ja) * 2018-05-14 2019-11-21 株式会社カスタム 非侵襲性血糖測定器
CN113397538A (zh) * 2021-07-20 2021-09-17 深圳市微克科技有限公司 一种可穿戴嵌入式系统的光学血糖算法
CN114098724A (zh) * 2021-11-22 2022-03-01 乐普（北京）医疗器械股份有限公司 基于光学信号特征及代谢热特征的血糖预测方法和装置

Family Cites Families (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
RU2007147839A (ru) * 2005-05-24 2009-06-27 Кониклейке Филипс Электроникс Н.В. (Nl) Датчик глюкозы
WO2007072300A2 (en) * 2005-12-22 2007-06-28 Koninklijke Philips Electronics N.V. System for non-invasive measurement of blood glucose concentration
CN104771181A (zh) * 2015-04-16 2015-07-15 桂林电子科技大学 一种反射式无创血糖检测仪
CN106361305A (zh) * 2016-09-19 2017-02-01 爱国者电子科技有限公司 糖代谢率的测量方法和装置
CN107174259A (zh) * 2017-06-26 2017-09-19 上海理工大学 基于多波长能量守恒法的无创血糖值采集装置及计算方法
CN109330607A (zh) * 2018-08-29 2019-02-15 桂林永成医疗科技有限公司 基于微创血糖值校准的无创血糖检测方法及其检测装置
CN112386252A (zh) * 2019-08-15 2021-02-23 博邦芳舟医疗科技（北京）有限公司 一种血糖测量探头和血糖测量装置和方法
CN112133442B (zh) * 2020-09-22 2024-02-13 博邦芳舟医疗科技（北京）有限公司 一种连续无创血糖检测装置及方法
CN113288132B (zh) * 2021-05-06 2023-04-14 广东工业大学 用于预测血糖值的方法、装置、存储介质及处理器

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20050043630A1 (en) * 2003-08-21 2005-02-24 Buchert Janusz Michal Thermal Emission Non-Invasive Analyte Monitor
CN102293654A (zh) * 2011-06-17 2011-12-28 清华大学 基于代谢热-光学方法的无创血糖检测仪
CN104665840A (zh) * 2015-03-02 2015-06-03 桂林麦迪胜电子科技有限公司 无创血糖测量方法及指端测量探头
CN107802255A (zh) * 2017-11-30 2018-03-16 杭州电子科技大学 一种基于代谢法的血糖数据处理方法及装置
JP2019198363A (ja) * 2018-05-14 2019-11-21 株式会社カスタム 非侵襲性血糖測定器
CN113397538A (zh) * 2021-07-20 2021-09-17 深圳市微克科技有限公司 一种可穿戴嵌入式系统的光学血糖算法
CN114098724A (zh) * 2021-11-22 2022-03-01 乐普（北京）医疗器械股份有限公司 基于光学信号特征及代谢热特征的血糖预测方法和装置

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117373586A (zh) * 2023-08-28 2024-01-09 北京华益精点生物技术有限公司 血糖数据比对方法及相关设备

Also Published As

Publication number Publication date
EP4437954A1 (en) 2024-10-02
CN114098724B (zh) 2024-03-26
CN114098724A (zh) 2022-03-01

Similar Documents

Publication Publication Date Title
WO2023087672A1 (zh) 基于光学信号特征及代谢热特征的血糖预测方法和装置
Chen et al. 5G-smart diabetes: Toward personalized diabetes diagnosis with healthcare big data clouds
US11350868B2 (en) Electrocardiogram information processing method and electrocardiogram workstation system
Habbu et al. Estimation of blood glucose by non-invasive method using photoplethysmography
CN106659392A (zh) 非侵扰式皮肤组织水合确定设备及相关的方法
Fan et al. Homecare-oriented intelligent long-term monitoring of blood pressure using electrocardiogram signals
Sung et al. Mobile physiological measurement platform with cloud and analysis functions implemented via IPSO
Chen et al. A new deep learning framework based on blood pressure range constraint for continuous cuffless BP estimation
Haque et al. A novel technique for non-invasive measurement of human blood component levels from fingertip video using DNN based models
Golap et al. Hemoglobin and glucose level estimation from PPG characteristics features of fingertip video using MGGP-based model
CN107689040A (zh) 血糖检测的移动终端
Pinge et al. A comparative study between ECG-based and PPG-based heart rate monitors for stress detection
Wang et al. Cuffless blood pressure estimation using dual physiological signal and its morphological features
Samanta et al. Optimized Tree Strategy with Principal Component Analysis Using Feature Selection‐Based Classification for Newborn Infant’s Jaundice Symptoms
Lavanya et al. [Retracted] Wearable Sensor‐Based Edge Computing Framework for Cardiac Arrhythmia Detection and Acute Stroke Prediction
CN108364690A (zh) 一种多功能生命体征检测系统及其工作方法
Jayakody et al. HemoSmart: a non-invasive, machine learning based device and mobile app for anemia detection
Skrivanos et al. Home Healthcare Technologies and Services: Heart-Rate Fetus Monitoring System Using an MCU ESP8266 Node
Lins et al. Accuracy of wearable electronic device compared to manual and automatic methods of blood pressure determination
WO2023197426A1 (zh) 一种结合大数据模型和个性化模型的血糖预测方法和装置
CN116602668B (zh) 一种全自动智能血糖检测系统
Tiwari et al. Non-invasive monitoring of health using sensor-rich wearables and smart devices
Turnip et al. PPG Signal-Based Blood Pressure Classification With Ensemble Bagged Trees Method
Raju et al. Real-Time Hemoglobin Measurement Using Smartphone Video and Artificial Neural Network
US20240156377A1 (en) yGT ESTIMATION DEVICE, yGT ESTIMATION METHOD, AND COMPUTER PROGRAM

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 22894215

Country of ref document: EP

Kind code of ref document: A1

WWE Wipo information: entry into national phase

Ref document number: 2022894215

Country of ref document: EP

NENP Non-entry into the national phase

Ref country code: DE

ENP Entry into the national phase

Ref document number: 2022894215

Country of ref document: EP

Effective date: 20240624