Random forest based feedback edema diagnosis and treatment integrated system

Abstract

The application discloses a random forest-based feedback type edema diagnosis and treatment integrated system, and belongs to the technical field of medical instruments.The system comprises a bioelectric impedance monitoring module, a control module and a far-infrared light irradiation array; the bioelectric impedance monitoring module and the far-infrared light module are connected with the control module; wherein the bioelectric impedance monitoring module measures the voltage of an edema site and transmits the voltage to the control module; the control module calculates the bioelectric impedance information according to the voltage, combines the patient's meta-information, evaluates the edema degree by using a random forest model, determines an irradiation scheme, and sends a control instruction containing the irradiation scheme to the far-infrared light irradiation array; and the far-infrared light irradiation array irradiates and treats the edema site of the patient according to the control instruction. The integrated system can be used at home by the patient to realize effective diagnosis and treatment of edema without excessive dependence on the treatment experience of doctors. The integrated system can shorten the recovery period of edema.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of digital medical instruments, and particularly relates to a feedback type edema diagnosis and treatment integrated system based on a random forest. BACKGROUND

[0002] Edema is a common pathological reaction, which is a swelling of tissue due to accumulation of fluid in the interstitial space, usually caused by abnormal elevation of venous hydrostatic pressure, abnormal decrease of plasma colloid osmotic pressure or lymphatic backflow obstruction. Patients often have edema after surgery due to long-term fasting, inflammatory reaction, bed rest and many other reasons, which seriously affects the postoperative recovery and daily life of patients.

[0003] Generally speaking, within three days after surgery, the surgical site of the patient will rapidly edema and the degree of edema will reach a peak, and then the degree of edema will decrease until it completely disappears as the patient recovers. Edema caused by soft tissue surgery often disappears in about a week, while edema caused by bone surgery often lasts for a month or even several months. However, due to the scarcity of medical resources, current edema treatment often only focuses on the first three days after surgery. During this period, the doctor suppresses the rapid occurrence of edema by having the patient take oral medication or intravenous injection of drugs. The long recovery period often has to be recovered by the patient at home.

[0004] In summary, the existing edema diagnosis and treatment method relies too much on the treatment experience of doctors. In the case of a shortage of medical resources, patients can only receive effective treatment in the first three days after surgery. The edema recovery time is long and it is difficult for patients to complete diagnosis and treatment at home. In order to accurately and effectively monitor and treat the edema site of the patient without increasing the medical pressure, the feedback type home edema diagnosis and treatment integrated system based on a random forest is proposed. SUMMARY

[0005] The purpose of the embodiment of the present application is to provide a feedback type edema diagnosis and treatment integrated system based on a random forest, which can solve the technical problem that the existing edema diagnosis and treatment method relies too much on the treatment experience of doctors. In the case of a shortage of medical resources, patients can only receive effective treatment in the first three days after surgery. The edema recovery time is long and it is difficult for patients to complete diagnosis and treatment at home.

[0006] In order to solve the above technical problems, the present application is implemented as follows:

[0007] The embodiment of the present application provides a feedback type edema diagnosis and treatment integrated system based on a random forest, which comprises a control module, a far infrared light treatment module and a bioelectrical impedance monitoring module.

[0008] The control module is connected with the far infrared light treatment module.

[0009] The bioelectrical impedance monitoring module is connected with the control module.

[0010] The bioelectrical impedance monitoring module is configured to monitor the electrical impedance information of the edema site and feed back the electrical impedance information to the control module.

[0011] The control module is configured to determine the radiation power and the radiation time by using a random forest model according to the electrical impedance information and the basic information of the patient, and send a control instruction containing the radiation power and the radiation time to the far-infrared light treatment module.

[0012] The far-infrared light treatment module is configured to perform irradiation treatment on the edema site of the patient according to the radiation power and the radiation time.

[0013] Optionally, the control module comprises a microcontroller, a power supply and a Bluetooth unit.

[0014] The microcontroller is connected with the mobile terminal of the patient through the Bluetooth unit to display the electrical impedance information of the patient and the corresponding irradiation scheme on the mobile terminal of the patient.

[0015] Optionally, the control module further comprises a training unit.

[0016] The training unit is configured to pre-collect a plurality of samples, the features of the samples being patient information, and the labels of the samples being irradiation schemes selected by doctors according to experience, divide the samples into a training set and a test set according to a preset ratio, divide the training set into N sample groups, train N decision trees respectively through the N sample groups to constitute a random forest model, and verify the decision results of the random forest model through the test set.

[0017] Optionally, the preset ratio between the training set and the test set is 8:2.

[0018] Optionally, N is 10, and the random forest model is constituted by 10 decision trees.

[0019] Optionally, each decision tree determines an irradiation scheme independently according to the electrical impedance information and the basic information of the patient, and the control module is configured to finally determine the radiation power and the radiation time according to the voting results of each decision tree.

[0020] Optionally, the bioelectrical impedance monitoring module comprises a current excitation submodule and a signal receiving submodule.

[0021] The current excitation submodule comprises a frequency generator, a first excitation electrode and a second excitation electrode.

[0022] The signal receiving submodule comprises an analog-to-digital conversion circuit (ADC) unit, a digital signal processing (DSP) unit, a first receiving electrode and a second receiving electrode.

[0023] The frequency generator is electrically connected with the two excitation electrodes, and the frequency generator generates a current signal, which is input to the patient's skin through the first excitation electrode, flows through the edema site, and then flows out of the human body through the second excitation electrode.

[0024] The ADC unit is electrically connected with the two receiving electrodes, and the ADC unit is used to acquire the excitation response, and the DSP unit is used to perform discrete Fourier transform on the acquisition result to calculate the electrical impedance information of the edema site.

[0025] Optionally, the first excitation electrode and the second excitation electrode are respectively arranged on the two sides of the edema site.

[0026] The first receiving electrode and the second receiving electrode are respectively arranged on the two sides of the edema site.

[0027] The distance between the first excitation electrode and the edema site is greater than the distance between the first receiving electrode and the edema site, and the distance between the second excitation electrode and the edema site is greater than the distance between the second receiving electrode and the edema site.

[0028] Optionally, the far infrared light treatment module includes an LED array.

[0029] Optionally, the LED array is composed of 10x20 LEDs.

[0030] In the embodiments of the present application, the edema is monitored by the bioelectrical impedance method, the edema is treated by the infrared light irradiation method, and the random forest model is used to dynamically adjust the irradiation scheme according to the monitoring result, so as to construct a feedback type diagnosis and treatment integrated device. The patient can realize effective diagnosis and treatment of edema by using the feedback type diagnosis and treatment integrated device at home, without excessive dependence on the treatment experience of the doctor, and the recovery period of the edema is greatly shortened. BRIEF DESCRIPTION OF DRAWINGS

[0031] Figure 1 It is a structure schematic diagram of a feedback type edema diagnosis and treatment integrated system based on a random forest provided by the embodiments of the present application.

[0032] 1-control module, 11-microcontroller, 12-power supply, 13-Bluetooth unit, 14-training module;

[0033] 2-far infrared light treatment module, 21-LED array;

[0034] 3-bioelectrical impedance monitoring module, 31-current excitation submodule, 311-frequency generator, 312-first excitation electrode, 313-second excitation electrode, 32-signal receiving submodule, 321-ADC unit, 322-DSP unit, 323-first receiving electrode, 324-second receiving electrode;

[0035] 4-edema site;

[0036] 5 - mobile terminal.

[0037] The objectives, functional characteristics and advantages of the present application will be further described with reference to the embodiments and in conjunction with the accompanying drawings. DETAILED DESCRIPTION

[0038] In order to make the objectives, technical solutions and advantages of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings of the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort belong to the scope of protection of the present application.

[0039] The large commercial complex security management level system provided by the embodiments of the present application will be described in detail below in conjunction with the drawings, specific embodiments and application scenarios.

[0040] Embodiment one

[0041] Reference Figure 1 Fig. 1 shows a structure schematic diagram of a feedback type edema diagnosis and treatment integrated system based on random forest provided by an embodiment of the present application.

[0042] The feedback type edema diagnosis and treatment integrated system based on random forest provided by the present application comprises a control module 1, a far infrared light treatment module 2 and a bioelectrical impedance monitoring module 3.

[0043] The control module 1 is connected with the far infrared light treatment module 2, and the bioelectrical impedance monitoring module 3 is connected with the control module 1.

[0044] According to the above connection relationship, a feedback type control can be formed, the edema site 4 is monitored in real time through the bioelectrical impedance monitoring module 3, and then the radiation power and radiation time of the far infrared light treatment module 2 are dynamically adjusted according to the monitoring result.

[0045] The bioelectrical impedance monitoring module 3 is used for monitoring the bioelectrical impedance information of the edema site 4 and feeding back the bioelectrical impedance information to the control module 1.

[0046] It should be noted that the interstitial fluid contains Na + , K + , Ca 2+ and other ions, has conductivity, and has similar electrical properties to resistance. When edema occurs in the lower limbs, the interstitial fluid in the corresponding site increases, and the electrical properties show a decrease in resistance. Therefore, by monitoring the resistance of the edema site, quantitative detection of the degree of edema can be realized.

[0047] Optionally, the bioelectrical impedance monitoring module 3 is an AD5933 type impedance converter. The AD5933 type impedance converter is a 12-bit impedance converter. The internal frequency generator of the AD5933 type impedance converter can send a pre-set signal to stimulate the bioelectrical impedance. The internal ADC (Analog-to-Digital Converter) collects the stimulation response. The internal DSP (Digital Signal Processing) performs discrete Fourier transform on the collected results, thereby calculating the bioelectrical resistance and bioelectrical reactance. Finally, the measurement results are sent to the microcontroller module through the I2C protocol. The input and output electrodes convert the electronic current of the external circuit into the ionic current inside the body, thereby forming a transition link between the external circuit and the inside of the body.

[0048] In actual application, the accuracy of the bioelectrical impedance monitoring module 3 can be tested by a bioelectrical impedance measurement experiment. The experimental method is mainly to divide the subjects into two categories. One category of subjects has no edema, and the other category of subjects has edema. The two categories of subjects are required to stretch their legs horizontally. The electrodes of the bioelectrical impedance measurement part of the present application are attached to the two ends of the edema site of the edema subject or the same site of the non-edema subject. The observed bioelectrical impedance results are measured.

[0049] The control module 1 is used to determine the radiation power and the radiation time by using the random forest model according to the bioelectrical impedance information and the basic information of the patient, and send the control instruction containing the radiation power and the radiation time to the far infrared light treatment module 2.

[0050] It should be noted that the random forest algorithm is a statistical learning method based on multiple decision trees. The decision tree is an effective prediction model, which essentially represents the mapping relationship between attributes and values. In the present application, the attributes are the patient's own information, including the basic information (such as age, gender, height, and weight) and the bioelectrical impedance information (bioelectrical resistance and bioelectrical reactance under different frequencies); the value is the patient's irradiation dose and irradiation time. Therefore, in the present application, the decision tree can achieve the purpose of determining the irradiation scheme according to the patient's state. However, a single decision tree is prone to overfitting, thereby limiting its prediction ability. The decision tree uses the BootStrap method to resample, divides the original sample into several different sample groups, and trains different decision trees according to these sample groups, thereby forming a random forest. When predicting, these decision trees independently predict and output the final prediction result in a voting manner. Compared with the single decision tree method, the random forest method is less prone to overfitting, and its prediction ability is often stronger.

[0051] The far infrared light treatment module 2 is used to perform irradiation treatment on the edema site 4 of the patient according to the radiation power and the radiation time.

[0052] It should be noted that the far infrared rays increase the kinetic energy of tissue molecules, produce a thermal effect, thereby expanding capillaries, increasing blood flow, improving blood circulation and enhancing metabolism, etc., have positive biological effects such as anti-inflammatory analgesic, promoting tissue regeneration, etc., and are suitable for treating postoperative edema of patients. However, the irradiation scheme of far infrared rays needs to be adjusted according to the patient's own condition to avoid adverse effects such as burns, which is also an important reason for introducing a feedback link.

[0053] In actual application process, the treatment effect of the integrated system provided by the application can be verified by treatment effect verification experiment. The specific method is: the edema subjects are divided into two categories, one category does not use the application, and the other category uses the application, on the one hand, the edema changes of each subject are compared longitudinally, and on the other hand, the edema changes between different subjects are compared horizontally.

[0054] In a possible implementation, the control module 1 includes a microcontroller 11, a power supply 12 and a Bluetooth unit 13, wherein the microcontroller 11 is the main body, and the power supply 12 and the Bluetooth unit 13 are necessary peripheral devices, and it can be understood that some necessary peripheral devices are not shown.

[0055] The microcontroller 11 is connected with the mobile terminal 5 of the patient through the Bluetooth unit 13 to display the electrical impedance information of the patient and the corresponding irradiation scheme on the mobile terminal 5 of the patient.

[0056] The microcontroller 11 is the main part of data processing. After receiving the measured bioelectrical impedance information, the microcontroller 11 first performs data denoising, and the denoising method can adopt mean filtering, and then inputs the bioelectrical impedance value and the patient basic information together as a feature vector into a random forest model to calculate the suitable irradiation scheme and send it to the far infrared light treatment module 2.

[0057] Optionally, the microcontroller 11 is an STM32F103C8T6 type microcontroller, and the Bluetooth unit 13 is a KT6368A type Bluetooth module.

[0058] In a possible implementation, the control module 1 further includes a training unit 14; the training unit 14 is used for pre-collecting a plurality of samples, the features of the samples are patient information, and the labels of the samples are the irradiation schemes selected by doctors according to experience, and the samples are divided into a training set and a test set according to a preset proportion; the training set is divided into N sample groups, and N decision trees are trained through the N sample groups to constitute a random forest model; and the decision results of the random forest model are verified through the test set.

[0059] In actual application process, the accuracy of the decision result of the random forest model can also be detected through random forest prediction experiment. The main experimental method is: two methods of inviting doctors to judge and using random forest prediction are adopted, and the irradiation scheme (irradiation time and irradiation power) is determined according to the information (bioelectrical impedance value and basic information) of the subjects independently, and the results given by the two methods are compared.

[0060] In a possible implementation, the preset ratio between the training set and the test set is 8:2. The preset ratio is set to 8:2 to balance between training and testing, and further improve the accuracy of the random forest model.

[0061] In a possible implementation, N is 10, and the random forest model is composed of 10 decision trees. N can also be other numbers, and a person skilled in the art can select a suitable number according to actual needs.

[0062] In a possible implementation, each decision tree independently determines the irradiation scheme according to the bioelectrical impedance information and the basic information of the patient, and the control module 1 is used to finally determine the radiation power and the radiation time according to the voting result of each decision tree. Compared with the single decision tree method, the random forest method is less likely to have overfitting phenomenon, and its prediction ability is often stronger.

[0063] In a possible implementation, the bioelectrical impedance monitoring module 3 includes a current excitation submodule 31 and a signal receiving submodule 32; the current excitation submodule 31 includes a frequency generator 311, a first excitation electrode 312 and a second excitation electrode 313; the signal receiving submodule 32 includes an analog-to-digital conversion circuit ADC unit 321, a digital signal processing DSP unit 322, a first receiving electrode 323 and a second receiving electrode 324; the frequency generator 311 is electrically connected with the two excitation electrodes, the frequency generator 311 generates a current signal, the current signal is input to the patient's skin through the first excitation electrode 312, and flows out of the human body through the second excitation electrode 313 after flowing through the edema site 4; the ADC unit 321 is electrically connected with the two receiving electrodes, and the ADC unit 321 is used to acquire an excitation response, and the DSP unit 322 is used to perform discrete Fourier transform on the acquisition result, and calculate the bioelectrical impedance information of the edema site 4.

[0064] In a possible implementation, the first excitation electrode 312 and the second excitation electrode 313 are respectively arranged on the two sides of the edema site 4; the first receiving electrode 323 and the second receiving electrode 324 are respectively arranged on the two sides of the edema site 4; the distance between the first excitation electrode 312 and the edema site 4 is greater than the distance between the first receiving electrode 323 and the edema site 4, and the distance between the second excitation electrode 313 and the edema site 4 is greater than the distance between the second receiving electrode 324 and the edema site 4.

[0065] In a possible implementation, the far-infrared light treatment module 2 includes an LED array 21.

[0066] In a possible implementation, the LED (Light-Emitting Diode) array 21 is composed of 10x20 LEDs, each of which can emit low-energy far-infrared light with a wavelength of 880 nm.

[0067] In a possible implementation, the microcontroller 11 sets an internal timer according to the calculated irradiation time, and when the internal timer ends, the microcontroller 11 receives an internal interrupt, and then stops delivering energy to the far-infrared light treatment module 2, thereby controlling the irradiation time; at the same time, the microcontroller 11 controls the current delivered to the far-infrared light treatment module 2 according to the calculated radiation power, thereby controlling the irradiation power.

[0068] In the embodiment, the edema is monitored by the bioelectrical impedance method, the edema is treated by the infrared light irradiation method, and the irradiation scheme is dynamically adjusted according to the monitoring result by using the random forest model, so as to construct a feedback type diagnosis and treatment integrated device. The patient can realize effective diagnosis and treatment of edema by using the feedback type diagnosis and treatment integrated device at home, without excessive dependence on the treatment experience of doctors, and the recovery period of edema is greatly shortened.

[0069] Embodiment two

[0070] Please refer to Figure 1 The embodiment provides a feedback type edema diagnosis and treatment integrated system based on a random forest, which comprises a control module 1, a far-infrared light irradiation array 2 and a bioelectrical impedance monitoring module 3.

[0071] The bioelectrical impedance monitoring module 3 and the far-infrared light irradiation array 2 are electrically connected with the control module 1.

[0072] The bioelectrical impedance monitoring module 3 is used for measuring the voltage value of the edema part of the patient and transmitting the voltage value to the control module 1.

[0073] The control module 1 is used for calculating bioelectrical impedance information according to the voltage value, evaluating the edema degree by using a random forest model in combination with the meta information of the patient, determining an irradiation scheme, and sending a control instruction containing the irradiation scheme to the far-infrared light irradiation array 2.

[0074] The far-infrared light irradiation array 2 is used for performing irradiation treatment on the edema part of the patient according to the control instruction.

[0075] The bioelectrical impedance information comprises static information and dynamic information.

[0076] The static information reflects the condition when the edema site is static, and the static information includes extracellular fluid resistance, bioelectric resistance at 8 predetermined frequencies, bioelectric reactance, and phase angle.

[0077] The dynamic information reflects the condition when the edema site is moving, and includes bioelectric resistance change rate and bioelectric reactance change rate at 8 predetermined frequencies.

[0078] The calculation method of the extracellular fluid resistance is as follows:

[0079] S301: Taking bioelectric resistance as the horizontal coordinate and bioelectric reactance as the vertical coordinate, the bioelectric resistance and reactance at 8 predetermined frequencies are mapped as 8 points on a complex plane.

[0080] S302: The 8 points are fitted by using the least square method, and according to the Cole-Cole theorem, the fitting result is an arc.

[0081] S303: The arc has two intersection points with the horizontal axis, and the value of the horizontal coordinate with the larger value is determined as the extracellular fluid resistance.

[0082] The calculation method of the phase angle is as follows:

[0083]

[0084] The calculation and measurement method of the bioelectric resistance change rate and bioelectric reactance change rate at 8 predetermined frequencies is as follows:

[0085] S401: The patient is required to flex and rotate the edema site, and the bioelectric resistance and bioelectric reactance curves during the movement are recorded.

[0086] S402: The maximum and minimum values of the bioelectric resistance and bioelectric reactance curves at 8 predetermined frequencies are found, and the maximum value of the bioelectric resistance at the i-th predetermined frequency is denoted as Ri max, the minimum value is denoted as Ri min, the maximum value of the bioelectric reactance is denoted as Xi max, and the minimum value is denoted as Xi min.

[0087] S403: The change rates of the bioelectric resistance and bioelectric reactance at 8 predetermined frequencies are calculated, that is, the change rate of the bioelectric resistance at the i-th predetermined frequency is:

[0088]

[0089] The change rate of the bioelectric reactance is:

[0090]

[0091] wherein X is the bioelectric reactance, and R is the bioelectric resistance.​​​​​​

[0092] Further, the bioelectrical impedance monitoring module 3 comprises a current excitation submodule 31 and a signal receiving submodule 32;

[0093] The current excitation submodule 31 comprises a frequency generator 311, a first excitation electrode 312 and a second excitation electrode 313;

[0094] The first excitation electrode 312 and the second excitation electrode 313 are both connected to the output end of the frequency generator 311;

[0095] The frequency generator 311 generates a sine current with a preset frequency and a preset amplitude, the sine current is injected into the human body through the first excitation electrode 312, and flows out of the human body through the second excitation electrode 313 after flowing through the edema site;

[0096] The signal receiving submodule 32 comprises a digital signal processing (DSP) unit 321, an analog-to-digital conversion (ADC) unit 322, a first receiving electrode 323 and a second receiving electrode 324;

[0097] The first receiving electrode 323 and the second receiving electrode 324 are electrically connected to the DSP unit 321 through the ADC unit 322;

[0098] The ADC unit 322 converts the voltage signal between the first receiving electrode 323 and the second receiving electrode 324 into a digital signal, and the DSP unit 321 filters and denoises the conversion result, and calculates the real part and the imaginary part of the voltage signal through discrete Fourier transform.

[0099] Further, the control module 1 comprises a microcontroller 11, a mobile power supply 12, a Bluetooth unit 13 and a self-correcting network 14;

[0100] The Bluetooth unit 13 and the self-correcting network 14 are both electrically connected to the microcontroller 11;

[0101] The Bluetooth unit 13 is used for wireless communication connection with the mobile terminal 5;

[0102] The mobile power supply 12 is electrically connected to all power-consuming modules for power supply;

[0103] The self-correcting network 14 calculates bioelectrical impedance information according to the real part and the imaginary part of the voltage signal under the instruction of the microcontroller 11, and performs self-correction every time it is restarted;

[0104] The microcontroller 11 evaluates the degree of edema by using a random forest model according to the bioelectrical impedance information and the patient meta-information, determines the irradiation scheme, and sends a control instruction containing the irradiation scheme to the far-infrared light irradiation array 2.

[0105] The Bluetooth unit 13 transmits the bioelectrical impedance information, the patient meta-information, the degree of edema, and the irradiation scheme to the mobile terminal for the patient to check.

[0106] Optionally, the training method of the random forest model is as follows:

[0107] S501: Collect a plurality of samples, the features of the samples are bioelectrical impedance information and patient meta-information, and the labels of the samples are the evaluations of the degree of edema by doctors;

[0108] S502: Subtract the mean value of each feature of each sample from all samples, and divide by the standard deviation of the feature of all samples, and replace the original feature with the standardized feature;

[0109] S503: Combine all features into a plurality of feature subsets;

[0110] S504: Train a random forest model for each feature subset, verify the performance of the model and calculate an index, select the model with the optimal index as the final model, and the corresponding feature subset as the final feature.

[0111] Further, the feature screening process is as follows:

[0112] Train a random forest using each feature, calculate the precision and recall of each model, arrange the 39 features in order from high to low according to the precision, and divide them into ten groups, of which the first nine groups each have four features, and the last group has three features, i.e., the first group of features is ; the same arrangement and grouping are performed according to the recall, and the features in the first group are ; ;

[0113] From and , take the intersection of the features and as a feature subset, and train a random forest model;

[0114] Then take and , repeat the above steps, and always keep , until and ; ​​​

[0115] Through the above process, 10 random forest models trained by different feature subsets are obtained, the AUC indicators of these models are compared, and the larger one is selected as the final random forest model, and the corresponding feature subset is the final used feature.

[0116] Further, the irradiation scheme is divided into four schemes, corresponding to four edema grading, including:

[0117] No irradiation, corresponding to no edema;

[0118] Weak irradiation, irradiation time is 10 minutes, single irradiation dose is 4 J / point, corresponding to mild edema;

[0119] Medium irradiation, irradiation time is 20 minutes, single irradiation dose is 5 J / point, corresponding to moderate edema;

[0120] Strong irradiation, irradiation time is 30 minutes, single irradiation dose is 6 J / point, corresponding to severe edema.

[0121] Optionally, the far infrared light irradiation array 2 includes a far infrared LED array 21, a temperature sensor 22 and a humidity sensor 23;

[0122] The far infrared LED array 21, the temperature sensor 22 and the humidity sensor 23 are electrically connected with the control module 1;

[0123] The far infrared LED array 21 is used to irradiate the edema of the patient according to the control instruction containing the irradiation scheme;

[0124] The temperature sensor 22 is used to monitor the temperature at the irradiation site, and when the temperature is too high, the irradiation is stopped;

[0125] The humidity sensor 23 is used to monitor the humidity at the irradiation site, and when the humidity is too low, the irradiation is stopped;

[0126] Wherein, the far infrared LED array is composed of 10*20 far infrared LEDs.

[0127] The above only describes the embodiments of the present application and is not used to limit the present application. For those skilled in the art, the present application can have various changes and variations. Any modification, equivalent replacement, improvement, etc. within the spirit and principles of the present application shall be included in the scope of claims of the present application.

Claims

1. A random forest-based feedback edema diagnosis and treatment integrated system, characterized in that, The control module (1), the far-infrared light irradiation array (2) and the bioelectrical impedance monitoring module (3) are included. The bioelectrical impedance monitoring module (3) and the far-infrared light irradiation array (2) are electrically connected with the control module (1). The bioelectrical impedance monitoring module (3) is used for measuring the voltage value of the edema site of a patient and transmitting the voltage value to the control module (1). The control module (1) is used for calculating bioelectrical impedance information according to the voltage value, evaluating the edema degree by combining the patient's meta information and using a random forest model, determining an irradiation scheme, and sending a control instruction containing the irradiation scheme to the far-infrared light irradiation array (2). The far-infrared light irradiation array (2) is used for performing irradiation treatment on the edema site of the patient according to the control instruction. The bioelectrical impedance information includes static information and dynamic information. The static information reflects the condition of the edema site when it is static, and includes extracellular fluid resistance, bioelectric resistance at 8 predetermined frequencies, bioelectric reactance and phase angle. The dynamic information reflects the condition of the edema site when it is moving, and includes bioelectric resistance change rate and bioelectric reactance change rate at 8 predetermined frequencies. The calculation method of the extracellular fluid resistance is as follows: S301: mapping the bioelectrical impedance at 8 predetermined frequencies to 8 points on a complex plane with bioelectric resistance as the horizontal coordinate and bioelectric reactance as the vertical coordinate; S302: fitting the 8 points by using the least square method, and according to the Cole-Cole theorem, the fitting result is an arc; S303: the arc has two intersection points with the horizontal axis, and the value of the horizontal coordinate with the larger value is determined as the extracellular fluid resistance; The calculation method of the phase angle is as follows: The calculation and measurement method of the bioelectric resistance change rate and the bioelectric reactance change rate at 8 predetermined frequencies is as follows: S401: requiring the patient to bend and stretch and rotate the edema site, and recording the bioelectric resistance and bioelectric reactance curves during the movement; S402: Find the maxima and minima of bioresistivity and bioimpedance curves at eight predetermined frequencies, and denote the values ​​of the curves at each frequency as... The maximum value of bioelectric resistance at a predetermined frequency is The minimum value is The maximum value of bioelectric reactance is The minimum value is ; S403: Calculate the rate of change of the bioelectric resistance and bioelectric reactance at 8 predetermined frequencies, i.e. the rate of change of the bioelectric resistance at the first predetermined frequency is: the rate of change of the bioelectric resistance at the first predetermined frequency is: The change rate of the bioelectric reactance is as follows: Wherein, X is the bioelectric reactance, and R is the bioelectric resistance.

2. The system of claim 1, wherein, The bioelectrical impedance monitoring module (3) includes a current excitation submodule (31) and a signal receiving submodule (32). The current excitation submodule (31) includes a frequency generator (311), a first excitation electrode (312) and a second excitation electrode (313). The first excitation electrode (312) and the second excitation electrode (313) are connected with the output end of the frequency generator (311). The frequency generator (311) generates a sine current with a preset frequency and a preset amplitude, the sine current is injected into the human body through the first excitation electrode (312), and flows out of the human body through the second excitation electrode (313) after flowing through the edema site. The signal receiving submodule (32) includes a digital signal processing (DSP) unit (321), an analog-to-digital conversion (ADC) unit (322), a first receiving electrode (323) and a second receiving electrode (324). The first receiving electrode (323) and the second receiving electrode (324) are electrically connected with the DSP unit (321) through the ADC unit (322); The ADC unit (322) converts the voltage signal between the first receiving electrode (323) and the second receiving electrode (324) into a digital signal, and the DSP unit (321) filters and denoises the conversion result and calculates the real part and the imaginary part of the voltage signal through discrete Fourier transform.

3. The system of claim 2, wherein, The control module (1) comprises a microcontroller (11), a mobile power supply (12), a Bluetooth unit (13) and a self-correcting network (14); The Bluetooth unit (13) and the self-correcting network (14) are electrically connected with the microcontroller (11); The Bluetooth unit (13) is used for wireless communication connection with a mobile terminal (5); The mobile power supply (12) is electrically connected with all power-consuming modules for power supply; The self-correcting network (14) calculates bioelectrical impedance information according to the real part and the imaginary part of the voltage signal under the instruction of the microcontroller (11) and performs self-correction every time it is restarted; The microcontroller (11) evaluates the degree of edema, determines an irradiation scheme and sends a control instruction containing the irradiation scheme to the far-infrared light irradiation array (2) according to bioelectrical impedance information and patient meta-information by using a random forest model; The Bluetooth unit (13) transmits bioelectrical impedance information, patient meta-information, the degree of edema and the irradiation scheme to the mobile terminal for the patient to check.

4. The system of claim 1, wherein, The training method of the random forest model is as follows: S501: Collect a plurality of samples, the features of the samples are bioelectrical impedance information and patient meta-information, and the labels of the samples are the evaluation of the degree of edema by doctors; S502: Replace the original features with standardized features by subtracting the mean value of all samples of each feature and dividing by the standard deviation of all samples of the feature for each sample; S503: Combine all features into a plurality of feature subsets; S504: Train a random forest model for each feature subset, verify the model performance and calculate an index, select the model with the optimal index as the final model, and the corresponding feature subset as the final feature.

5. The system of claim 4, wherein, The feature selection process is as follows: The precision and recall of each model are calculated by training random forest with each feature respectively; 39 features are arranged in order of precision from high to low and divided into ten groups, the first nine groups have four features each, and the last group has three features, i.e. the first group of features is ; the same arrangement and grouping are performed according to the recall, and the features of the first group are ; From And Start, take the intersection of the features that satisfy The characteristics And The characteristics Of a set of feature subsets, and train a random forest model; Subsequently, take and repeat the above steps, always keeping until and . Through the above process, 10 random forest models trained by different feature subsets are obtained, the AUC indexes of these models are compared, and the larger one is selected as the final random forest model, and the corresponding feature subset is the final feature used.

6. The system of claim 5, wherein, The irradiation scheme is divided into four schemes corresponding to four edema grading, including: No irradiation, corresponding to no edema; Weak irradiation, irradiation time is 10 minutes, single irradiation dose is 4 J / point, corresponding to mild edema; Medium irradiation, irradiation time is 20 minutes, single irradiation dose is 5 J / point, corresponding to moderate edema; Strong irradiation, irradiation time is 30 minutes, single irradiation dose is 6 J / point, corresponding to severe edema.

7. The system of claim 1, wherein, The far-infrared light irradiation array (2) comprises a far-infrared LED array (21), a temperature sensor (22) and a humidity sensor (23). The far infrared LED array (21), the temperature sensor (22) and the humidity sensor (23) are electrically connected with the control module (1); The far infrared LED array (21) is used for irradiating the edema of the patient according to the control instruction containing an irradiation scheme; The temperature sensor (22) is used for monitoring the temperature of the irradiation position, and the irradiation is stopped when the temperature is too high; The humidity sensor (23) is used for monitoring the humidity of the irradiation position, and the irradiation is stopped when the humidity is too low; The far infrared LED array is composed of 10*20 far infrared LEDs.

Images