Children primary immune thrombocytopenia identification method and device

Abstract

The invention belongs to the technical field of thrombocytopenia detection methods, and particularly discloses a children primary immune thrombocytopenia identification method and device.The method comprises the following steps that clinical data of a patient are collected and preprocessed; performing single-factor analysis on the preprocessed clinical data, and screening primary parameters; performing importance sorting on the primary parameters based on a machine learning model, and screening important parameters; and constructing a multi-classification logistic regression model, inputting the important parameters into the multi-classification logistic regression model, and outputting an identification result. By adopting the technical scheme, the important parameters are obtained, and the multi-classification logistic regression model is utilized to realize the I TP identification of the patient.

Description

Technical Field

[0001] The present invention belongs to the technical field of thrombocytopenia detection methods, and relates to a method and device for identifying primary immune thrombocytopenia in children. Background Art

[0002] Primary immune thrombocytopenia (ITP) in children is an acquired autoimmune disease characterized by isolated thrombocytopenia and bleeding. It is also the most common bleeding disease in children, with an incidence of (4-5) / 100,000, which has a significant impact on quality of life.

[0003] ITP is most common in children aged 4-6 years, with no significant gender differences. Older children at diagnosis tend to have a longer course of disease. ITP in children has an acute onset and a short course. 50%-70% of children with ITP recover within 3 months, 80% have normal platelet counts within 1 year, and 20% develop chronic ITP after a course of up to 1 year.

[0004] The pathogenesis of ITP in children remains unclear, and challenges persist, including a lack of a gold standard for diagnosis and a poor diagnostic timeline. Limited understanding of ITP and inadequate hardware make testing for it inadequate. For physicians outside specialized children's hospitals, testing standards vary widely, and methods are not standardized. This can lead to misdiagnosis and mistreatment, delaying early intervention and putting children at risk of severe bleeding. Summary of the Invention

[0005] The purpose of the present invention is to provide a method and device for identifying primary immune thrombocytopenia in children, using a multi-classification logistic regression model to achieve patient ITP identification.

[0006] In order to achieve the above object, the basic scheme of the present invention is: a method for identifying primary immune thrombocytopenia in children, comprising the following steps:

[0007] Collect patients' clinical data and perform preprocessing;

[0008] Univariate analysis was performed on the pre-processed clinical data to screen primary parameters;

[0009] Based on the machine learning model, the primary parameters are ranked by importance and important parameters are selected;

[0010] Construct a multi-classification logistic regression model, input important parameters into the multi-classification logistic regression model, and output the identification results.

[0011] The working principle and beneficial effects of this basic solution are as follows: This technical solution uses clinical data for secondary screening, first obtaining primary parameters, then using machine learning models to screen important parameters, gradually streamlining the data to obtain data that is more conducive to classification and identification, making it easier to use. Based on a multi-classification logistic regression model, identification results are obtained, improving identification efficiency and accuracy, and facilitating early intervention and treatment.

[0012] Furthermore, the patient's clinical data includes:

[0013] Demographic characteristics: gender, age;

[0014] Hematological indicators: hemoglobin, mean corpuscular volume, platelet count, absolute lymphocyte count, platelet / lymphocyte ratio, absolute neutrophil count, absolute monocyte count, large platelet ratio, white blood cell count, red blood cell count;

[0015] Biochemical indicators: lactate dehydrogenase, creatinine, aspartate aminotransferase, urea, uric acid, alkaline phosphatase, total protein, total bilirubin, alanine aminotransferase, glutamyl transpeptidase, albumin, globulin, aspartate / alanine ratio;

[0016] Electrolyte indicators: potassium ion, sodium ion, chloride ion, total calcium, phosphorus, magnesium;

[0017] Coagulation function indicators: thrombin time, prothrombin time, fibrinogen, PT international normalized ratio, activated partial thromboplastin time;

[0018] Other indicators: hepatitis B core antibody, hepatitis B e antibody, hepatitis B e antigen, congenital heart disease, and acquired heart disease.

[0019] Collect the required parameters for subsequent analysis.

[0020] Furthermore, the method for preprocessing the patient's clinical data is as follows:

[0021] Data cleaning: patients aged >18 years were excluded, and outliers outside the mean ± 3 times the standard deviation were removed;

[0022] Data normalization: standardize continuous variables;

[0023] Dataset division: The data were randomly divided into a modeling group and a validation group in a ratio of 7:3.

[0024] Preprocess clinical data and optimize data.

[0025] Furthermore, the important parameters include platelet count, absolute neutrophil count, hemoglobin, platelet / lymphocyte ratio (PLR), lactate dehydrogenase, age, white blood cell count, mean corpuscular volume, alkaline phosphatase, and absolute lymphocyte count.

[0026] Obtaining important parameters is conducive to improving identification accuracy.

[0027] Furthermore, the multi-classification logistic regression model is:

[0028] G_SLE=-10.6769+0.0259*platelet count-0.1895*neutrophil absolute value-0.0301*hemoglobin+0.0214*PLR+0.5741*age+0.0542*white blood cell count+0.0822*mean corpuscular volume-0.0126*alkaline phosphatase-0.4300*lymphocyte absolute value;

[0029] G_AL=-5.4630+0.0175*platelet count-0.2973*neutrophil absolute value-0.0727*hemoglobin+0.0222*PLR+0.1996*age+0.0902*white blood cell count+0.1184*mean corpuscular volume-0.0071*alkaline phosphatase-0.0366*lymphocyte absolute value;

[0030] G_AA=-3.3477+0.0132*platelet count-0.4862*neutrophil absolute value-0.0852*hemoglobin+0.0228*PLR+0.1751*age+0.0418*white blood cell count+0.1292*mean corpuscular volume-0.3119*lymphocyte absolute value;

[0031] G_ITP=0 (control group).

[0032] Then, G_SLE, G_AL, G_AA, and G_ITP are substituted into the following formula to obtain the corresponding probabilities of the four diseases:

[0033] P_SLE=exp(G_SLE) / [exp(G_SLE)+exp(G_AL)+exp(G_AA)+exp(G_ITP)];

[0034] P_AL=exp(G_AL) / [exp(G_SLE)+exp(G_AL)+exp(G_AA)+exp(G_ITP)];

[0035] P_AA=exp(G_AA) / [exp(G_SLE)+exp(G_AL)+exp(G_AA)+exp(G_ITP)];

[0036] P_ITP=exp(G_ITP) / [exp(G_SLE)+exp(G_AL)+exp(G_AA)+exp(G_ITP)].

[0037] Among them, P_SLE, P_AL, P_AA, and P_ITP are the classification probabilities of the corresponding diseases;

[0038] Compare the sizes of P_SLE, P_AL, P_AA, and P_ITP and sort them. The disease category corresponding to the largest value is the identification result.

[0039] The calculation is simple and easy to operate.

[0040] Furthermore, the machine learning model adopts one of random forest, support vector machine, extreme gradient boosting, naive Bayes, and artificial neural network machine learning models.

[0041] Choose the appropriate model according to your needs for easy use.

[0042] The present invention also provides a nursing device for children with primary immune thrombocytopenia, comprising a bed frame, a bed board, a monitoring mechanism, a human-computer interaction module and an identification module;

[0043] The bed board includes a first board and a second board, wherein the adjacent sides of the first board and the second board are hinged, and a movable groove is provided at the bottom of the first board and the second board, and a movable block is provided in the movable groove and moves along the movable groove. A telescopic mechanism is provided below the movable block, and the telescopic mechanism is installed on the bed frame, and the telescopic end of the telescopic mechanism is hinged to the bottom of the movable block;

[0044] A fence is provided around the bed board, the fence being mounted on the bed frame, the monitoring mechanism comprising a ring groove, a camera, and a moving member, the ring groove being provided at the top of the fence, the camera being mounted on the moving member, the moving member being provided in the ring groove and being movable along the ring groove, the camera facing the inside of the fence, and the output end of the camera being connected to the human-computer interaction module and the remote terminal;

[0045] The human-computer interaction module and the identification module are both installed on the fence, the human-computer interaction module is used to collect clinical data of the patient, and the output end of the human-computer interaction module is connected to the input end of the identification module;

[0046] The identification module executes the method of the present invention and outputs the patient's identification result to the human-computer interaction module and the remote terminal for display.

[0047] The bed board of the device is divided into board one and board two, and board one and board two are hinged. In this way, the telescopic mechanism is activated to control the corresponding board one and board two to rise or fall. Board one and board two can rise or fall at the same time to adjust the overall height of the bed board.

[0048] You can also independently control the swing of board one or board two around the hinge, adjusting the angle between boards one and two to meet the patient's needs, such as sit-stand support, reclining, head-high-feet-low position (to prevent intracranial hemorrhage), etc. The angle and height of the bed board can be adjusted to meet different usage needs.

[0049] Fences are set up on all four sides of the bed to facilitate protection and improve safety. Ring grooves are set on the fences. The camera moves within the ring grooves based on the moving parts to collect circumferential images of the patient on the bed, which has a more comprehensive image acquisition range and better monitoring effect.

[0050] The human-computer interaction module and the identification module are combined to obtain the patient's identification results for easy viewing.

[0051] Furthermore, the monitoring mechanism further includes a pressure sensor, a vibration sensor, a color sensor, a pressure comparator, and a vibration comparator;

[0052] The outer sides of the fence and the bed board are provided with an elastic protective layer, which is divided into several partitions, each of which is equipped with a pressure sensor, a vibration sensor, and a color sensor;

[0053] The first input end of the pressure comparator is connected to the output end of the pressure sensor, the second input end of the pressure comparator is connected to the pressure threshold memory, the output end of the pressure comparator is connected to the pressure threshold alarm, and the pressure threshold alarm is installed on the bed frame and / or the remote terminal;

[0054] A first input end of the vibration comparator is connected to an output end of the vibration sensor, a second input end of the vibration comparator is connected to a vibration threshold memory, an output end of the vibration comparator is connected to a vibration threshold alarm, and the vibration threshold alarm is mounted on the bed frame and / or the remote terminal;

[0055] The output ends of all color sensors are connected to a parallel counter, the output end of the parallel counter is connected to a first input end of a numerical comparator, the second input end of the numerical comparator is connected to a numerical memory, and the output end of the numerical comparator is connected to a bleeding alarm, which is installed on the bed frame and / or the remote terminal.

[0056] A pressure sensor detects the pressure applied by the patient to the elastic protective layer and transmits it to a pressure comparator. The pressure comparator compares the collected pressure signal with the pressure threshold stored in a pressure threshold memory. If the collected pressure signal exceeds the pressure threshold, it indicates that the patient is applying excessive force to the corresponding area of ​​the elastic protective layer, which could easily cause skin damage. At this point, the pressure comparator outputs a control signal to the pressure threshold alarm, which sounds an alarm to alert the patient and medical staff.

[0057] Similarly, the vibration sensor collects vibration information applied by the patient to the elastic protective layer and transmits it to a vibration comparator. The vibration comparator compares the collected vibration signal with the vibration threshold stored in the vibration threshold memory. If the collected vibration signal value is greater than the vibration threshold, it indicates that the patient's movement in the corresponding area of ​​the elastic protective layer is too intense, which may cause skin damage. At this time, the vibration comparator outputs a control signal to the vibration threshold alarm, which sounds an alarm signal to alert the patient and medical staff.

[0058] Furthermore, it also includes a fence lifting mechanism, which includes a piston cylinder, a lifter and an airbag;

[0059] The piston cylinder is arranged below the fence and is mounted on the bed frame. A horizontal piston plate is sealingly and slidingly connected in the piston cylinder. Gas is interposed between the piston plate and the top of the piston cylinder. A vertical piston rod is connected to the bottom of the piston plate.

[0060] The lift is mounted on the bed frame, the lift end of the lift is connected to the bottom of the fence, and the piston rod is connected to the lift end of the lift via a connecting rod;

[0061] The air bag is installed on the inner side of the top of the fence, and the air bag is communicated with the top of the piston cylinder.

[0062] Use the elevator to control the lifting and lowering displacement of the fence and adjust the height of the fence relative to the bed board to meet different usage requirements.

[0063] During the lift, the lifting end of the lift, via a connecting rod, drives the piston rod upward or downward. This upward movement of the piston rod drives the piston plate. As the piston plate moves upward, it pushes the gas between the piston plate and the top of the piston cylinder into the airbag, causing it to expand. This causes the fence to move upward, and the expansion of the airbag enhances the fence's protective effect. The further the piston plate moves upward, the more gas enters the airbag, the greater the expansion, and the better the fall and shock protection.

[0064] When the piston plate moves downward, the space between the piston plate and the top of the piston cylinder increases to form negative pressure, which draws the gas in the airbag into the piston cylinder, and the airbag gradually shrinks and resets. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] Figure 1 This is a front view structural schematic diagram of a nursing device for children with primary immune thrombocytopenia according to the present invention;

[0066] Figure 2 This is a schematic top view of the structure of the nursing device for children with primary immune thrombocytopenia according to the present invention;

[0067] Figure 3The present invention is a schematic structural diagram of a piston cylinder and a lifter of a nursing device for children with primary immune thrombocytopenia.

[0068] The reference numerals in the drawings of the specification include: bed frame 1, bed board 2, monitoring mechanism 3, human-computer interaction module 4, identification module 5;

[0069] Plate 1 11, plate 2 12, moving groove 13, moving block 14, telescopic mechanism 15;

[0070] Fence 21 , annular groove 22 , camera 23 , moving part 24 , piston cylinder 25 , lifter 26 , airbag 27 , piston plate 28 , piston rod 29 , connecting rod 30 . DETAILED DESCRIPTION

[0071] The following describes embodiments of the present invention in detail. Examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended only to explain the present invention and are not to be construed as limiting the present invention.

[0072] In the description of the present invention, it should be understood that the terms "longitudinal", "transverse", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", etc., indicating the orientation or position relationship, are based on the orientation or position relationship shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention.

[0073] In the description of the present invention, unless otherwise specified and limited, it should be noted that the terms "installed", "connected" and "connected" should be understood in a broad sense. For example, it can be a mechanical connection or an electrical connection, or it can be the internal communication between two components. It can be a direct connection or an indirect connection through an intermediate medium. For ordinary technicians in this field, the specific meanings of the above terms can be understood according to the specific circumstances.

[0074] The present invention discloses a method for identifying primary immune thrombocytopenia in children, comprising the following steps:

[0075] Collect patients' clinical data and perform preprocessing;

[0076] Univariate analysis was performed on the pre-processed clinical data to screen primary parameters, as shown in Table 1:

[0077] Table 1 Primary parameters

[0078] index ITP (N=1270) AA (N=1834) AL (N=1388) SLE (N=1165) P-value Female (number of cases, %) 549(43.23％) 815(44.44％) 579(41.71％) 978(83.95％) <0.001 Age (years) 7(5,10) 9(7,13) 9(6,12) 14(12,16) <0.001 Hemoglobin (g / L) 114(103,124) 82(72,92) 87(75,103) 123(106,133) <0.001 Mean corpuscular volume (fL) 80.26±6.43 87.03±7.57 86.05±7.79 86.51±7.34 <0.001 Platelet count (10^9 / L) 22.00(10.00,39.00) 48.00(15.00,101.75) 65.00(28.00,143.50) 248.00(187.00,306.00) <0.001 Absolute lymphocyte count (10^9 / L) 3.01(1.90,4.49) 1.34(0.53,2.30) 2.99(1.38,5.67) 1.57(1.05,2.18) <0.001 Platelet / lymphocyte ratio 7.50(3.15,14.69) 37.20(14.33,92.36) 21.96(6.29,72.09) 149.44(101.56,227.01) <0.001 Absolute neutrophil count (10^9 / L) 2.64(1.48,4.58) 0.44(0.16,1.01) 1.05(0.38,2.82) 3.66(2.34,5.86) <0.001 Absolute monocyte count (10^9 / L) 0.31(0.21,0.45) 0.09(0.04,0.17) 0.21(0.08,0.52) 0.31(0.21,0.44) <0.001 Lactate dehydrogenase (U / L) 277(236.00,333.00) 217.00(175.00,273.98) 373(256.00,796.48) 227.00(193.00,283.00) <0.001 Creatinine (umol / L) 27.00(22.00,34.00) 32.90(25.00,41.88) 33.00(26.58,40.13) 45.00(37.00,55.00) <0.001 Aspartate aminotransferase (U / L) 38.35(30.13,50.23) 28.00(21.83,37.00) 33.00(24.00,51.00) 23.50(19,31) <0.001 Urea (mmol / L) 4.20(3.10,5.32) 4.85(3.74,6.10) 4.30(3.40,5.26) 4.42(3.50,6.07) <0.001 Uric acid (μmol / L) 252(209.00,302.08) 249.00(193,317.83) 290.90(227.00,385.25) 319.00(266.00,395.00) <0.001 Alkaline phosphatase (U / L) 199(159.20,239.23) 165.00(130.00,211.85) 142.2(111.00,187.03) 108.00(75.00,159.00) <0.001 Total protein (g / L) 68.47±6.99 68.48±7.38 66.39±7.27 66.42±9.29 <0.001 Total bilirubin (μmol / L) 6.60(4.50,9.68) 8.50(5.80,11.90) 6.60(4.40,9.83) 6.70(4.60,9.50) <0.001 Alanine aminotransferase (U / L) 24.00(16.00,35.58) 24.00(14.83,38.00) 20.60(13.00,35.00) 17.00(12.00,28.00) <0.001 Glutamyl transpeptidase (U / L) 13.00(10.00,18.00) 17.00(13.00,25.00) 16.00(11.48,29.00) 16.00(11.00,26.00) <0.001 Albumin (g / L) 43.67±3.79 41.98±4.47 40.79±5.09 40.52±7.51 <0.001 Globulin (g / L) 24.79±6.00 26.51±5.90 25.59±5.83 25.90±6.67 <0.001 Aspartate aminotransferase / alanine aminotransferase 1.57(1.23,2.10) 1.22(0.83,1.80) 1.64(1.08,2.37) 1.43(1.00,1.89) <0.001 Potassium ion (mmol / L) 4.13±0.47 4.10±0.53 4.08±0.48 4.08±0.49 0.04 Sodium ion (mmol / L) 138.54±2.50 137.63±3.03 138.09±2.78 140.04±2.89 <0.001 Chloride ion (mmol / L) 104.34±2.36 103.26±3.35 103.35±3.82 105.59±3.22 <0.001 Total calcium (mmol / L) 2.43±0.16 2.32±0.14 2.32±0.18 2.28±0.19 <0.001 Phosphorus (mmol / L) 1.69±0.26 1.49±0.29 1.57±0.33 1.46±0.31 <0.001 Magnesium (mmol / L) 0.87±0.07 0.80±0.12 0.87±0.09 0.81±0.09 <0.001 Large platelet ratio (%) 39.73±9.42 30.99±8.65 29.02±8.73 30.51±10.29 <0.001 White blood cell count (10^9 / L) 6.74(4.93,9.28) 2.37(1.23,4.00) 5.44(2.33,16.20) 5.92(4.23,8.23) <0.001 Red blood cell count (10^12 / L) 4.19(3.68,4.59) 2.90(2.50,3.37) 3.17(2.67,3.77) 4.24(3.65,4.64) <0.001 Hepatitis B core antibody (s / co) 0.20(0.07,2.21) 0.34(0.09,1.97) 0.10(0.06,0.32) 0.15(0.08,1.58) <0.001 Hepatitis B e antibody (s / co) 1.65(1.38,1.82) 1.70(1.44,1.87) 1.80(1.67,1.92) 1.68(1.51,1.80) <0.001 Hepatitis B e antigen (s / co) 0.38(0.14,0.43) 0.40(0.35,0.46) 0.41(0.36,0.47) 0.39(0.30,0.44) <0.001 Thrombin time (seconds) 17.81±2.13 16.49±3.39 17.08±2.56 17.44±3.41 <0.001 Prothrombin time (seconds) 11.21±0.99 11.87±1.44 12.34±1.80 10.88±1.36 <0.001 Fibrinogen (g / L) 2.36±0.79 3.51±1.48 3.12±1.47 2.87±1.14 <0.001 PT international normalized ratio 0.97±0.09 1.04±0.13 1.09±0.17 0.94±0.13 <0.001 Activated partial thromboplastin time (seconds) 28.99±4.32 28.82±5.04 28.93±4.76 28.26±5.32 0.001 Congenital heart disease (number of cases, %) 29(2.28％) 91(4.96％) 115(8.29％) 54(4.64％) <0.001 Acquired heart disease (number of cases, %) 32(2.52％) 159(8.67％) 235(16.93％) 138(11.85％) <0.001

[0079] Based on the machine learning model, the primary parameters were ranked by importance and important parameters were screened, including platelet count, absolute neutrophil value, hemoglobin, platelet / lymphocyte ratio (PLR), lactate dehydrogenase, age, white blood cell count, mean corpuscular volume, alkaline phosphatase, and absolute lymphocyte value. Preferably, the machine learning model used one of the random forest, support vector machine, extreme gradient boosting, naive Bayes, and artificial neural network machine learning models.

[0080] Construct a multi-classification logistic regression model, input important parameters into the multi-classification logistic regression model, and output the identification results.

[0081] Secondary screening using clinical data first obtains primary parameters, then uses machine learning models to screen important parameters, gradually streamlining the data to obtain data that is more conducive to classification and identification, making it easier to use. Based on a multi-classification logistic regression model, identification results are obtained, improving identification efficiency and accuracy, and facilitating early intervention and treatment.

[0082] In a preferred embodiment of the present invention, the patient's clinical data includes:

[0083] Demographic characteristics: gender, age;

[0084] Hematological indicators: hemoglobin, mean corpuscular volume, platelet count, absolute lymphocyte count, platelet / lymphocyte ratio, absolute neutrophil count, absolute monocyte count, large platelet ratio, white blood cell count, red blood cell count;

[0085] Biochemical indicators: lactate dehydrogenase, creatinine, aspartate aminotransferase, urea, uric acid, alkaline phosphatase, total protein, total bilirubin, alanine aminotransferase, glutamyl transpeptidase, albumin, globulin, aspartate / alanine ratio;

[0086] Electrolyte indicators: potassium ion, sodium ion, chloride ion, total calcium, phosphorus, magnesium;

[0087] Coagulation function indicators: thrombin time, prothrombin time, fibrinogen, PT international normalized ratio, activated partial thromboplastin time;

[0088] Other indicators: hepatitis B core antibody, hepatitis B e antibody, hepatitis B e antigen, congenital heart disease, and acquired heart disease.

[0089] In a preferred embodiment of the present invention, the method for preprocessing the patient's clinical data is:

[0090] Data cleaning: patients aged >18 years were excluded, and outliers outside the mean ± 3 times the standard deviation were removed;

[0091] Data normalization: standardize continuous variables;

[0092] Dataset division: The data were randomly divided into a modeling group and a validation group in a ratio of 7:3.

[0093] In a preferred embodiment of the present invention, the multi-classification logistic regression model is:

[0094] G_SLE=-10.6769+0.0259*platelet count-0.1895*neutrophil absolute value-0.0301*hemoglobin+0.0214*PLR+0.5741*age+0.0542*white blood cell count+0.0822*mean corpuscular volume-0.0126*alkaline phosphatase-0.4300*lymphocyte absolute value, as shown in Table 2:

[0095] Table 2 Logistic regression analysis of SLE (systemic lupus erythematosus) vs reference class (ITP)

[0096] variable Beta coefficient Standard error Wald statistic P-value Odds ratio (OR) OR 95% confidence interval (intercept) -10.6769 0.0136 -785.65 <2e-16 0 (0to 0) Platelet count 0.0259 0.002 13.18 <2e-16 1.03 (1.02 to 1.03) Absolute neutrophil count -0.1895 0.0295 -6.43 0 0.83 (0.78 to 0.88) Hemoglobin -0.0301 0.0041 -7.35 0 0.97 (0.96 to 0.98) PLR 0.0214 0.004 5.37 0 1.02 (1.01 to 1.03) Lactate dehydrogenase 0.0018 0.0004 4.09 0 1.00 (1.00 to 1.00) age 0.5741 0.0255 22.48 <2e-16 1.78 (1.69 to 1.87) White blood cell count 0.0542 0.0159 3.41 0.001 1.06 (1.02 to 1.09) mean corpuscular volume 0.0822 0.0061 13.56 <2e-16 1.09 (1.07 to 1.10) Alkaline phosphatase -0.0126 0.0011 -11.00 <2e-16 0.99 (0.99 to 0.99) Absolute lymphocyte count -0.4300 0.0663 -6.48 0 0.65 (0.57 to 0.74)

[0097] G_AL = -5.4630 + 0.0175 * platelet count - 0.2973 * absolute value of neutrophils - 0.0727 * hemoglobin + 0.0222 * PLR + 0.1996 * age + 0.0902 * white blood cell count + 0.1184 * mean corpuscular volume - 0.0071 * alkaline phosphatase - 0.0366 * absolute value of lymphocytes, as shown in Table 3:

[0098] Table 3 Logistic regression analysis of AL (acute leukemia) vs reference class (ITP)

[0099] variable Beta coefficient Standard error Wald statistic P-value Odds ratio (OR) OR 95% confidence interval (intercept) -5.4630 0.2789 -19.59 <2e-16 0 (0to 0.01) Platelet count 0.0175 0.0019 9.32 <2e-16 1.02 (1.01 to 1.02) Absolute neutrophil count -0.2973 0.0275 -10.82 <2e-16 0.74 (0.70 to 0.78) Hemoglobin -0.0727 0.0036 -20.45 <2e-16 0.93 (0.92 to 0.94) PLR 0.0222 0.004 5.61 0 1.02 (1.01 to 1.03) Lactate dehydrogenase 0.0030 0.0004 8.53 <2e-16 1.00 (1.00 to 1.00) age 0.1996 0.0193 10.32 <2e-16 1.22 (1.18 to 1.27) White blood cell count 0.0902 0.013 6.92 0 1.09 (1.07 to 1.12) mean corpuscular volume 0.1184 0.0052 22.57 <2e-16 1.13 (1.11 to 1.14) Alkaline phosphatase -0.0071 0.0009 -8.13 0 0.99 (0.99 to 0.99) Absolute lymphocyte count -0.0366 0.0335 -1.09 0.275 0.96 (0.90 to 1.03)

[0100] G_AA=-3.3477+0.0132*platelet count-0.4862*neutrophil absolute value-0.0852*hemoglobin+0.0228*PLR+0.1751*age+0.0418*white blood cell count+0.1292*mean corpuscular volume-0.3119*lymphocyte absolute value, as shown in Table 4:

[0101] Table 4 Logistic regression analysis of AA (aplastic anemia) vs reference category (ITP)

[0102]

[0103]

[0104] G_ITP=0 (control group).

[0105] Then, G_SLE, G_AL, G_AA, and G_ITP are substituted into the following formula to obtain the corresponding probabilities of the four diseases:

[0106] P_SLE=exp(G_SLE) / [exp(G_SLE)+exp(G_AL)+exp(G_AA)+exp(G_ITP)];

[0107] P_AL=exp(G_AL) / [exp(G_SLE)+exp(G_AL)+exp(G_AA)+exp(G_ITP)];

[0108] P_AA=exp(G_AA) / [exp(G_SLE)+exp(G_AL)+exp(G_AA)+exp(G_ITP)];

[0109] P_ITP=exp(G_ITP) / [exp(G_SLE)+exp(G_AL)+exp(G_AA)+exp(G_ITP)].

[0110] Among them, P_SLE, P_AL, P_AA, and P_ITP are the classification probabilities of the corresponding diseases;

[0111] Compare the sizes of P_SLE, P_AL, P_AA, and P_ITP and sort them. The disease category corresponding to the largest value is the identification result.

[0112] The present invention also provides a nursing device for children with primary immune thrombocytopenia, such as Figure 1 and Figure 2 As shown, it includes a bed frame 1, a bed board 2, a monitoring mechanism 3, a human-computer interaction module 4 and an identification module 5. The human-computer interaction module 4 can be a computer, an intelligent display screen, etc.

[0113] The bed board 2 includes a first board 11 and a second board 12. The adjacent sides of the first board 11 and the second board 12 are hingedly connected (e.g., by a rotating hinge). A movable groove 13 is provided at the bottom of each board 11 and the second board 12. A movable block 14 is provided within the movable groove 13 and moves along the movable groove 13. A telescopic mechanism 15 is provided below the movable block 14. The telescopic end of the telescopic mechanism 15 is hingedly connected (e.g., by a universal joint) to the bottom of the movable block 14. The telescopic mechanism 15 can utilize an existing pneumatic, electric, or hydraulic telescopic rod structure. The telescopic mechanism 15 is mounted (e.g., by welding, bonding, riveting, or clamping) to the bed frame 1.

[0114] A fence 21 is provided around the bed board 2 and mounted on the bed frame 1. The monitoring mechanism 3 includes an annular groove 22, a camera 23, and a movable member 24. The annular groove 22 is provided at the top of the fence 21, and the camera 23 is mounted on the movable member 24 via a universal pan / tilt head (riveted, clamped, bonded, etc.). The movable member 24 can be an existing device such as an electric mobile trolley. The movable member 24 is disposed within the annular groove 22 and can move along the annular groove 22. The camera 23 faces the inside of the fence 21. The output end of the camera 23 is electrically connected to the human-computer interaction module 4 and a remote terminal (such as a mobile phone, computer, etc.) via a wireless transmission module (such as WiFi, Bluetooth, 5G network, etc.).

[0115] The human-computer interaction module 4 and the identification module 5 are both installed (such as riveted, clamped, bonded, etc.) on the fence 21. The human-computer interaction module 4 is used to collect clinical data of the patient. The output end of the human-computer interaction module 4 is electrically connected to the input end of the identification module 5.

[0116] The identification module 5 executes the method of the present invention and outputs the patient's identification result to the human-computer interaction module 4 and the remote terminal for display.

[0117] The bed board 2 of this device is divided into board 1 1 and board 2 12, and board 1 11 and board 2 12 are hinged. In this way, the telescopic mechanism 15 is activated to control the corresponding board 1 11 and board 2 12 to rise or fall. Board 1 11 and board 2 12 can rise or fall at the same time to adjust the overall height of the bed board 2.

[0118] The first board 11 or the second board 12 can also be independently controlled to swing around the hinge, adjusting the angle between the first board 11 and the second board 12 to meet the patient's needs, such as sitting support, reclining, head-high and feet-low position (to prevent intracranial hemorrhage), etc. The angle and height of the bed board 2 can be adjusted to meet different usage needs.

[0119] Fences 21 are provided on all four sides of the bed 2 for protection and improved safety. Ring grooves 22 are provided on the fences 21. The camera moves within the ring grooves 22 based on the moving member 24 to capture circumferential images of the patient on the bed 2. This provides a more comprehensive image acquisition range and better monitoring effect.

[0120] The human-computer interaction module 4 and the identification module 5 are combined to obtain the patient's identification results for easy viewing.

[0121] In a preferred embodiment of the present invention, the monitoring mechanism 3 further includes a pressure sensor, a vibration sensor, a color sensor, a pressure comparator, and a vibration comparator. An elastic protective layer (e.g., sponge) is provided on the outside of the fence 21 and the bed board 2. The elastic protective layer is divided into several sections, each of which is equipped with a pressure sensor, a vibration sensor, and a color sensor.

[0122] The first input end of the pressure comparator is electrically connected to the output end of the pressure sensor, the second input end of the pressure comparator is electrically connected to the pressure threshold memory, the output end of the pressure comparator is electrically connected to the pressure threshold alarm, and the pressure threshold alarm is installed (such as riveted, clamped, bonded, etc.) on the bed frame 1 and / or the remote terminal.

[0123] The first input end of the vibration comparator is electrically connected to the output end of the vibration sensor, the second input end of the vibration comparator is electrically connected to the vibration threshold memory, the output end of the vibration comparator is electrically connected to the vibration threshold alarm, and the vibration threshold alarm is installed (such as riveted, clamped, bonded, welded, clamped, etc.) on the bed frame 1 and / or the remote terminal.

[0124] The output ends of all color sensors are connected to a parallel counter, the output end of the parallel counter is connected to a first input end of a numerical comparator, the second input end of the numerical comparator is connected to a numerical memory, and the output end of the numerical comparator is connected to a bleeding alarm, which is installed on the bed frame 1 and / or the remote terminal.

[0125] A pressure sensor detects the pressure applied by the patient to the elastic protective layer and transmits it to a pressure comparator. The pressure comparator compares the collected pressure signal with the pressure threshold stored in a pressure threshold memory. If the collected pressure signal exceeds the pressure threshold, it indicates that the patient is applying excessive force to the corresponding area of ​​the elastic protective layer, which could easily cause skin damage. At this point, the pressure comparator outputs a control signal to the pressure threshold alarm, which sounds an alarm to alert the patient and medical staff.

[0126] Similarly, the vibration sensor collects vibration information applied by the patient to the elastic protective layer and transmits it to a vibration comparator. The vibration comparator compares the collected vibration signal with the vibration threshold stored in the vibration threshold memory. If the collected vibration signal value is greater than the vibration threshold, it indicates that the patient's movement in the corresponding area of ​​the elastic protective layer is too intense, which may cause skin damage. At this time, the vibration comparator outputs a control signal to the vibration threshold alarm, which sounds an alarm signal to alert the patient and medical staff.

[0127] The output ends of all color sensors are connected to a parallel counter. The parallel counter (such as 74LS161) counts the number of high-level inputs and compares them with the numerical threshold through a numerical comparator (such as 74LS688). If multiple color sensors all collect red signals, they output multiple high levels. If the number of high-level inputs is greater than or equal to the numerical threshold (such as 3), the numerical comparator outputs a control signal to the bleeding alarm, which sends out a corresponding alarm signal, indicating that the patient is bleeding abnormally so that the patient himself or medical staff can deal with it in time.

[0128] In a preferred embodiment of the present invention, Figure 3 As shown, the child primary immune thrombocytopenia care device further includes a fence lifting mechanism, which includes a piston cylinder 25, a lifter 26 and an airbag 27.

[0129] The piston cylinder 25 is arranged below the fence 21. The piston cylinder 25 is installed (such as riveting, clamping, bonding, etc.) on the bed frame 1. A horizontal piston plate 28 is sealed and slidably connected inside the piston cylinder 25. Gas is placed between the piston plate 28 and the top of the piston cylinder 25. The bottom of the piston plate 28 is connected to a vertical piston rod 29.

[0130] The lifter 26 is mounted (e.g., by riveting, clamping, or bonding) on ​​the bed frame 1. The lifting end of the lifter 26 is connected to the bottom of the fence 21. The piston rod 29 is connected (e.g., by clamping, bonding, or welding) to the lifting end of the lifter 26 via a connecting rod 30. Preferably, the connecting rod 30 can have a bent structure, such as a Z-shaped structure, or an arc structure.

[0131] Airbag 27 is mounted (e.g., clipped or bonded) to the inside of the top of fence 21 and communicates with the top of piston cylinder 25. Airbag 27 can be a single annular airbag 27 or multiple strip-shaped airbags 27, as needed. The size of airbag 27 and piston cylinder 25 can be adjusted as needed, and an auxiliary inflation and deflation mechanism, such as a fan, can also be connected to airbag 27.

[0132] The lift 26 is used to control the lifting displacement of the fence 21 and adjust the height of the fence 21 relative to the bed board 2 to meet different usage requirements. Two or more lifts 26 can be set. If two lifts 26 are set, the lifts 26 are placed on both sides of the bottom of the fence 21.

[0133] During the lifting process, the lifting end of elevator 26 drives piston rod 29 upward or downward synchronously via connecting rod 30. The upward and downward movement of piston rod 29 drives piston plate 28 synchronously. As piston plate 28 moves upward, it pushes the gas between piston plate 28 and the top of piston cylinder 25 into airbag 27, causing it to expand. This causes fence 21 to move upward, and the expansion of airbag 27 enhances the protective effect of fence 21. Furthermore, the more upward piston plate 28 moves, the more gas enters airbag 27, and the greater the expansion of airbag 27, resulting in greater protection against falls and shocks.

[0134] When the piston plate 28 moves downward, the space between the piston plate 28 and the top of the piston cylinder 25 increases to form a negative pressure, which draws the gas in the airbag 27 into the piston cylinder 25, and the airbag 27 gradually shrinks and returns to its original position.

[0135] The present invention can provide a plurality of control buttons on the bed frame 1 to respectively control the start and stop and the running direction of the telescopic mechanism 15, the elevator 26 and other mechanisms. In the description of this specification, the reference terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" and the like mean that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0136] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to the embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the claims and their equivalents.

[0137] The experimental data and results are as follows:

[0138] Table 5 Partial data

[0139]

[0140] Table 6 Experimental results of some data

[0141]

Claims

1. A method for identifying primary immune thrombocytopenia in children, characterized in that: The steps include: Collect patients' clinical data and perform preprocessing; Univariate analysis was performed on the pre-processed clinical data to screen primary parameters; Based on the machine learning model, the primary parameters are ranked by importance and important parameters are selected; Construct a multi-classification logistic regression model, input important parameters into the multi-classification logistic regression model, and output the identification results.

2. The method for identifying primary immune thrombocytopenia in children according to claim 1, wherein: The patient's clinical data included: Demographic characteristics: gender, age; Hematological indicators: hemoglobin, mean corpuscular volume, platelet count, absolute lymphocyte count, platelet / lymphocyte ratio, absolute neutrophil count, absolute monocyte count, large platelet ratio, white blood cell count, red blood cell count; Biochemical indicators: lactate dehydrogenase, creatinine, aspartate aminotransferase, urea, uric acid, alkaline phosphatase, total protein, total bilirubin, alanine aminotransferase, glutamyl transpeptidase, albumin, globulin, aspartate / alanine ratio; Electrolyte indicators: potassium ion, sodium ion, chloride ion, total calcium, phosphorus, magnesium; Coagulation function indicators: thrombin time, prothrombin time, fibrinogen, PT international normalized ratio, activated partial thromboplastin time; Other indicators: hepatitis B core antibody, hepatitis B e antibody, hepatitis B e antigen, congenital heart disease, and acquired heart disease.

3. The method for identifying primary immune thrombocytopenia in children according to claim 1, wherein: The method for preprocessing the patient's clinical data is: Data cleaning: patients aged >18 years were excluded, and outliers outside the mean ± 3 times the standard deviation were removed; Data normalization: standardize continuous variables.

4. The method for identifying primary immune thrombocytopenia in children according to claim 1, wherein: The important parameters include platelet count, absolute neutrophil count, hemoglobin, platelet / lymphocyte ratio (PLR), lactate dehydrogenase, age, white blood cell count, mean corpuscular volume, alkaline phosphatase, and absolute lymphocyte count.

5. The method for identifying primary immune thrombocytopenia in children according to claim 1, wherein: The multi-classification logistic regression model is: G_SLE=-10.6769+0.0259*platelet count-0.1895*neutrophil absolute value-0.0301*hemoglobin+0.0214*PLR+0.5741*age+0.0542*white blood cell count+0.0822*mean corpuscular volume-0.0126*alkaline phosphatase-0.4300*lymphocyte absolute value; G_AL=-5.4630+0.0175*platelet count-0.2973*neutrophil absolute value-0.0727*hemoglobin+0.0222*PLR+0.1996*age+0.0902*white blood cell count+0.1184*mean corpuscular volume-0.0071*alkaline phosphatase-0.0366*lymphocyte absolute value; G_AA=-3.3477+0.0132*platelet count-0.4862*neutrophil absolute value-0.0852*hemoglobin+0.0228*PLR+0.1751*age+0.0418*white blood cell count+0.1292*mean corpuscular volume-0.3119*lymphocyte absolute value; G_ITP=0 (control group) Substituting G_SLE, G_AL, G_AA, and G_ITP into the formula, we can finally get the corresponding probabilities of the four diseases: P_SLE=exp(G_SLE) / [exp(G_SLE)+exp(G_AL)+exp(G_AA)+exp(G_ITP)]; P_AL=exp(G_AL) / [exp(G_SLE)+exp(G_AL)+exp(G_AA)+exp(G_ITP)]; P_AA=exp(G_AA) / [exp(G_SLE)+exp(G_AL)+exp(G_AA)+exp(G_ITP)]; P_ITP=exp(G_ITP) / [exp(G_SLE)+exp(G_AL)+exp(G_AA)+exp(G_ITP)]; Among them, P_SLE, P_AL, P_AA, and P_ITP are the classification probabilities of the corresponding diseases; Compare the sizes of P_SLE, P_AL, P_AA, and P_ITP and sort them. The disease category corresponding to the largest value is the identification result.

6. The method for identifying primary immune thrombocytopenia in children according to claim 1, wherein: The machine learning model adopts one of random forest, support vector machine, extreme gradient boosting, naive Bayes, and artificial neural network machine learning models.

7. A nursing device for children with primary immune thrombocytopenia, characterized in that: It includes bed frame, bed board, monitoring mechanism, human-computer interaction module and identification module; The bed board includes a first board and a second board, wherein the adjacent sides of the first board and the second board are hinged, and a movable groove is provided at the bottom of the first board and the second board, and a movable block is provided in the movable groove and moves along the movable groove. A telescopic mechanism is provided below the movable block, and the telescopic mechanism is installed on the bed frame, and the telescopic end of the telescopic mechanism is hinged to the bottom of the movable block; A fence is provided around the bed board, the fence being mounted on the bed frame, the monitoring mechanism comprising a ring groove, a camera, and a moving member, the ring groove being provided at the top of the fence, the camera being mounted on the moving member, the moving member being provided in the ring groove and being movable along the ring groove, the camera facing the inside of the fence, and the output end of the camera being connected to the human-computer interaction module and the remote terminal; The human-computer interaction module and the identification module are both installed on the fence, the human-computer interaction module is used to collect clinical data of the patient, and the output end of the human-computer interaction module is connected to the input end of the identification module; The identification module executes the method according to any one of claims 1 to 6, and outputs the patient's identification result to the human-computer interaction module and the remote terminal for display.

8. The child primary immune thrombocytopenia nursing device according to claim 7, characterized in that: The monitoring mechanism also includes a pressure sensor, a vibration sensor, a color sensor, a pressure comparator, and a vibration comparator; The outer sides of the fence and the bed board are provided with an elastic protective layer, which is divided into several partitions, each of which is equipped with a pressure sensor, a vibration sensor, and a color sensor; The first input end of the pressure comparator is connected to the output end of the pressure sensor, the second input end of the pressure comparator is connected to the pressure threshold memory, the output end of the pressure comparator is connected to the pressure threshold alarm, and the pressure threshold alarm is installed on the bed frame and / or the remote terminal; A first input end of the vibration comparator is connected to an output end of the vibration sensor, a second input end of the vibration comparator is connected to a vibration threshold memory, an output end of the vibration comparator is connected to a vibration threshold alarm, and the vibration threshold alarm is mounted on the bed frame and / or the remote terminal; The output ends of all color sensors are connected to a parallel counter, the output end of the parallel counter is connected to a first input end of a numerical comparator, the second input end of the numerical comparator is connected to a numerical memory, and the output end of the numerical comparator is connected to a bleeding alarm, which is installed on the bed frame and / or the remote terminal.

9. The child primary immune thrombocytopenia nursing device according to claim 7, characterized in that: Also included is a fence lifting mechanism, which includes a piston cylinder, a lifter, and an airbag; The piston cylinder is arranged below the fence and is mounted on the bed frame. A horizontal piston plate is sealingly and slidingly connected in the piston cylinder. Gas is interposed between the piston plate and the top of the piston cylinder. A vertical piston rod is connected to the bottom of the piston plate. The lift is mounted on the bed frame, the lift end of the lift is connected to the bottom of the fence, and the piston rod is connected to the lift end of the lift via a connecting rod; The air bag is installed on the inner side of the top of the fence, and the air bag is communicated with the top of the piston cylinder.

Images