TWI836884B - Artificial Intelligence Prediction Method for Uric Acid Stones - Google Patents
Artificial Intelligence Prediction Method for Uric Acid Stones Download PDFInfo
- Publication number
- TWI836884B TWI836884B TW112103294A TW112103294A TWI836884B TW I836884 B TWI836884 B TW I836884B TW 112103294 A TW112103294 A TW 112103294A TW 112103294 A TW112103294 A TW 112103294A TW I836884 B TWI836884 B TW I836884B
- Authority
- TW
- Taiwan
- Prior art keywords
- model
- uric acid
- artificial intelligence
- input variables
- history
- Prior art date
Links
- LEHOTFFKMJEONL-UHFFFAOYSA-N Uric Acid Chemical compound N1C(=O)NC(=O)C2=C1NC(=O)N2 LEHOTFFKMJEONL-UHFFFAOYSA-N 0.000 title claims abstract description 107
- TVWHNULVHGKJHS-UHFFFAOYSA-N Uric acid Natural products N1C(=O)NC(=O)C2NC(=O)NC21 TVWHNULVHGKJHS-UHFFFAOYSA-N 0.000 title claims abstract description 107
- 229940116269 uric acid Drugs 0.000 title claims abstract description 107
- 239000004575 stone Substances 0.000 title claims abstract description 77
- 238000013473 artificial intelligence Methods 0.000 title claims abstract description 57
- 238000000034 method Methods 0.000 title claims abstract description 31
- 238000010801 machine learning Methods 0.000 claims abstract description 68
- 208000000913 Kidney Calculi Diseases 0.000 claims abstract description 61
- 206010029148 Nephrolithiasis Diseases 0.000 claims abstract description 61
- 210000002700 urine Anatomy 0.000 claims abstract description 37
- 239000013078 crystal Substances 0.000 claims abstract description 31
- 230000024924 glomerular filtration Effects 0.000 claims abstract description 20
- 230000006870 function Effects 0.000 claims description 49
- DDRJAANPRJIHGJ-UHFFFAOYSA-N creatinine Chemical compound CN1CC(=O)NC1=N DDRJAANPRJIHGJ-UHFFFAOYSA-N 0.000 claims description 30
- 238000013528 artificial neural network Methods 0.000 claims description 25
- 238000004364 calculation method Methods 0.000 claims description 18
- 229940109239 creatinine Drugs 0.000 claims description 15
- 206010012601 diabetes mellitus Diseases 0.000 claims description 15
- 210000002966 serum Anatomy 0.000 claims description 15
- 201000004538 Bacteriuria Diseases 0.000 claims description 11
- 201000005569 Gout Diseases 0.000 claims description 11
- 206010020772 Hypertension Diseases 0.000 claims description 8
- 239000000203 mixture Substances 0.000 claims description 3
- 230000001434 glomerular Effects 0.000 claims 1
- 238000012795 verification Methods 0.000 description 22
- 238000010200 validation analysis Methods 0.000 description 18
- 230000004913 activation Effects 0.000 description 7
- 238000012549 training Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 238000010276 construction Methods 0.000 description 2
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 2
- 238000001914 filtration Methods 0.000 description 2
- 210000003734 kidney Anatomy 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 238000002636 symptomatic treatment Methods 0.000 description 2
- 238000004566 IR spectroscopy Methods 0.000 description 1
- 208000001145 Metabolic Syndrome Diseases 0.000 description 1
- 208000008589 Obesity Diseases 0.000 description 1
- 201000000690 abdominal obesity-metabolic syndrome Diseases 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000002591 computed tomography Methods 0.000 description 1
- 230000037213 diet Effects 0.000 description 1
- 235000005911 diet Nutrition 0.000 description 1
- 201000010099 disease Diseases 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000004750 isotope dilution mass spectroscopy Methods 0.000 description 1
- 208000017169 kidney disease Diseases 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 235000015097 nutrients Nutrition 0.000 description 1
- 235000020824 obesity Nutrition 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 230000005855 radiation Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 239000000243 solution Substances 0.000 description 1
- 238000001356 surgical procedure Methods 0.000 description 1
Images
Landscapes
- Investigating Or Analysing Biological Materials (AREA)
Abstract
一種人工智慧尿酸結石預測方法,包含:建立一人工智慧尿酸結石預測系統,該人工智慧尿酸結石預測系統包含一包括數個模型輸入變數及一模型輸出變數之機器學習預測模型,該等模型輸入變數包括一腎絲球過濾率估計值,及一尿液酸鹼值,該模型輸出變數表示腎結石成分為尿酸結晶的機率;將一罹患有腎結石之病患之數個分別對應於該等模型輸入變數的模型輸入值輸入至該機器學習預測模型;及該機器學習預測模型根據該等模型輸入值,運算出一相對應於該模型輸出變數之模型輸出值,該模型輸出值表示該病患之腎結石之成分為尿酸結晶的機率。 An artificial intelligence uric acid stone prediction method comprises: establishing an artificial intelligence uric acid stone prediction system, the artificial intelligence uric acid stone prediction system comprises a machine learning prediction model including a plurality of model input variables and a model output variable, the model input variables include a glomerular filtration rate estimation value and a urine acid-base value, the model output variable indicates the probability that the kidney stone component is uric acid crystals; inputting a plurality of model input values of a patient suffering from kidney stones corresponding to the model input variables respectively into the machine learning prediction model; and the machine learning prediction model calculates a model output value corresponding to the model output variable according to the model input values, the model output value indicates the probability that the patient's kidney stone component is uric acid crystals.
Description
本發明是有關於一種結石預測方法,特別是指一種人工智慧(AI)尿酸結石預測方法。 The present invention relates to a stone prediction method, in particular to an artificial intelligence (AI) uric acid stone prediction method.
尿酸腎結石大約佔所有腎結石之10%至15%。不同於其他類型腎結石,尿酸結石是與代謝綜合症及營養分配障礙(包括糖尿病及肥胖)的若干特徵有關。此外,不同於需要進行手術之其他類型腎結石,大部分尿酸結石只要以保守方式治療即可。因此,當知道病患罹患有腎結石後,在進行治療之前,重要的是能夠進一步分辨出腎結石是純尿酸結石,或是非尿酸結石。 Uric acid kidney stones account for approximately 10% to 15% of all kidney stones. Unlike other types of kidney stones, uric acid stones are associated with several features of metabolic syndrome and nutrient distribution disorders, including diabetes and obesity. In addition, unlike other types of kidney stones that require surgery, most uric acid stones can be treated conservatively. Therefore, when it is known that a patient has kidney stones, before treatment, it is important to further distinguish whether the kidney stones are pure uric acid stones or non-uric acid stones.
目前已有相關研究報告指出,雙能量電腦斷層(dual-energy computed tomography)掃描可用於在對於腎結石進行治療之前,準確分辨出腎結石是純尿酸結石,或是非尿酸結石。然而,雙能量電腦斷層掃描會產生相當高的輻射劑量,且其設備之費用較為高昂。故,有必要尋求解決之道。 At present, relevant research reports have pointed out that dual-energy computed tomography (DECT) scans can be used to accurately distinguish whether kidney stones are pure uric acid stones or non-uric acid stones before treating kidney stones. However, DECT scans produce a very high radiation dose and the cost of the equipment is relatively high. Therefore, it is necessary to find a solution.
因此,本發明的目的,即在提供一種人工智慧尿酸結石預測方法。 Therefore, the purpose of the present invention is to provide an artificial intelligence method for predicting uric acid stones.
於是,本發明人工智慧尿酸結石預測方法,包含:(A)建立一人工智慧尿酸結石預測系統,其中,該人工智慧尿酸結石預測系統包含一包括數個模型輸入變數及一模型輸出變數之機器學習預測模型,該等模型輸入變數包括一腎絲球過濾率估計值,及一尿液酸鹼值,該模型輸出變數表示腎結石成分為尿酸結晶的機率;(B)將一罹患有腎結石之病患之數個分別對應於該等模型輸入變數的模型輸入值輸入至該機器學習預測模型;及(C)該機器學習預測模型根據該等模型輸入值,運算出一相對應於該模型輸出變數之模型輸出值,該模型輸出值表示該病患之腎結石之成分為尿酸結晶的機率。 Therefore, the artificial intelligence uric acid stone prediction method of the present invention includes: (A) establishing an artificial intelligence uric acid stone prediction system, wherein the artificial intelligence uric acid stone prediction system includes a machine learning including several model input variables and a model output variable Prediction models. The input variables of these models include an estimated value of glomerular filtration rate and a urine pH value. The output variables of the model represent the probability that the component of kidney stones is uric acid crystals; (B) a patient suffering from kidney stones. A number of patient model input values corresponding to the model input variables are input to the machine learning prediction model; and (C) the machine learning prediction model calculates a corresponding model output based on the model input values. The model output value of the variable represents the probability that the patient's kidney stones are composed of uric acid crystals.
本發明的另一目的,即在提供另一種人工智慧尿酸結石預測方法。 Another purpose of the present invention is to provide another artificial intelligence uric acid stone prediction method.
於是,本發明人工智慧尿酸結石預測方法,包含:(a)建立一人工智慧尿酸結石預測系統,其中,該人工智慧尿酸結石預測系統包含數個系統輸入變數,及一包括數個模型輸入變數及一模型輸出變數之機器學習預測模型,該等系統輸入變數包括一血清肌酸酐濃度、一性別、一年齡,及一尿液酸鹼值,該等模型輸入變數至 少包括該尿液酸鹼值,該模型輸出變數表示腎結石成分為尿酸結晶的機率;(b)將一罹患有腎結石之病患之數個分別對應於該等系統輸入變數的系統輸入值輸入至該人工智慧尿酸結石預測系統;及(c)該機器學習預測模型根據該等系統輸入值,運算出一相對應於該模型輸出變數之模型輸出值,其中,該模型輸出值表示該病患之腎結石之成分為尿酸結晶的機率。 Therefore, the artificial intelligence uric acid stone prediction method of the present invention comprises: (a) establishing an artificial intelligence uric acid stone prediction system, wherein the artificial intelligence uric acid stone prediction system comprises a plurality of system input variables, and a machine learning prediction model comprising a plurality of model input variables and a model output variable, wherein the system input variables comprise a serum creatinine concentration, a gender, an age, and a urine acid-base value, wherein the model input variables at least comprise the urine acid-base value , the model output variable represents the probability that the kidney stone is composed of uric acid crystals; (b) a plurality of system input values of a patient suffering from kidney stones corresponding to the system input variables are input into the artificial intelligence uric acid stone prediction system; and (c) the machine learning prediction model calculates a model output value corresponding to the model output variable according to the system input values, wherein the model output value represents the probability that the patient's kidney stones are composed of uric acid crystals.
本發明的功效在於:(1)只要將罹患有腎結石之病患之2個容易獲得的臨床變數(腎絲球過濾率估計值,及尿液酸鹼值)做為該機器學習預測模型之模型輸入變數,即可運算出該病患之腎結石之成分為尿酸結晶的機率,供醫療人員判斷該病患之腎結石類型是否為尿酸結石,以即早對症治療下藥;(2)只要將罹患有腎結石之病患之血清肌酸酐濃度、性別、年齡及尿液酸鹼值做為系統輸入值輸入至該人工智慧尿酸結石預測系統,該機器學習預測模型即可輸出該模型輸出值,其表示腎結石成分為尿酸結晶的機率,可供醫療人員判斷該病患之腎結石類型是否為尿酸結石,以即早對症治療下藥。 The functions of the present invention are: (1) As long as two easily obtained clinical variables (estimated glomerular filtration rate and urine pH) of patients suffering from kidney stones are used as the machine learning prediction model By inputting variables into the model, the probability that the patient's kidney stones are composed of uric acid crystals can be calculated, allowing medical personnel to determine whether the patient's kidney stones are uric acid stones and to prescribe appropriate treatment as soon as possible; (2) as long as The serum creatinine concentration, gender, age and urine pH of patients with kidney stones are input into the artificial intelligence uric acid stone prediction system as system input values, and the machine learning prediction model can output the model output value. It indicates the probability that the kidney stone component is uric acid crystals, which can be used by medical personnel to determine whether the patient's kidney stone type is uric acid stone, so as to prescribe symptomatic treatment as soon as possible.
2:人工智慧尿酸結石預測系統 2: Artificial intelligence uric acid stone prediction system
21:腎絲球過濾率估計值(eGFR)運算單元 21: Estimated glomerular filtration rate (eGFR) calculation unit
22:機器學習預測模型 22: Machine learning prediction model
S91~S93:步驟 S91~S93: steps
本發明的其他的特徵及功效,將於參照圖式的實施方式中清楚地呈現,其中: 圖1是一方塊圖,說明用於實施本發明人工智慧(AI)尿酸結石預測方法之一些實施例之人工智慧尿酸結石預測系統,其中,該人工智慧尿酸結石預測系統包含一用於運算一腎絲球過濾率估計值(eGFR)之腎絲球過濾率估計值運算單元,及一機器學習預測模型;圖2是一方塊圖,說明用於實施本發明人工智慧尿酸結石預測方法之另一些實施例之人工智慧尿酸結石預測系統,其中,該人工智慧尿酸結石預測系統包含該腎絲球過濾率估計值運算單元;圖3是一方塊圖,說明未設置上述圖1、2中的該腎絲球過濾率估計值運算單元之變化例;圖4是一流程圖,說明本發明人工智慧尿酸結石預測方法之主要實施例;圖5是一曲線圖,說明在本發明人工智慧尿酸結石預測方法之第一實施例中,當數個模型輸入變數包括一腎絲球過濾率估計值,及一尿液酸鹼值時,該機器學習預測模型應用在A組驗證數據時所得的驗證接收者操作特徵(Receiver Operating Characteristic,ROC)曲線;圖6是一曲線圖,說明在該第一實施例中,該機器學習預測模型應用在B組驗證數據時所得的驗證ROC曲線;圖7是一曲線圖,說明在本發明人工智慧尿酸結石預測方法之第 二實施例中,當該等模型輸入變數包括該腎絲球過濾率估計值、該尿液酸鹼值,及一身體質量指數(BMI)時,該機器學習預測模型應用在A組驗證數據時所得的驗證ROC曲線;圖8是一曲線圖,說明在該第二實施例中,該機器學習預測模型應用在B組驗證數據時所得的驗證ROC曲線;圖9是一曲線圖,說明在本發明人工智慧尿酸結石預測方法之第三實施例中,當該等模型輸入變數包括一年齡、該腎絲球過濾率估計值、該尿液酸鹼值,及該身體質量指數時,該機器學習預測模型應用在A組驗證數據時所得的驗證ROC曲線;圖10是一曲線圖,說明在該第三實施例中,該機器學習預測模型應用在B組驗證數據時所得的驗證ROC曲線;圖11是一曲線圖,說明在本發明人工智慧尿酸結石預測方法之第四實施例中,當該等模型輸入變數包括一性別、該年齡、該腎絲球過濾率估計值、該尿液酸鹼值,及該身體質量指數時,該機器學習預測模型應用在A組驗證數據時所得的驗證ROC曲線;圖12是一曲線圖,說明在該第四實施例中,該機器學習預測模型應用在B組驗證數據時所得的驗證ROC曲線;圖13是一曲線圖,說明在本發明人工智慧尿酸結石預測方法之第五實施例中,當該等模型輸入變數包括該性別、該年齡、該腎絲球過濾率估計值、該尿液酸鹼值、該身體質量指數,及是否有糖尿 病病史時,該機器學習預測模型應用在A組驗證數據時所得的驗證ROC曲線;圖14是一曲線圖,說明在該第五實施例中,該機器學習預測模型應用在B組驗證數據時所得的驗證ROC曲線;圖15是一曲線圖,說明在本發明人工智慧尿酸結石預測方法之第六實施例中,當該等模型輸入變數包括該性別、該年齡、該腎絲球過濾率估計值、該尿液酸鹼值、該身體質量指數、是否有糖尿病病史、是否有痛風病史,及是否有菌尿症病史時,該機器學習預測模型應用在A組驗證數據時所得的驗證ROC曲線;圖16是一曲線圖,說明在該第六實施例中,該機器學習預測模型應用在B組驗證數據時所得的驗證ROC曲線;圖17是一曲線圖,說明在本發明人工智慧尿酸結石預測方法之第七實施例中,當該等模型輸入變數包括該性別、該年齡、該腎絲球過濾率估計值、該尿液酸鹼值、該身體質量指數、是否有糖尿病病史、是否有痛風病史、是否有菌尿症病史,及是否有高血壓病史時,該機器學習預測模型應用在A組驗證數據時所得的驗證ROC曲線;及圖18是一曲線圖,說明在該第七實施例中,該機器學習預測模型應用在B組驗證數據時所得的驗證ROC曲線。 Other features and effects of the present invention will be clearly presented in the embodiments with reference to the drawings, in which: Figure 1 is a block diagram illustrating an artificial intelligence (AI) uric acid stone prediction system for implementing some embodiments of the present invention's artificial intelligence (AI) uric acid stone prediction method, wherein the artificial intelligence uric acid stone prediction system includes a device for calculating a kidney Estimated glomerular filtration rate (eGFR) estimated glomerular filtration rate calculation unit, and a machine learning prediction model; Figure 2 is a block diagram illustrating other implementations of the artificial intelligence uric acid stone prediction method of the present invention For example, the artificial intelligence uric acid stone prediction system includes the renal filament estimated value calculation unit; Figure 3 is a block diagram illustrating that the renal filament in Figures 1 and 2 is not set Variations of the ball filtration rate estimation value calculation unit; Figure 4 is a flow chart illustrating the main embodiment of the artificial intelligence uric acid stone prediction method of the present invention; Figure 5 is a graph illustrating the artificial intelligence uric acid stone prediction method of the present invention. In the first embodiment, when several model input variables include an estimated value of glomerular filtration rate and a urine pH value, the machine learning prediction model applies the verification receiver operating characteristic obtained from the verification data of Group A. (Receiver Operating Characteristic, ROC) curve; Figure 6 is a graph illustrating the verification ROC curve obtained when the machine learning prediction model is applied to Group B verification data in the first embodiment; Figure 7 is a graph, It is explained that in the first step of the artificial intelligence uric acid stone prediction method of the present invention In the second embodiment, when the model input variables include the estimated glomerular filtration rate, the urine pH, and a body mass index (BMI), the machine learning prediction model is applied to the validation data of Group A. The obtained verification ROC curve; Figure 8 is a graph illustrating the verification ROC curve obtained when the machine learning prediction model is applied to Group B verification data in the second embodiment; Figure 9 is a graph illustrating the verification ROC curve obtained in this second embodiment. In the third embodiment of the invention of the artificial intelligence uric acid stone prediction method, when the model input variables include an age, the estimated glomerular filtration rate, the urine pH, and the body mass index, the machine learning The verification ROC curve obtained when the prediction model is applied to the verification data of Group A; Figure 10 is a graph illustrating the verification ROC curve obtained when the machine learning prediction model is applied to the verification data of Group B in the third embodiment; Figure 11 is a graph illustrating that in the fourth embodiment of the artificial intelligence uric acid stone prediction method of the present invention, when the model input variables include a gender, the age, the estimated glomerular filtration rate, and the urine acid-base value, and the body mass index, the verification ROC curve obtained when the machine learning prediction model is applied to group A verification data; Figure 12 is a graph illustrating that in the fourth embodiment, the machine learning prediction model is applied to The verification ROC curve obtained when verifying data in Group B; Figure 13 is a graph illustrating that in the fifth embodiment of the artificial intelligence uric acid stone prediction method of the present invention, when the model input variables include the gender, the age, the kidney Estimated silk ball filtration rate, urine pH, body mass index, and whether there is diabetes The validation ROC curve obtained when the machine learning prediction model is applied to the verification data of Group A when the disease history is determined; Figure 14 is a graph illustrating that in the fifth embodiment, the machine learning prediction model is applied to the verification data of Group B The obtained verification ROC curve; Figure 15 is a graph illustrating that in the sixth embodiment of the artificial intelligence uric acid stone prediction method of the present invention, when the model input variables include the gender, the age, and the glomerular filtration rate estimate value, the urine pH, the body mass index, whether there is a history of diabetes, whether there is a history of gout, and whether there is a history of bacteriuria, the validation ROC curve obtained when the machine learning prediction model is applied to the validation data of Group A ; Figure 16 is a graph illustrating the verification ROC curve obtained when the machine learning prediction model is applied to Group B verification data in the sixth embodiment; Figure 17 is a graph illustrating the artificial intelligence uric acid stones of the present invention In the seventh embodiment of the prediction method, when the model input variables include the gender, the age, the estimated glomerular filtration rate, the urine pH, the body mass index, whether there is a history of diabetes, whether there is The validation ROC curve obtained when the machine learning prediction model is applied to the validation data of Group A when there is a history of gout, a history of bacteriuria, and a history of hypertension; and Figure 18 is a graph illustrating the seventh implementation In this example, the validation ROC curve obtained when the machine learning prediction model is applied to Group B validation data.
參閱圖1、4,本發明人工智慧尿酸結石預測方法之第一實施例,包含如圖4流程圖所示的步驟。首先,如圖4步驟S91所示,建立如圖1所示之一人工智慧尿酸結石預測系統2。其中,該人工智慧尿酸結石預測系統2包含數個系統輸入變數、一用於運算一腎絲球過濾率估計值(eGFR)之腎絲球過濾率估計值(eGFR)運算單元21,以及一包括數個模型輸入變數及一模型輸出變數之機器學習預測模型22。其中,該模型輸出變數表示腎結石成分為尿酸結晶的機率。
Referring to Figures 1 and 4, the first embodiment of the artificial intelligence uric acid stone prediction method of the present invention includes the steps shown in the flowchart of Figure 4. First, as shown in step S91 of Figure 4, an artificial intelligence uric acid
需特別說明的是,雖然圖1中有繪示其中一個輸入變數x5是身體質量指數(BMI),但是該輸入變數x5是在第二至第七實施例(稍後將詳述)才會用到的輸入變數,因此,以下將在暫時忽略圖1中之該輸入變數x5的情況下詳加說明本第一實施例。亦即,在本第一實施例中,該等系統輸入變數僅包括x0=「血清肌酸酐濃度(簡稱Scr)」、x1=「性別」、x2=「年齡」,及x4=「尿液酸鹼值」。 It should be noted that although one of the input variables x 5 is shown in Figure 1 as body mass index (BMI), this input variable x 5 is only used in the second to seventh embodiments (which will be described in detail later). The input variable will be used. Therefore, the first embodiment will be described in detail below while temporarily ignoring the input variable x 5 in FIG. 1 . That is, in this first embodiment, the system input variables only include x 0 = "serum creatinine concentration (Scr)", x 1 = "gender", x 2 = "age", and x 4 = "Urine pH".
在本第一實施例中,該eGFR運算單元21是根據該等系統輸入變數x0=「血清肌酸酐濃度」、x1=「性別」,x2=「年齡」,利用以下關於腎病飲食的同位素稀釋質譜溯源改良(isotope dilution mass spectrometry traceable Modification of Diet in Renal Disease)之公式計算出其中一模型輸入變數x3=「eGFR」,
使得在本第一實施例中,該等模型輸入變數包括x3=「eGFR」,及x4=「尿液酸鹼值」:eGFR[mL/min/1.73m2]=175×(血清肌酸酐濃度)-1.154×(年齡)-0.203×[0.742若為女性]。
In the first embodiment, the
在本第一實施例中,該機器學習預測模型22之結構如下:
在本第一實施例中,該機器學習預測模型22之建構乃是根據罹患有腎結石之A組(Cohort A)病患數據,及罹患有腎結石之B組(Cohort B)病患數據等兩組病患數據。其中,A組病患數據被分為用於模型訓練之訓練集(60%),及用於模型驗證之驗證集(40%)。
In this first embodiment, the construction of the machine
其中,在該機器學習預測模型22之建置過程中,A組病
患數據包括1098個病患之數據,且該等1098個病患中的146個(13.3%)罹患有純尿酸結石,而B組病患數據包括71個病患之數據,且該等71個病患中的3個(4.23%)罹患有純尿酸結石。另外,附帶一提的是,可使用紅外光譜對於罹患有腎結石之A組病患及B組病患之手術取出的結石進行分析,以分析出腎結石是純尿酸結石,或非尿酸結石,其中,在本第一實施例中將純尿酸結石定義為1,且將非尿酸結石定義為0。
Among them, during the construction process of the machine
在本第一實施例中,透過計算A組病患數據之純尿酸結石患者與非尿酸結石患者的患病率所得之計算結果可顯示出,A組病患數據之訓練集與驗證集之間的統計差異非常小。而B組病患數據則僅用於驗證。 In this first embodiment, the calculation results obtained by calculating the prevalence of pure uric acid stone patients and non-uric acid stone patients in group A patient data show that the statistical difference between the training set and the validation set of group A patient data is very small. Group B patient data is only used for validation.
於是,根據A組病患數據之訓練集產生該機器學習預測模型22之後,可進一步使用接收者操作特徵(Receiver Operating Characteristic,ROC)曲線之曲線下面積(Area Under the ROC Curve,AUC)分析方式,基於約登指數(Youden’s index)來求取預測陰性與陽性之最佳門檻值,繼而根據所得之最佳門檻值,計算該機器學習預測模型22之靈敏度(sensitivity)、特異度(specificity)、陽性預測值(positive predictive value,PPV),及陰性預測值(negative predictive value,NPV),以評估該機器學習預測模型22之模型效能。然後,將該機器學習預測模型22
應用於A組病患數據之驗證集,及B組病患數據上,以驗證該機器學習預測模型22是否能夠正確識別出尿酸結石患者。
Therefore, after the machine
在本第一實施例中,由於該等模型輸入變數僅有x3=「eGFR」,及x4=「尿液酸鹼值」,故函數f(x)可簡化為下列形式:f(x)=w0+w3f3(x3)+w4f4(x4),其中,w0、w3、w4為常數參數,f3(x3)、f4(x4)函數則皆是以全連通之兩層神經網路所建構之非線性函數。該兩層神經網路的第一層之輸出維度是mi(i=3,4),其中,mi為1至100之間的整數,而第二層之輸出維度則是1。函數f3(x3)及f4(x4)分別為以下形式:fi(xi)=σi,2(Ai,2σi,1(Ai,1xi+Bi,1)+Bi,2),i=3,4。 In the first embodiment, since the model input variables are only x 3 = "eGFR" and x 4 = "urine acid-base value", the function f(x) can be simplified to the following form: f(x)=w 0 +w 3 f 3 (x 3 )+w 4 f 4 (x 4 ), where w 0 , w 3 , and w 4 are constant parameters, and the functions f 3 (x 3 ) and f 4 (x 4 ) are both nonlinear functions constructed by a fully connected two-layer neural network. The output dimension of the first layer of the two-layer neural network is mi (i=3,4), where mi is an integer between 1 and 100, and the output dimension of the second layer is 1. The functions f 3 (x 3 ) and f 4 (x 4 ) are of the following forms respectively: fi ( xi )=σi ,2 (Ai ,2σi ,1 (Ai ,1xi + Bi ,1 )+Bi ,2 ),i=3,4.
其中,Ai,1及Bi,1(i=3,4)皆是mi-by-1之向量變數,其每一個變數包括mi個可調參數。而Ai,2(i=3,4)則皆是1-by-mi之向量變數,其每一個變數也包括mi個可調參數。B3,2及B4,2則皆是純量變數。而σi,1及σi,2(i=3,4)則是建構神經網路時常使用的標準非線性激活函數。非線性函數f(x)共包括個可訓練模型參數。在本第一實施例中,所使用之mi(i=3,4)皆為20,且Bi,1(i=3,4)皆設為零,共85個可訓練模型參數。 Among them, A i,1 and B i,1 (i=3,4) are vector variables of m i -by-1, and each variable includes m i adjustable parameters. And A i,2 (i=3,4) are all 1-by- mi vector variables, and each variable also includes m i adjustable parameters. B 3,2 and B 4,2 are both scalar variables. σ i,1 and σ i,2 (i=3,4) are standard nonlinear activation functions commonly used in constructing neural networks. The nonlinear function f(x) includes trainable model parameters. In this first embodiment, the m i (i=3,4) used are all 20, and the Bi ,1 (i=3,4) are all set to zero, resulting in a total of 85 trainable model parameters.
參閱圖4、5、6,於是,當將本第一實施例之該步驟S91所建立的該機器學習預測模型22應用在A組病患數據之驗證集之
後,可計算出其曲線下面積(Area Under the ROC Curve,AUC)為0.8213,如圖5所示。同理,當將該機器學習預測模型22應用在B組病患數據之後,可計算出其曲線下面積(AUC)為0.8382,如圖6所示。
Referring to Figures 4, 5, and 6, when the machine
接著,如圖1及圖4之步驟S92所示,在本第一實施例中,若欲知曉一罹患有腎結石之病患之腎結石成分為尿酸結晶的機率,則需先將該病患之分別對應於該等系統輸入變數x0=「血清肌酸酐濃度、x1=「性別」、x2=「年齡」及x4=「尿液酸鹼值」之四個系統輸入值,輸入至該人工智慧尿酸結石預測系統2,於是該eGFR運算單元21根據該等系統輸入值x0、x1、x2,利用前述eGFR公式運算出該病患之eGFR,並將其做為該模型輸入值x3。
Next, as shown in step S92 of FIG. 1 and FIG. 4 , in the first embodiment, if one wishes to know the probability that the kidney stones of a patient suffering from kidney stones are composed of uric acid crystals, one must first input the patient's four system input values corresponding to the system input variables x 0 = "serum creatinine concentration", x 1 = "sex", x 2 = "age" and x 4 = "urine acid-base value" into the artificial intelligence uric acid
然後,如步驟S93所示,該機器學習預測模型22能夠根據該病患之該等模型輸入值x3=「eGFR」,及x4=「尿液酸鹼值」,運算出一相對應於該模型輸出變數y之模型輸出值,其能夠表示該病患之腎結石之成分為尿酸結晶的機率。
Then, as shown in step S93, the machine
參閱圖1、7、8,在本發明之一第二實施例中,該等模型輸入變數包括x3=「eGFR」、x4=「尿液酸鹼值」,及x5=「身體質量指數(BMI)」,故函數f(x)可簡化為下列形式:f(x)=w0+w3f3(x3)+w4f4(x4)+w5f5(x5),其中,w0、w3、w4、w5為常數參數,f3(x3)、f4(x4)、 f5(x5)函數則皆是以全連通之兩層神經網路所建構之非線性函數。該兩層神經網路的第一層之輸出維度是mi(i=3至5),其中,mi為1至100之間的整數,而第二層之輸出維度則是1。函數f3(x3)至f5(x5)分別為以下形式:fi(xi)=σi,2(Ai,2σi,1(Ai,1xi+Bi,1)+Bi,2),i=3,4,5。 Referring to Figures 1, 7, and 8, in a second embodiment of the present invention, the model input variables include x 3 = "eGFR", x 4 = "urine pH value", and x 5 = "body mass" Index (BMI)", so the function f(x) can be simplified to the following form: f(x)=w 0 +w 3 f 3 (x 3 )+w 4 f 4 (x 4 )+w 5 f 5 (x 5 ), where w 0 , w 3 , w 4 , and w 5 are constant parameters, and the functions f 3 (x 3 ), f 4 (x 4 ), and f 5 (x 5 ) are all fully connected two-layer Nonlinear functions constructed by neural networks. The output dimension of the first layer of the two-layer neural network is m i (i=3 to 5), where m i is an integer between 1 and 100, and the output dimension of the second layer is 1. The functions f 3 (x 3 ) to f 5 (x 5 ) respectively have the following forms: f i (x i )=σ i,2 (A i,2 σ i,1 (A i,1 x i +B i, 1 )+B i,2 ),i=3,4,5.
其中,Ai,1及Bi,1(i=3至5)皆是mi-by-1之向量變數,其每一個變數包括mi個可調參數。而Ai,2(i=3至5)則皆是1-by-mi之向量變數,其每一個變數也包括mi個可調參數。B3,2到B5,2則皆是純量變數。而σi,1及σi,2(i=3至5)則是建構神經網路時常使用的標準非線性激活函數。非線性函數f(x)共包括個可訓練模型參數。在本第二實施例中,所使用之mi(i=3至5)皆為20,且Bi,1(i=3至5)皆設為零,共127個可訓練模型參數。 Among them, A i,1 and B i,1 (i=3 to 5) are vector variables of m i -by-1, and each variable includes m i adjustable parameters. And A i,2 (i=3 to 5) are all 1-by-m i vector variables, each of which also includes m i adjustable parameters. B 3,2 to B 5,2 are all scalar variables. σ i,1 and σ i,2 (i=3 to 5) are standard nonlinear activation functions commonly used in constructing neural networks. The nonlinear function f(x) includes trainable model parameters. In this second embodiment, the m i (i=3 to 5) used are all 20, and the Bi ,1 (i=3 to 5) are all set to zero, resulting in a total of 127 trainable model parameters.
參閱圖4、7、8,於是,當將本第二實施例之該步驟S91所建立的該機器學習預測模型22應用在A組病患數據之驗證集之後,可計算出其曲線下面積(AUC)為0.8238,如圖7所示。同理,當將該機器學習預測模型22應用在B組病患數據之後,可計算出其曲線下面積(AUC)為0.848,如圖8所示。
Referring to Figures 4, 7, and 8, when the machine
接著,如圖1及圖4之步驟S92所示,在本第二實施例中,若欲知曉罹患有腎結石之病患之腎結石成分為尿酸結晶的機率,則先將該病患之分別對應於該等系統輸入變數x0=「血清肌酸酐濃
度、x1=「性別」、x2=「年齡」、x4=「尿液酸鹼值」及x5=「BMI」之五個系統輸入值,輸入至該人工智慧尿酸結石預測系統2,於是該eGFR運算單元21根據系統輸入值x0、x1、x2,利用前述eGFR公式運算出該病患之eGFR。
Next, as shown in step S92 of Figures 1 and 4, in this second embodiment, if one wants to know the probability that the kidney stones of a patient suffering from kidney stones are composed of uric acid crystals, the five system input values of the patient corresponding to the system input variables x0 = "serum creatinine concentration", x1 = "gender", x2 = "age", x4 = "urine acid-base value" and x5 = "BMI" are first input into the artificial intelligence uric acid
然後,如步驟S93所示,在本第二實施例中,該機器學習預測模型22能夠根據該病患之該等模型輸入值x3=「eGFR」、x4=「尿液酸鹼值」,及x5=「BMI」,運算出該模型輸出值y,其表示該病患之腎結石之成分為尿酸結晶的機率。
Then, as shown in step S93, in the second embodiment, the machine
參閱圖2,在本發明之一第三實施例中,該等模型輸入變數還包括x2=「年齡」。需特別說明的是,雖然圖2中有繪示該系統輸入變數x1「性別」也可做為模型輸入變數,且還繪示了輸入變數x6「是否有糖尿病病史」、x7「是否有痛風病史」、x8「是否有菌尿症病史」、x9「是否有高血壓病史」,但是模型輸入變數x1「性別」以及輸入變數x6~x9是在其他實施例(稍後將詳述)才會用到的輸入變數,因此,以下將在暫時忽略該等系統輸入變數x1及x6~x9之情況下詳加說明本第三實施例。亦即,在本第三實施例中,該等系統輸入變數包括x0=「血清肌酸酐濃度」、x1=「性別」、x2=「年齡」、x4=「尿液酸鹼值」,及x5=「BMI」,而該等模型輸入變數則包括x2=「年齡」、x3=「eGFR」、x4=「尿液酸鹼值」,及x5=「BMI」。 Referring to Figure 2, in a third embodiment of the present invention, the model input variables also include x 2 = "age". It should be noted that although the system input variable x 1 "gender" is shown in Figure 2, it can also be used as a model input variable, and the input variables x 6 "whether there is a history of diabetes" and x 7 "whether "Do you have a history of gout", x 8 "Do you have a history of bacteriuria", x 9 "Do you have a history of hypertension", but the model input variable x 1 "gender" and the input variables x 6 ~ x 9 are in other embodiments (later (will be described in detail later), therefore, the third embodiment will be described in detail below while temporarily ignoring the system input variables x 1 and x 6 ~x 9 . That is, in this third embodiment, the system input variables include x 0 = "serum creatinine concentration", x 1 = "gender", x 2 = "age", x 4 = "urine pH value" ”, and x 5 = “BMI”, and the model input variables include x 2 = “age”, x 3 = “eGFR”, x 4 = “urine pH”, and x 5 = “BMI” .
在本第三實施例中,函數f(x)可簡化為下列形式:f(x)=w0+w2f2(x2)+w3f3(x3)+w4f4(x4)+w5f5(x5),其中,w0、w2,...,w5為常數參數,f2(x2),...,f5(x5)函數則皆是以全連通之兩層神經網路所建構之非線性函數。該兩層神經網路的第一層之輸出維度是mi(i=2至5),其中,mi為1至100之間的整數,而第二層之輸出維度則是1。函數f2(x2)至f5(x5)分別為以下形式:fi(xi)=σi,2(Ai,2σi,1(Ai,1xi+Bi,1)+Bi,2),i=2,3,4,5。 In the third embodiment, the function f(x) can be simplified to the following form: f(x)=w 0 +w 2 f 2 (x 2 )+w 3 f 3 (x 3 )+w 4 f 4 (x 4 )+w 5 f 5 (x 5 ), where w 0 , w 2 ,...,w 5 are constant parameters, and the functions f 2 (x 2 ),...,f 5 (x 5 ) are all nonlinear functions constructed by a fully connected two-layer neural network. The output dimension of the first layer of the two-layer neural network is mi (i=2 to 5), where mi is an integer between 1 and 100, and the output dimension of the second layer is 1. Functions f 2 (x 2 ) to f 5 (x 5 ) are of the following forms respectively: fi ( xi )=σi ,2 (Ai , 2σi,1 (Ai ,1xi + Bi ,1 )+Bi ,2 ),i=2,3,4,5.
其中,Ai,1及Bi,1(i=2至5)皆是mi-by-1之向量變數,其每一個變數包括mi個可調參數。而Ai,2(i=2至5)則皆是1-by-mi之向量變數,其每一個變數也包括mi個可調參數。B2,2到B5,2則皆是純量變數。而σi,1及σi,2(i=2至5)則是建構神經網路時常使用的標準非線性激活函數。非線性函數f(x)共包括個可訓練模型參數。在本第三實施例中,所使用之mi(i=2至5)皆為20,且Bi,1(i=2至5)皆設為零,共169個可訓練模型參數。 Among them, Ai,1 and Bi,1 (i=2 to 5) are mi -by-1 vector variables, each of which includes mi adjustable parameters. Ai,2 (i=2 to 5) are 1-by- mi vector variables, each of which also includes mi adjustable parameters. B 2,2 to B 5,2 are all scalar variables. σ i,1 and σ i,2 (i=2 to 5) are standard nonlinear activation functions commonly used in constructing neural networks. The nonlinear function f(x) includes In the third embodiment, mi (i=2 to 5) used is 20, and Bi ,1 (i=2 to 5) is set to zero, for a total of 169 trainable model parameters.
參閱圖2、4、9、10,於是,當將本第三實施例之該步驟S91所建立的該機器學習預測模型22應用在A組病患數據之驗證集之後,可計算出其曲線下面積(AUC)為0.8253,如圖9所示。同理,當將該機器學習預測模型22應用在B組病患數據之後,可計算出其曲線下面積(AUC)為0.8529,如圖10所示。
Referring to Figures 2, 4, 9, and 10, when the machine
接著,如圖2及圖4之步驟S92所示,在本第三實施例中,若欲知曉罹患有腎結石之病患之腎結石成分為尿酸結晶的機率,則先將該病患之分別對應於該等系統輸入變數x0=「血清肌酸酐濃度」、x1=「性別」、x2=「年齡」、x4=「尿液酸鹼值」及x5=「BMI」之五個系統輸入值,輸入至該人工智慧尿酸結石預測系統2,於是該eGFR運算單元21根據系統輸入值x0、x1、x2,利用前述eGFR公式運算出該病患之eGFR。
Next, as shown in step S92 of FIG. 2 and FIG. 4 , in the third embodiment, if you want to know the probability that the kidney stone component of a patient suffering from kidney stones is uric acid crystals, first classify the patients. Corresponding to the system input variables x 0 = "serum creatinine concentration", x 1 = "sex", x 2 = "age", x 4 = "urine pH" and x 5 = "BMI" five Each system input value is input to the artificial intelligence uric acid
然後,如步驟S93所示,在本第三實施例中,該機器學習預測模型22根據該病患之該等模型輸入值x2=「年齡」、x3=「eGFR」、x4=「尿液酸鹼值」,及x5=「BMI」,運算出該模型輸出值y,其表示該病患之腎結石之成分為尿酸結晶的機率。
Then, as shown in step S93, in this third embodiment, the machine
參閱圖2,在本發明之一第四實施例中,該等模型輸入變數還包括x1=「性別」,但該等系統輸入變數及該等模型輸入變數則尚不需包括x6至x9。 Referring to Figure 2, in a fourth embodiment of the present invention, the model input variables also include x 1 = "gender", but the system input variables and the model input variables do not need to include x 6 to x 9 .
在本第四實施例中,函數f(x)可簡化為下列形式:f(x)=w0+w1f1(x2)+w2f2(x2)+w3f3(x3)+w4f4(x4)+w5f5(x5),其中,w0、w1,...,w5為常數參數,f1(x1),...,f5(x5)函數則皆是以全連通之兩層神經網路所建構之非線性函數。該兩層神經網路的第一層之輸出維度是mi(i=1至5),其中,mi為1至100之間 的整數,而第二層之輸出維度則是1。函數f1(x1)至f5(x5)分別為以下形式:fi(xi)=σi,2(Ai,2σi,1(Ai,1xi+Bi,1)+Bi,2),i=1,...,5。 In this fourth embodiment, the function f(x) can be simplified to the following form: f(x)=w 0 +w 1 f 1 (x 2 )+w 2 f 2 (x 2 )+w 3 f 3 ( x 3 )+w 4 f 4 (x 4 )+w 5 f 5 (x 5 ), where w 0 , w 1 ,...,w 5 are constant parameters, f 1 (x 1 ),... , f 5 (x 5 ) functions are all nonlinear functions constructed with fully connected two-layer neural networks. The output dimension of the first layer of the two-layer neural network is m i (i=1 to 5), where m i is an integer between 1 and 100, and the output dimension of the second layer is 1. The functions f 1 (x 1 ) to f 5 (x 5 ) respectively have the following forms: f i (x i )=σ i,2 (A i,2 σ i,1 (A i,1 x i +B i, 1 )+B i,2 ),i=1,...,5.
其中,Ai,1及Bi,1(i=1至5)皆是mi-by-1之向量變數,其每一個變數包括mi個可調參數。而Ai,2(i=1至5)則皆是1-by-mi之向量變數,其每一個變數也包括mi個可調參數。B1,2到B5,2則皆是純量變數。而σi,1及σi,2(i=1至5)則是建構神經網路時常使用的標準非線性激活函數。非線性函數f(x)共包括個可訓練模型參數。在本第四實施例中,所使用之mi(i=1至5)皆為20,且Bi,1(i=1至5)皆設為零,共211個可訓練模型參數。 Among them, Ai,1 and Bi,1 (i=1 to 5) are mi -by-1 vector variables, each of which includes mi adjustable parameters. Ai,2 (i=1 to 5) are 1-by- mi vector variables, each of which also includes mi adjustable parameters. B 1,2 to B 5,2 are all scalar variables. σ i,1 and σ i,2 (i=1 to 5) are standard nonlinear activation functions commonly used in constructing neural networks. The nonlinear function f(x) includes In the fourth embodiment, mi (i=1 to 5) used is 20, and Bi ,1 (i=1 to 5) is set to zero, for a total of 211 trainable model parameters.
參閱圖2、4、11、12,於是,當將本第四實施例之該步驟S91所建立的該機器學習預測模型22應用在A組病患數據之驗證集之後,可計算出其曲線下面積(AUC)為0.8259,如圖11所示。同理,當將該機器學習預測模型22應用在B組病患數據之後,可計算出其曲線下面積(AUC)為0.8627,如圖12所示。
Referring to Figures 2, 4, 11, and 12, when the machine
接著,如圖2及圖4之步驟S92所示,在本第四實施例中,若欲知曉罹患有腎結石之病患之腎結石成分為尿酸結晶的機率,則先將該病患之分別對應於該等系統輸入變數x0=「血清肌酸酐濃度」、x1=「性別」、x2=「年齡」、x4=「尿液酸鹼值」及x5=「BMI」之五個系統輸入值,輸入至該人工智慧尿酸結石預測系統2,於是
該eGFR運算單元21根據系統輸入值x0、x1、x2,利用前述eGFR公式運算出該病患之eGFR。
Next, as shown in step S92 of FIG. 2 and FIG. 4 , in the fourth embodiment, if you want to know the probability that the kidney stone component of a patient suffering from kidney stones is uric acid crystals, first classify the patients. Corresponding to the system input variables x 0 = "serum creatinine concentration", x 1 = "sex", x 2 = "age", x 4 = "urine pH" and x 5 = "BMI" five Each system input value is input to the artificial intelligence uric acid
然後,如步驟S93所示,在本第四實施例中,該機器學習預測模型22根據該病患之該等模型輸入值x1=「性別」、x2=「年齡」、x3=「eGFR」、x4=「尿液酸鹼值」,及x5=「BMI」,運算出該模型輸出值y,其表示該病患之腎結石之成分為尿酸結晶的機率。
Then, as shown in step S93, in the fourth embodiment, the machine
參閱圖2,在本發明之一第五實施例中,該等系統輸入變數及該等模型輸入變數皆還包括x6=「是否有糖尿病病史」,但該等系統輸入變數及該等模型輸入變數則尚不需包括x7至x9。 Referring to Figure 2, in a fifth embodiment of the present invention, the system input variables and the model input variables also include x 6 = "whether there is a history of diabetes", but the system input variables and the model input variables The variables do not yet need to include x 7 to x 9 .
在本第五實施例中,函數f(x)可簡化為下列形式:f(x)=w0+w1f1(x2)+w2f2(x2)+w3f3(x3)+w4f4(x4)+w5f5(x5)+w6f6(x6),其中,w0、w1,...,w6為常數參數,f1(x1),...,f6(x6)函數則皆是以全連通之兩層神經網路所建構之非線性函數。該兩層神經網路的第一層之輸出維度是mi(i=1至6),其中,mi為1至100之間的整數,而第二層之輸出維度則是1。函數f1(x1)至f6(x6)分別為以下形式:fi(xi)=σi,2(Ai,2σi,1(Ai,1xi+Bi,1)+Bi,2),i=1,...,6。 In this fifth embodiment, the function f(x) can be simplified to the following form: f(x)=w 0 +w 1 f 1 (x 2 )+w 2 f 2 (x 2 )+w 3 f 3 ( x 3 )+w 4 f 4 (x 4 )+w 5 f 5 (x 5 )+w 6 f 6 (x 6 ), where w 0 , w 1 ,...,w 6 are constant parameters, f 1 (x 1 ),...,f 6 (x 6 ) functions are all nonlinear functions constructed with fully connected two-layer neural networks. The output dimension of the first layer of the two-layer neural network is m i (i=1 to 6), where m i is an integer between 1 and 100, and the output dimension of the second layer is 1. The functions f 1 (x 1 ) to f 6 (x 6 ) respectively have the following forms: f i (x i )=σ i,2 (A i,2 σ i,1 (A i,1 x i +B i, 1 )+B i,2 ),i=1,...,6.
其中,Ai,1及Bi,1(i=1至6)皆是mi-by-1之向量變數,其 每一個變數包括mi個可調參數。而Ai,2(i=1至6)則皆是1-by-mi之向量變數,其每一個變數也包括mi個可調參數。B1,2到B6,2則皆是純量變數。而σi,1及σi,2(i=1至6)則是建構神經網路時常使用的標準非線性激活函數。非線性函數f(x)共包括個可訓練模型參數。在本第五實施例中,所使用之mi(i=1至6)皆為20,且Bi,1(i=1至6)皆設為零,共253個可訓練模型參數。 Among them, A i,1 and B i,1 (i=1 to 6) are vector variables of m i -by-1, and each variable includes m i adjustable parameters. And A i,2 (i=1 to 6) are all 1-by-m i vector variables, each of which also includes m i adjustable parameters. B 1,2 to B 6,2 are all scalar variables. σ i,1 and σ i,2 (i=1 to 6) are standard nonlinear activation functions commonly used in constructing neural networks. The nonlinear function f(x) includes trainable model parameters. In the fifth embodiment, the m i (i=1 to 6) used are all 20, and the Bi ,1 (i=1 to 6) are all set to zero, resulting in a total of 253 trainable model parameters.
參閱圖2、4、13、14,於是,當將本第五實施例之該步驟S91所建立的該機器學習預測模型22應用在A組病患數據之驗證集之後,可計算出其曲線下面積(AUC)為0.829,如圖13所示。同理,當將該機器學習預測模型22應用在B組病患數據之後,可計算出其曲線下面積(AUC)為0.9363,如圖14所示。
Referring to Figures 2, 4, 13, and 14, when the machine
接著,如圖2及圖4之步驟S92所示,在本第五實施例中,若欲知曉罹患有腎結石之病患之腎結石成分為尿酸結晶的機率,則先將該病患之分別對應於該等系統輸入變數x0=「血清肌酸酐濃度」、x1=「性別」、x2=「年齡」、x4=「尿液酸鹼值」、x5=「BMI」、x6=「是否有糖尿病病史」之六個系統輸入值,輸入至該人工智慧尿酸結石預測系統2,於是該eGFR運算單元21根據系統輸入值x0、x1、x2,利用前述eGFR公式運算出該病患之eGFR。
Next, as shown in step S92 of FIG. 2 and FIG. 4 , in the fifth embodiment, if you want to know the probability that the kidney stone component of a patient suffering from kidney stones is uric acid crystals, first classify the patients. Corresponding to the system input variables x 0 = "serum creatinine concentration", x 1 = "sex", x 2 = "age", x 4 = "urine pH", x 5 = "BMI", x The six system input values 6 = "whether there is a history of diabetes" are input into the artificial intelligence uric acid
然後,如步驟S93所示,在本第五實施例中,該機器學習預測模型22根據該病患之該等模型輸入值x1=「性別」、x2=「年
齡」、x3=「eGFR」、x4=「尿液酸鹼值」、x5=「BMI」、x6=「是否有糖尿病病史」,運算出該模型輸出值y,其表示該病患之腎結石之成分為尿酸結晶的機率。
Then, as shown in step S93, in the fifth embodiment, the machine
參閱圖2,在本發明之一第六實施例中,該等系統輸入變數及該等模型輸入變數皆還包括x7=「是否有痛風病史」及x8=「是否有菌尿症病史」,但是該等系統輸入變數及該等模型輸入變數則尚不需包括x9。 Referring to Figure 2, in a sixth embodiment of the present invention, the system input variables and the model input variables also include x 7 = "whether there is a history of gout" and x 8 = "whether there is a history of bacteriuria" , but the system input variables and the model input variables do not need to include x 9 yet.
在本第六實施例中,函數f(x)可簡化為下列形式:f(x)=w0+w1f1(x2)+w2f2(x2)+w3f3(x3)+w4f4(x4)+w5f5(x5)+w6f6(x6,x7,x8),其中,w0、w1,...,w6為常數參數,f1(x1),...,f6(x6,x7,x8)函數則皆是以全連通之兩層神經網路所建構之非線性函數。該兩層神經網路的第一層之輸出維度是mi(i=1至6),其中,mi為1至100之間的整數,而第二層之輸出維度則是1。函數f1至f6分別為以下形式:fi(xi)=σi,2(Ai,2σi,1(Ai,1xi+Bi,1)+Bi,2),i=1,...,5;f6(x6,x7,x8)=σ6,2(A6,2σ6,1(A6,1x6+A7,1x7+A8,1x8+B6,1)+B6,2), In the sixth embodiment, the function f(x) can be simplified to the following form: f(x)= w0 + w1f1 ( x2 ) +w2f2( x2)+w3f3(x3 ) + w4f4 ( x4 ) + w5f5 ( x5 ) + w6f6 ( x6 , x7 , x8 ), wherein w0 , w1 ,..., w6 are constant parameters, and the functions f1 ( x1 ),..., f6 ( x6 , x7 , x8 ) are all nonlinear functions constructed by a fully connected two-layer neural network. The output dimension of the first layer of the two-layer neural network is mi (i=1 to 6), where mi is an integer between 1 and 100, and the output dimension of the second layer is 1. The functions f1 to f6 are of the following forms: fi ( xi ) = σi ,2 (Ai ,2 σi ,1 (Ai ,1 xi +Bi ,1 ) + Bi ,2 ), i = 1, ..., 5; f6 ( x6 , x7 , x8 ) = σ6,2 ( A6,2 σ6,1 ( A6,1 x6 + A7,1 x7 + A8,1 x8 + B6,1 ) + B6,2 ),
其中,Ai,1及Bi,1(i=1至6)皆是mi-by-1之向量變數,其每一個變數包括mi個可調參數。向量變數A7,1與A8,1之維度與A6,1相同。而Ai,2(i=1至6)則皆是1-by-mi之向量變數,其每一個變數 也包括mi個可調參數。B1,2到B6,2則皆是純量變數。而σi,1及σi,2(i=1至6)則是建構神經網路時常使用的標準非線性激活函數。非線性函數f(x)共包括個可訓練模型參數。在本第六實施例中,所使用之mi(i=1至6)皆為20,且Bi,1(i=1至6)皆設為零,共293個可訓練模型參數。 Among them, A i,1 and B i,1 (i=1 to 6) are vector variables of m i -by-1, and each variable includes m i adjustable parameters. The vector variables A 7,1 and A 8,1 have the same dimensions as A 6,1 . And A i,2 (i=1 to 6) are all 1-by-m i vector variables, each of which also includes m i adjustable parameters. B 1,2 to B 6,2 are all scalar variables. σ i,1 and σ i,2 (i=1 to 6) are standard nonlinear activation functions commonly used in constructing neural networks. The nonlinear function f(x) includes trainable model parameters. In this sixth embodiment, the m i (i=1 to 6) used are all 20, and the Bi ,1 (i=1 to 6) are all set to zero, resulting in a total of 293 trainable model parameters.
參閱圖2、4、15、16,於是,當將本第六實施例之該步驟S91所建立的該機器學習預測模型22應用在A組病患數據之驗證集之後,可計算出其曲線下面積(AUC)為0.8417,如圖15所示。同理,當將該機器學習預測模型22應用在B組病患數據之後,可計算出其曲線下面積(AUC)為0.9461,如圖16所示。
Referring to Figures 2, 4, 15, and 16, when the machine
接著,如圖2及圖4之步驟S92所示,在本第六實施例中,若欲知曉罹患有腎結石之病患之腎結石成分為尿酸結晶的機率,則先將該病患之分別對應於該等系統輸入變數x0=「血清肌酸酐濃度」、x1=「性別」、x2=「年齡」、x4=「尿液酸鹼值」、x5=「BMI」、x6=「是否有糖尿病病史」、x7=「是否有痛風病史」、x8=「是否有菌尿症病史」之八個系統輸入值,輸入至該人工智慧尿酸結石預測系統2,於是該eGFR運算單元21根據系統輸入值x0、x1、x2,利用前述eGFR公式運算出該病患之eGFR。
Next, as shown in step S92 of Figures 2 and 4, in the sixth embodiment, if one wants to know the probability that the kidney stones of a patient suffering from kidney stones are composed of uric acid crystals, the eight system input values of the patient corresponding to the system input variables x0 = "serum creatinine concentration", x1 = "gender", x2 = "age", x4 = "urine acid-base value", x5 = "BMI", x6 = "whether there is a history of diabetes", x7 = "whether there is a history of gout", x8 = "whether there is a history of bacteriuria" are first input into the artificial intelligence uric acid
然後,如步驟S93所示,在本第六實施例中,該機器學習預測模型22根據該病患之該等模型輸入值x1=「性別」、x2=「年
齡」、x3=「eGFR」、x4=「尿液酸鹼值」、x5=「BMI」、x6=「是否有糖尿病病史」、x7=「是否有痛風病史」、x8=「是否有菌尿症病史」,運算出該模型輸出值y,其表示該病患之腎結石之成分為尿酸結晶的機率。
Then, as shown in step S93, in this sixth embodiment, the machine
參閱圖2,在本發明之一第七實施例中,該等系統輸入變數及該等模型輸入變數皆還包括x9=「是否有高血壓病史」。 Referring to FIG. 2 , in a seventh embodiment of the present invention, the system input variables and the model input variables also include x 9 = "whether there is a history of hypertension".
在本第七實施例中,函數f(x)可簡化為下列形式:f(x)=w0+w1f1(x2)+w2f2(x2)+w3f3(x3)+w4f4(x4)+w5f5(x5)+w6(x9)f6(x6,x7,x8), 其中,w0、w1,...,w5為常數參數,f1(x1),...,f5(x5),f6(x6,x7,x8)函數則皆是以全連通之兩層神經網路所建構之非線性函數。該兩層神經網路的第一層之輸出維度是mi(i=1至6),其中,mi為1至100之間的整數,而第二層之輸出維度則是1。函數w6(x9)也是一個以全連通之兩層神經網路所建構之非線性函數,其第一層之輸出維度是m7,m7為1至100之間的整數,而第二層之輸出維度則是1。函數f1至f6分別為以下形式:fi(xi)=σi,2(Ai,2σi,1(Ai,1xi+Bi,1)+Bi,2),i=1,...,5,f6(x6,x7,x8)=σ6,2(A6,2σ6,1(A6,1x6+A7,1x7+A8,1x8+B6,1)+B6,2),w6(x9)=w6,0.σ7,2(A7,2σ7,1(A9,1x9+B7,1)+B7,2)。 In this seventh embodiment, the function f(x) can be simplified to the following form: f(x)=w 0 +w 1 f 1 (x 2 )+w 2 f 2 (x 2 )+w 3 f 3 ( x 3 )+w 4 f 4 (x 4 )+w 5 f 5 (x 5 )+w 6 (x 9 )f 6 (x 6 ,x 7 ,x 8 ), where w 0 , w 1 ,. ..,w 5 is a constant parameter, f 1 (x 1 ),...,f 5 (x 5 ),f 6 (x 6 ,x 7 ,x 8 ) functions are all based on fully connected two-layer neural Nonlinear functions constructed by the network. The output dimension of the first layer of the two-layer neural network is m i (i=1 to 6), where m i is an integer between 1 and 100, and the output dimension of the second layer is 1. Function w 6 (x 9 ) is also a nonlinear function constructed with a fully connected two-layer neural network. The output dimension of the first layer is m 7 , m 7 is an integer between 1 and 100, and the output dimension of the second layer The output dimension of the layer is 1. The functions f 1 to f 6 respectively have the following forms: f i (x i )=σ i,2 (A i,2 σ i,1 (A i,1 x i +B i,1 )+B i,2 ) ,i=1,...,5,f 6 (x 6 ,x 7 ,x 8 )=σ 6,2 (A 6,2 σ 6,1 (A 6,1 x 6 +A 7,1 x 7 +A 8,1 x 8 +B 6,1 )+B 6,2 ), w 6 (x 9 )=w 6,0 . σ 7,2 (A 7,2 σ 7,1 (A 9,1 x 9 +B 7,1 )+B 7,2 ).
其中,Ai,1及Bi,1(i=1至6)皆是mi-by-1之向量變數,其 每一個變數包括mi個可調參數。向量變數A7,1與A8,1之維度與A6,1相同。向量變數A9,1與B7,1之維度則是m7-by-1,其每一個變數包括m7個可調參數。而Ai,2(i=1至7)則皆是1-by-mi之向量變數,其每一個變數也包括mi個可調參數。B1,2到B7,2以及w6,0則皆是純量變數。而σi,1及σi,2(i=1至7)則是建構神經網路時常使用的標準非線性激活函數。非線性函數f(x)共包括個可訓練模型參數。在本第七實施例中,所使用之mi(i=1至6)皆為20,m7為5,且Bi,1(i=1至7)皆設為零,共304個可訓練模型參數。 Among them, Ai,1 and Bi,1 (i=1 to 6) are both mi -by-1 vector variables, each of which includes mi adjustable parameters. The dimensions of vector variables A 7,1 and A 8,1 are the same as A 6,1 . The dimensions of vector variables A 9,1 and B 7,1 are m 7 -by-1, each of which includes m 7 adjustable parameters. Ai,2 (i=1 to 7) are all 1-by- mi vector variables, each of which also includes mi adjustable parameters. B 1,2 to B 7,2 and w 6,0 are all scalar variables. σ i,1 and σ i,2 (i=1 to 7) are standard nonlinear activation functions commonly used in constructing neural networks. The nonlinear function f(x) includes In the seventh embodiment, mi (i=1 to 6) is 20, m7 is 5, and Bi ,1 (i=1 to 7) is set to zero, for a total of 304 trainable model parameters.
參閱圖2、4、17、18,於是,當將本第七實施例之該步驟S91所建立的該機器學習預測模型22應用在A組病患數據之驗證集之後,可計算出其曲線下面積(AUC)為0.8446,如圖17所示。同理,當將該機器學習預測模型22應用在B組病患數據之後,可計算出其曲線下面積(AUC)為0.951,如圖18所示。
Referring to Figures 2, 4, 17, and 18, when the machine
接著,如圖2及圖4之步驟S92所示,在本第七實施例中,若欲知曉罹患有腎結石之病患之腎結石成分為尿酸結晶的機率,則先將該病患之分別對應於該等系統輸入變數x0~x9,輸入至該人工智慧尿酸結石預測系統2,於是該eGFR運算單元21根據系統輸入值x0、x1、x2,利用前述eGFR公式運算出該病患之eGFR。
Next, as shown in step S92 of Figures 2 and 4, in the seventh embodiment, if one wants to know the probability that the kidney stones of a patient suffering from kidney stones are composed of uric acid crystals, the patient's corresponding system input variables x0 ~ x9 are first input into the artificial intelligence uric acid
然後,如步驟S93所示,在本第七實施例中,該機器學習預測模型22根據該病患之該等模型輸入值x1至x9,運算出該模型
輸出值y,其表示該病患之腎結石之成分為尿酸結晶的機率。
Then, as shown in step S93, in the seventh embodiment, the machine
參閱圖3之變化例,需特別提出的是,雖然在前述圖1、2實施例中,必須先利用該eGFR運算單元21從該等系統輸入變數「血清肌酸酐濃度」、「性別」及「年齡」來計算出該模型輸入變數「eGFR」,但是在本發明各實施例之變化例中,也可以不需在該人工智慧尿酸結石預測系統2中設置該eGFR運算單元21,而是該等系統輸入變數及該等模型輸入變數可完全相同,且皆至少包括該eGFR及該尿液酸鹼值,例如,如圖3所示,該等系統輸入變數及該等模型輸入變數都包括x1=「性別」、x2=「年齡」、x3=「eGFR」、x4=「尿液酸鹼值」、x5=「BMI」、x6=「是否有糖尿病病史」、x7=「是否有痛風病史」、x8=「是否有菌尿症病史」,及x9=「是否有高血壓病史」。
Referring to the variation example of Figure 3, it should be noted that although in the aforementioned embodiments of Figures 1 and 2, the
綜上所述,本發明人工智慧尿酸結石預測方法之優點與功效在於,基於圖5至18所呈現的高AUC,可知該機器學習預測模型22具有準確度很高的預測能力,故只要將罹患有腎結石之病患之2至9個容易獲得的臨床變數做為該機器學習預測模型22之模型輸入變數,讓該機器學習預測模型22運算出該病患之腎結石之成分為尿酸結晶的機率y,該機率y即可供醫療人員判斷該病患之腎結石類型是否為尿酸結石,以即早對症治療下藥;所以確實能達成本發明的目的。
In summary, the advantages and effectiveness of the artificial intelligence uric acid stone prediction method of the present invention are that, based on the high AUC shown in Figures 5 to 18, it can be seen that the machine
惟以上所述者,僅為本發明的實施例而已,當不能以此限定本發明實施的範圍,凡是依本發明申請專利範圍及專利說明書內容所作的簡單的等效變化與修飾,皆仍屬本發明專利涵蓋的範圍內。 However, the above are only examples of the present invention and should not be used to limit the scope of the present invention. All simple equivalent changes and modifications made based on the patent scope of the present invention and the content of the patent specification are still within the scope of the present invention. within the scope covered by the patent of this invention.
S91~S93:步驟 S91~S93: Steps
Claims (10)
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
TW112103294A TWI836884B (en) | 2023-01-31 | 2023-01-31 | Artificial Intelligence Prediction Method for Uric Acid Stones |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
TW112103294A TWI836884B (en) | 2023-01-31 | 2023-01-31 | Artificial Intelligence Prediction Method for Uric Acid Stones |
Publications (2)
Publication Number | Publication Date |
---|---|
TWI836884B true TWI836884B (en) | 2024-03-21 |
TW202433493A TW202433493A (en) | 2024-08-16 |
Family
ID=91269893
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
TW112103294A TWI836884B (en) | 2023-01-31 | 2023-01-31 | Artificial Intelligence Prediction Method for Uric Acid Stones |
Country Status (1)
Country | Link |
---|---|
TW (1) | TWI836884B (en) |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109785976A (en) * | 2018-12-11 | 2019-05-21 | 青岛中科慧康科技有限公司 | A kind of goat based on Soft-Voting forecasting system by stages |
CN113962992A (en) * | 2021-12-21 | 2022-01-21 | 青岛大学附属医院 | Urinary calculus flat scanning CT image recognition system based on deep learning and training method |
TWI768288B (en) * | 2020-02-10 | 2022-06-21 | 中國醫藥大學 | Renal function assessment method, renal function assessment system and kidney care device |
CN115205213A (en) * | 2022-06-06 | 2022-10-18 | 四川大学华西医院 | System for predicting in-vivo urinary calculus components |
-
2023
- 2023-01-31 TW TW112103294A patent/TWI836884B/en active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109785976A (en) * | 2018-12-11 | 2019-05-21 | 青岛中科慧康科技有限公司 | A kind of goat based on Soft-Voting forecasting system by stages |
TWI768288B (en) * | 2020-02-10 | 2022-06-21 | 中國醫藥大學 | Renal function assessment method, renal function assessment system and kidney care device |
CN113962992A (en) * | 2021-12-21 | 2022-01-21 | 青岛大学附属医院 | Urinary calculus flat scanning CT image recognition system based on deep learning and training method |
CN115205213A (en) * | 2022-06-06 | 2022-10-18 | 四川大学华西医院 | System for predicting in-vivo urinary calculus components |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Lin et al. | Statistical methods in assessing agreement: Models, issues, and tools | |
CN111933281B (en) | Disease typing determination system, method, device and storage medium | |
Osman et al. | SOM-LWL method for identification of COVID-19 on chest X-rays | |
CN106339593A (en) | Kawasaki disease classification and prediction method based on medical data modeling | |
TWI726574B (en) | Model training method and electronic device | |
Thorwarth et al. | Machine learning of biomarkers and clinical observation to predict eosinophilic chronic rhinosinusitis: a pilot study | |
CN116579982A (en) | Pneumonia CT image segmentation method, device and equipment | |
Ngo et al. | Federated deep learning for the diagnosis of cerebellar ataxia: Privacy preservation and auto-crafted feature extractor | |
TWI836884B (en) | Artificial Intelligence Prediction Method for Uric Acid Stones | |
Cohen et al. | A computer aided medical classification system of COVID-19 ct lung scans using convolution neural networks | |
KR102200039B1 (en) | Method and apparatus for providing a prediction value for spontaneous ureter stone passage | |
TW202433493A (en) | Artificial intelligence uric acid stone prediction method | |
CN110010246A (en) | A kind of disease Intelligent Diagnosis Technology based on neural network and confidence interval | |
CN111148999A (en) | Method and computer program for predicting neonatal bilirubin levels | |
TWI768288B (en) | Renal function assessment method, renal function assessment system and kidney care device | |
Wang et al. | Prediction of target range of intact parathyroid hormone in hemodialysis patients with artificial neural network | |
Kang et al. | Evaluation of biomarkers for treatment selection using individual participant data from multiple clinical trials | |
TWI817795B (en) | Cancer progression discriminant method and system thereof | |
TWI826081B (en) | Cancer progression assessment method and system thereof | |
TWI853646B (en) | Drug Dosage Assessment Methods | |
Dasare et al. | Covid19 Infection Detection and Classification Using CNN On Chest X-ray Images | |
CN116504407B (en) | Branch occlusion risk prediction method and system for coronary left trunk bifurcation | |
Fauzi et al. | Classification of Cervical Cancer Images Using Deep Residual Network Architecture | |
Braun et al. | Extending Mendelian risk prediction models to handle misreported family history | |
CN116168850B (en) | Method and device for calculating characteristic probability of iodine uptake of thyroid cancer patient |