CN116978582A - Modeling method and prediction system of prostate cancer prediction model - Google Patents
Modeling method and prediction system of prostate cancer prediction model Download PDFInfo
- Publication number
- CN116978582A CN116978582A CN202310750568.4A CN202310750568A CN116978582A CN 116978582 A CN116978582 A CN 116978582A CN 202310750568 A CN202310750568 A CN 202310750568A CN 116978582 A CN116978582 A CN 116978582A
- Authority
- CN
- China
- Prior art keywords
- prostate cancer
- prediction
- risk
- score
- value
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 208000000236 Prostatic Neoplasms Diseases 0.000 title claims abstract description 97
- 206010060862 Prostate cancer Diseases 0.000 title claims abstract description 94
- 238000000034 method Methods 0.000 title claims abstract description 36
- 238000007477 logistic regression Methods 0.000 claims abstract description 33
- 238000012795 verification Methods 0.000 claims abstract description 10
- 238000012360 testing method Methods 0.000 claims abstract description 9
- 238000007689 inspection Methods 0.000 claims abstract description 8
- 238000011156 evaluation Methods 0.000 claims abstract description 5
- 210000002307 prostate Anatomy 0.000 claims description 19
- 108010072866 Prostate-Specific Antigen Proteins 0.000 claims description 15
- 210000002966 serum Anatomy 0.000 claims description 9
- 238000003384 imaging method Methods 0.000 claims description 8
- 238000012216 screening Methods 0.000 claims description 7
- 238000012545 processing Methods 0.000 claims description 6
- 230000009471 action Effects 0.000 claims description 4
- 230000007170 pathology Effects 0.000 claims description 4
- 102000007066 Prostate-Specific Antigen Human genes 0.000 claims 1
- 230000001419 dependent effect Effects 0.000 abstract description 5
- 238000010827 pathological analysis Methods 0.000 abstract description 3
- 238000003759 clinical diagnosis Methods 0.000 abstract 1
- 102100038358 Prostate-specific antigen Human genes 0.000 description 14
- 238000011088 calibration curve Methods 0.000 description 9
- 206010028980 Neoplasm Diseases 0.000 description 8
- 238000003745 diagnosis Methods 0.000 description 7
- 229920001481 poly(stearyl methacrylate) Polymers 0.000 description 7
- 238000001574 biopsy Methods 0.000 description 6
- 230000035945 sensitivity Effects 0.000 description 6
- 238000012549 training Methods 0.000 description 6
- 238000001514 detection method Methods 0.000 description 5
- 230000001575 pathological effect Effects 0.000 description 5
- 238000004364 calculation method Methods 0.000 description 3
- 210000004027 cell Anatomy 0.000 description 3
- 201000010099 disease Diseases 0.000 description 3
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 238000001356 surgical procedure Methods 0.000 description 3
- 238000010200 validation analysis Methods 0.000 description 3
- 102100041003 Glutamate carboxypeptidase 2 Human genes 0.000 description 2
- 101000892862 Homo sapiens Glutamate carboxypeptidase 2 Proteins 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 201000011510 cancer Diseases 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000036541 health Effects 0.000 description 2
- 230000003902 lesion Effects 0.000 description 2
- 238000002595 magnetic resonance imaging Methods 0.000 description 2
- 230000036210 malignancy Effects 0.000 description 2
- 230000002265 prevention Effects 0.000 description 2
- 230000005180 public health Effects 0.000 description 2
- 230000011218 segmentation Effects 0.000 description 2
- 208000024891 symptom Diseases 0.000 description 2
- 101100261173 Arabidopsis thaliana TPS7 gene Proteins 0.000 description 1
- 102000001554 Hemoglobins Human genes 0.000 description 1
- 108010054147 Hemoglobins Proteins 0.000 description 1
- 206010020772 Hypertension Diseases 0.000 description 1
- 238000012351 Integrated analysis Methods 0.000 description 1
- 208000007433 Lymphatic Metastasis Diseases 0.000 description 1
- 238000005481 NMR spectroscopy Methods 0.000 description 1
- 241000208125 Nicotiana Species 0.000 description 1
- 235000002637 Nicotiana tabacum Nutrition 0.000 description 1
- GLQOALGKMKUSBF-UHFFFAOYSA-N [amino(diphenyl)silyl]benzene Chemical compound C=1C=CC=CC=1[Si](C=1C=CC=CC=1)(N)C1=CC=CC=C1 GLQOALGKMKUSBF-UHFFFAOYSA-N 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000003556 assay Methods 0.000 description 1
- 239000000090 biomarker Substances 0.000 description 1
- 210000004369 blood Anatomy 0.000 description 1
- 239000008280 blood Substances 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 208000020832 chronic kidney disease Diseases 0.000 description 1
- 238000002591 computed tomography Methods 0.000 description 1
- 238000004590 computer program Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 206010012601 diabetes mellitus Diseases 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000000556 factor analysis Methods 0.000 description 1
- 230000010365 information processing Effects 0.000 description 1
- 201000010985 invasive ductal carcinoma Diseases 0.000 description 1
- 238000012417 linear regression Methods 0.000 description 1
- 238000013178 mathematical model Methods 0.000 description 1
- 238000011119 multifactor regression analysis Methods 0.000 description 1
- 238000000491 multivariate analysis Methods 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 210000000056 organ Anatomy 0.000 description 1
- 230000008520 organization Effects 0.000 description 1
- 238000004223 overdiagnosis Methods 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 201000011046 prostatic acinar adenocarcinoma Diseases 0.000 description 1
- 238000011002 quantification Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000000611 regression analysis Methods 0.000 description 1
- 230000001105 regulatory effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000004083 survival effect Effects 0.000 description 1
- 230000008685 targeting Effects 0.000 description 1
- 210000001519 tissue Anatomy 0.000 description 1
- 238000002604 ultrasonography Methods 0.000 description 1
- 210000002229 urogenital system Anatomy 0.000 description 1
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H70/00—ICT specially adapted for the handling or processing of medical references
- G16H70/60—ICT specially adapted for the handling or processing of medical references relating to pathologies
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/27—Regression, e.g. linear or logistic regression
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/70—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for mining of medical data, e.g. analysing previous cases of other patients
Abstract
The application discloses a modeling method and a prediction system of a prostate cancer prediction model, wherein the method comprises the following steps: s1, acquiring clinical examination and inspection data of a study object; s2, taking the clinically significant prostate cancer as a dependent variable through pathological diagnosis or not, and carrying out single-factor logistic regression analysis by utilizing the clinical examination and test data to obtain the clinically significant prostate cancer prediction factor; s3, performing multi-factor logistic regression analysis on the prediction factors to obtain independent prediction factors of the prostate cancer with clinical significance, and constructing a prediction model; s4, constructing an alignment chart according to independent prediction factors; and S5, performing performance evaluation and verification on the constructed prediction model. The application accurately predicts the risk of the patient suffering from the prostate cancer by quantifying the evaluation result of the prostate cancer, provides guidance for clinical diagnosis of the patient suffering from the prostate cancer, and has important clinical significance.
Description
Technical Field
The application relates to the technical field of biology, in particular to a modeling method and a prediction system of a prostate cancer prediction model.
Background
Prostate cancer (PCa) is one of the most common malignancies in the male genitourinary system, with its incidence located at position 2 among all malignancies in men worldwide, as counted by the World Health Organization (WHO) 2018 GLOBOCAN. Gleason grading is the most widely used grading system for histologically assessing prostate cancer at present, and Gleason score is the sum of Gleason grades of major and minor components (> 5%) of tumors. With reference to the PI-RADSv2.1 guideline, prostate cancer of clinical significance is defined as medium-high risk PCa with a Gleason score of 3+4 points, while Gleason score of 3+3 points is defined as inert PCa.
The diagnosis of prostate cancer is usually dependent on various routes of prostate cancer aspiration biopsy procedures, which are invasive examinations and are not suitable for screening of prostate cancer. Early identification of prostate cancer requires efficient screening means, and since early prostate cancer is generally free of typical symptoms, prostate cancer symptoms are also not useful as an early screening means. Currently, serum prostate specific antigen (prostate specific antigen, PSA) assays are the most widely used biomarker for screening PCa. However, the major drawbacks of PSA testing are low organ specificity, resulting in a high false positive rate, overdiagnosis and unnecessary punctures. The Chinese patent with publication number of CN105243283A reports in the diagnosis and treatment information collection system of prostate cancer: because the pure utilization of PSA and digital rectal examination can cause that some people suffer from the risk of prostate cancer to be underestimated, and the risk of another part of people to be overestimated, the prostate cancer diagnosis and treatment information acquisition system is provided, and comprises a patient basic information acquisition module, a blood detection information acquisition module, an image information acquisition module (comprising CT or MRI) and a puncture biopsy information acquisition module, the information collected by the modules is transmitted to a network cloud for summarization, and the information collected by the modules is subjected to parameter processing through an information processing module. The multiple parameters are integrated and multivariate analysis is provided through the computer-aided technology, so that the diagnosis process is more objective, and the diagnosis efficiency is improved.
Whereas prostate multiparameter magnetic resonance (mpMRI) has been widely accepted in the routine management of men suspected or diagnosed with prostate cancer. This is an effective image examination with considerable accuracy in detecting tumors and stage. However, mpMRI limits its clinical application due to relatively low specificity (75%), positive predictive value and reproducibility. Such as the application of the webpage link https:// www.sohu.com/a/437522721_328736 name to multiparameter magnetic resonance imaging in prostate cancer diagnosis is reported in the text. As another chinese patent with publication No. CN114022462a, the patent name is reported in methods, systems, processors and computer readable storage media for implementing multi-parameter nmr image lesion segmentation: the method and the system realize the focus segmentation of the multi-parameter nuclear magnetic resonance image based on the deep neural network model.
At present 68 Ga-PSMA PET-CT (positron emission computed tomography) is a relatively new molecular imaging means, and is also a powerful tool for detecting biochemical recurrence, lymph node metastasis and primary PCa. In contrast to the mpMRI method, the method comprises, 68 Ga-PSMA PET-CT has higher sensitivity. However, prostate Specific Membrane Antigen (PSMA) is not up-regulated in all PCa cells, which affects 68 Application effect of Ga-PSMA PET-CT.
The screening efficiency of the prostate cancer is improved, the high risk group of the prostate cancer can be identified early, and timely prostate puncture inspection and unnecessary puncture reduction are key points for suppressing the threat of the prostate cancer to human health. Recent studies have shown that 68 The combination of Ga-PSMA PET-CT and mpMRI can achieve higher PCa detection rates as reported in DOI papers 10.2214/AJR.18.19585 and 10.2967/jnumed.118.221010.
Although the above prior art mentions the use of various combinations of methods to improve the accuracy of tumor detection and identification, there is still a lack of effective basis 68 Ga-PSMA PET-CT and mpMRI in combination, laboratory examinations and other risk factors.
Disclosure of Invention
The application provides a modeling method and a prediction system of a prostate cancer prediction model, which are based on 68 Ga-PSMA PET-CT and mpMRI are combined to construct a prediction model, a prediction model and a prediction system for predicting the prostate cancer with clinical significance are obtained, and a logistic regression model nomogram can be constructed by using clinical examination and test data of a patient, and the nomogram can predict the risk of the prostate cancer with clinical significance of the patient.
In order to achieve the above object, one of the technical solutions of the present application is a modeling method of a prostate cancer prediction model, the method comprising the steps of:
s1, acquiring clinical examination and inspection data of a study object.
S2, taking the clinically significant prostate cancer as a dependent variable through pathological diagnosis or not, and carrying out single-factor logistic regression analysis by utilizing the clinical examination and test data to obtain the clinically significant prostate cancer prediction factor; the single-factor logistic regression analysis is to fit only one factor into the logistic regression model when constructing the logistic regression model, and to determine whether the distribution difference of the group mean OR the rate has statistical significance (P value), the estimated value of the partial regression coefficient (β), the estimated value of the effect (OR, RR value), and the like.
S3, performing multi-factor logistic regression analysis on the predicted factors to obtain independent predicted factors of the prostate cancer with clinical significance; and selecting a variable with a smaller P value obtained after single-factor logistic regression analysis, and carrying out multi-factor analysis by adopting a stepwise regression method. And a predictive model is constructed, and the prediction model is constructed,
logitP＝ln[P/(1-P)]＝-4.0359+1.6926*a+0.2341*b+c
where P is the likelihood of developing a clinically significant prostate cancer, a is the PSAD value, b is the SUVmax value, and c is the value corresponding to the PI-RADS score.
S4, constructing an alignment chart according to independent prediction factors; nomograms are a graphical representation of a clinical predictive model that computes a total score based on the numerical value of an individual's individual predictive variables, and then computes the risk or probability of occurrence of an event based on the total score.
And S5, performing performance evaluation and verification on the constructed prediction model.
Further, the clinical examination and test data in the step S1 include: basic data, hematology index, serum PSA concentration, imaging exam report, pathology exam report.
Further, the step S3 of screening to obtain independent prediction factors is as follows: prostate imaging report and data system (Prostate Imaging Reporting and Data System, PSAd), maximum standard uptake value (Standardized uptake value maximum, SUVmax), prostate specific antigen density (Prostate-specific antigen density, PI-RAD) S score.
Further, in the nomogram, the first action is a score scale, and the score range is 0-100 minutes; the second behavior PSAD ranges from 0 to 2.4, and the corresponding score ranges from 0 to 28.9; the third behavior PI-RADS score ranges from 1,2,3,4,5, 11.8 points if 4, 15.1 points if 5, or 0 points if not; the fourth behavior SUVmax ranges from 0 to 60, and the corresponding score ranges from 0 to 100 minutes; the fifth action total score ranges from 0 to 160 minutes; the sixth behavior has a clinically significant likelihood of developing prostate cancer ranging from 0.01 to 0.999.
Further, in the alignment chart, the second row to the fourth row are related to factors, and different factor states correspond to different scores of the scale; the total score of the fifth row is divided into the total score of each factor, the risk of the sixth row has a corresponding relation with the total score of the fifth row, and the risk is projected to the corresponding position according to different scores, namely the corresponding risk.
Further, the model comprises a risk value calculation formula, and is constructed by a prediction model
(logitP＝ln[P/(1-P)]＝-4.0359+1.6926*a+0.2341*b+c)
The risk value formula is calculated as follows,
wherein a is a PSAD value, b is a SUVmax value, c is a value corresponding to a PI-RADS score, and c is a value corresponding to the PI-RADS score as follows:
the second technical scheme of the application is a prostate cancer prediction system, which comprises a data acquisition module M1 and a risk prediction module M2, wherein:
the data processing module M1 is used for acquiring clinical examination and checking data of a user;
the risk prediction module M2 is configured to calculate whether the user is at risk of developing prostate cancer according to a pre-trained logistic regression model.
The prostate cancer prediction model with clinical significance constructed by the method and the system of the application utilizes the clinical examination and the inspection data of patients to construct a logistic regression model nomogram, which can well detect the high risk group of the prostate cancer, and realize early detection and early treatment of the prostate cancer by early prostate puncture of the high risk group; the low risk group of the prostate cancer is eliminated, so that unnecessary prostate puncture is avoided, the pain of patients is reduced, and a large amount of medical resources are saved. The risk prediction is carried out by the ordinary people through the device, so that the risk of the prostate cancer can be queried quickly and conveniently at any time and any place according to the actual situation of the people, and the cancer prevention consciousness is improved. In addition, the device is beneficial to medical institutions for targeted treatment according to results, has important public health significance and has important clinical significance.
Drawings
FIG. 1 is a flow chart of a method for modeling a predictive model of prostate cancer in accordance with the present application.
Fig. 2 is a structured alignment chart.
Fig. 3 is a constructed training set ROC curve.
Fig. 4 is a constructed training set calibration curve.
Fig. 5 is a graph of the constructed validation set ROC.
FIG. 6 is a calibration curve of the constructed validation set.
FIG. 7 is a block diagram of a predictive system of the present application.
Detailed Description
It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the application.
The technical solutions of the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings, and it is apparent that the described embodiments are only some embodiments of the present application, but not all embodiments, and all other embodiments obtained by those skilled in the art without making creative efforts based on the embodiments of the present application are within the protection scope of the present application.
The modeling method of the prostate cancer prediction model of the present embodiment, referring to fig. 1, includes the following steps:
s1: clinical examination and inspection data of the study object are obtained.
Prostate biopsy surgery is the most reliable means of diagnosing prostate cancer, and thus, in one embodiment, the subject of choice is the subject on which the prostate biopsy surgery was performed and completed prior to surgery 68 Ga-PSMA PET-CT and mpMRI.
In one embodiment, the subject is in need of excluding the following patients: patients with preoperative lack of mpMRI, PET-CT, serum PSA examination; clinical critical data such as pathology report, hematology index, operation record, admission record, etc.
In a specific embodiment, the accuracy of the prediction result is improved, and the clinical examination and test data obtained by the method comprises the following steps: basic data, hematology index, serum PSA concentration, imaging exam report, pathology exam report, wherein: the patient's basis includes: age, height, weight, history of tobacco and wine, family history and co-morbid condition of the patient;
the co-condition of the patient includes: hypertension, diabetes, chronic kidney disease;
the hematology index of the patient includes: red cell count, white cell count, and classification, platelet count, and hemoglobin content;
the serum PSA of the patient comprises: tPSA (serum total PSA), fPSA (serum free PSA) and PSAd (prostate specific antigen density);
the patient's imaging exam report includes: prostate size and PIRADS score obtained by mpMRI; prostate size and SUVmax obtained by PET-CT;
the report of the pathological examination of the patient includes: histological type, gleason score, tumor tissue quantification.
S2: and taking the clinically significant prostate cancer as a dependent variable through pathological diagnosis or not, and carrying out single-factor logistic regression analysis by utilizing the clinical examination and test data to obtain the clinically significant prostate cancer prediction factor.
The logistic regression model is a generalized linear regression analysis model and can be used in the field of automatic disease diagnosis, including the discussion of risk factors causing diseases and the prediction of the occurrence probability of the diseases according to the risk factors. The single-factor logistic regression analysis is to fit only one factor into the logistic regression model when constructing the logistic regression model, and to determine whether the distribution difference of the group mean OR the rate has statistical significance (P value), the estimated value of the partial regression coefficient (β), the estimated value of the effect (OR, RR value), and the like.
For suspected prostate cancer patients selected as subjects, whether or not pathologically diagnosed with clinically significant prostate cancer is used as a dependent variable in a logistic regression model.
In one embodiment, the prostate tissue may be obtained by a prostate biopsy procedure, and after a pathological examination, the subject may be divided into clinically significant groups of prostate cancers and non-clinically significant groups of prostate cancers (non-cancerous and non-clinically significant prostate cancers) based on the pathological examination.
In one embodiment, a patient's pathological examination report may be used to determine whether the patient has clinically significant prostate cancer: referring to PI-radsv2.1 guidelines, patients with a Gleason score of 7 or more (including 4+3=7, 3+4=7) belonging to prostate acinar adenocarcinoma and ductal adenocarcinoma in pathological examination are defined as clinically significant prostate cancer, while Gleason score of 3+3 is defined as inert PCa.
In a specific embodiment, the prostate biopsy procedure includes ultrasound guided prostate system aspiration and mpMRI and PET-CT based prostate targeting aspiration.
In one embodiment, a single-factor logistic regression analysis is used to obtain a clinically significant prostate cancer predictor, which refers to a statistically different correlation factor between clinically significant and non-clinically significant prostate cancer groups, using R4.1.1 (R software v4.1.1).
Single-factor logistic regression analysis showed that age (p=0.008), TPSA (p < 0.001), PSAd (p < 0.001), P I-RADS 4 (p=0.001), P I-RADS 5 (p < 0.001) and SUVmax (p < 0.001) were predictive factors for prostate cancer of clinical significance.
S3: and carrying out multi-factor logistic regression analysis on the predicted factors to obtain independent predicted factors of the prostate cancer with clinical significance.
In a specific embodiment, R4.1.1 (R software v4.1.1) is used to select a prediction factor of prostate cancer with clinical significance after single-factor logistic regression analysis according to the P value, and then multiple-factor logistic regression analysis is performed by stepwise regression to select an independent prediction factor of prostate cancer with clinical significance, where the independent prediction factor is a related factor with independent prediction capability.
Multiple regression analysis showed P I-RADS (P I-RADS 4, p=0.037; PI-RADS 5, p=0.006; P I-RADS.ltoreq.3 as a reference) and SUVmax (p < 0.001) as independent predictors of prostate cancer of clinical significance. Considering literature reports in recent years, PSAD is an important prediction index of csPCa compared with traditional tPSA, and has wider application prospect. The mpMRI combined PSAD can effectively help P I-RADS 1-3 minutes and lesions of the peripheral zone of the prostate, and has extremely high negative predictive value. Even though the PSAD corresponding p value in the multi-factor regression analysis is 0.86, we have incorporated PSAD as an important clinical index into the construction of the predictive model.
Using R4.1.1 (R software v4.1.1) to determine the estimated value of each partial regression coefficient (β), a predictive model is obtained as:
(logitP＝ln[P/(1-P)]＝-4.0359+1.6926*a+0.2341*b+c)
wherein a is a PSAD value, b is a SUVmax value, and c is a value corresponding to the PI-RADS score.
S4: constructing an alignment graph according to the independent prediction factors.
Nomograms are a graphical representation of predictive models, in which total scores are calculated based on the values of individual predictive variables, and the risk or survival probability of an event is calculated from the total scores.
Assigning a score to each value level of each influence factor according to the regression coefficient of each influence factor in the model, adding the scores to obtain a total score, and finally calculating the predicted value of the individual ending event through the function conversion relation between the total score and the ending event occurrence probability.
In a specific embodiment, a Nomogram (Nomogram) was constructed by multi-factor logistic regression analysis using R4.1.1 (R software v4.1.1), which can be seen in fig. 2.
In the nomogram, the first line is a score scale, and the score range is 0-100 minutes; the second behavior PSAD ranges from 0 to 2.4, and the corresponding score ranges from 0 to 28.9; third behavior P I-RADS score ranging from 1,2,3,4,5, 11.8 points if 4, 15.1 points if 5, or 0 points otherwise; the fourth behavior SUVmax ranges from 0 to 60, and the corresponding score ranges from 0 to 100 minutes; the fifth action total score ranges from 0 to 160 minutes; the sixth behavior has a clinically significant likelihood of developing prostate cancer ranging from 0.01 to 0.999.
In the alignment chart, the second row to the fourth row are related to factors, and different factor states correspond to different scores of the scale; the total score of the fifth row is divided into the total score of each factor, the risk of the sixth row has a corresponding relation with the total score of the fifth row, and the risk is projected to the corresponding position according to different scores, namely the corresponding risk.
In one embodiment, the risk value of the model is formulated as follows,
wherein a is PSAD value, b is SUVmax value, c is corresponding value of P I-RADS score, and c corresponding value of P I-RADS score is as follows:
if P I-RADS scores 4 points then c is 1.6696, if 5 points then c is 2.1270, otherwise c is 0 points. For example, the patient Li Mou is an examination subject, the PSAd value is 0.67, the SUVmax value is 13.5, the pi-RADS score is 5 points, c= 2.1270, and the PSAd, SUVmax, P I-RADS values are substituted into the above formula, so that the risk value is 0.9157143.
And S5, performing performance evaluation and verification on the constructed prediction model.
To evaluate the performance of the model, in one embodiment, the ROC curve for constructing the training set using R4.1.1 (R software v4.1.1) is shown in FIG. 3 and the calibration curve for constructing the training set is shown in FIG. 4. As can be seen from fig. 3 and 4, the AUC (area under ROC curve) of the alignment chart is 0.936 (95% C I: 0.888-0.984), which indicates that the accuracy of the prediction result of the alignment chart is higher, and the alignment chart has higher diagnostic value. As can be seen from fig. 4, in the calibration curve of the alignment chart: the abscissa is the predicted clinically significant probability of prostate cancer and the ordinate is the actual clinically significant probability of prostate cancer. The 45-degree dotted line in the figure represents an ideal calibration curve, the solid line is an actual calibration curve obtained by calculation, the two directions are consistent and closely overlapped, the model shows good calibration degree, and further the nomogram is further explained to accurately estimate the probability of the research population suffering from the prostate cancer with clinical significance.
To verify the above model, in a specific embodiment, R4.1.1 (R software v4.1.1) is used to verify the internal validity and the external validity of the model. The training set is utilized to carry out internal verification through a 400-time 5-fold cross verification method, the utilization efficiency of the method on samples is relatively high, the method can be used for checking the repeatability of a model development process, and the model is prevented from being excessively fitted to cause overestimation of the performance of the model. The model is characterized in AUC (0.940), R 2 (0.577), D index (0.551) and Brier value (0.097). External verification is performed by incorporating an additional 61 patient build verification set, the model being in an external verification queueThe ROC curve of the constructed validation set, which showed good discrimination, is shown in FIG. 5 with an AUC of 0.924 (95% CI: 0.857-0.990). The calibration curve for constructing the training set is shown in fig. 6, the 45-degree dotted line represents an ideal calibration curve, the solid line is an actual calibration curve obtained by calculation, the two curves are consistent and closely overlapped, and the model shows good calibration degree in an external verification queue.
According to the application, three good prediction indexes of SUVmax, PI-RADS and PSAD are determined through single-factor logistic regression analysis, multi-factor logistic regression analysis and the like, and a nomogram is constructed according to independent prediction factors, so that the risk of suffering from prostate cancer with clinical significance is predicted, and a high-accuracy prediction result is obtained. The mathematical model predicts that the AUC of the prostate cancer with clinical significance is 0.936, the sensitivity is 0.882, the specificity is 0.910, the positive predictive value is 0.811, and the negative predictive value is 0.947 when the cutoff value is 0.316. The prediction model has good sensitivity and specificity, and is obviously higher than serum PSAD (sensitivity: 0.676; specificity: 0.859; AUC: 0.812); mpMRI (sensitivity: 0.794; specificity: 0.821; AUC: 0.806); PET-CT (sensitivity: 0.912; specificity: 0.795; AUC: 0.903) alone.
As shown in fig. 7, another embodiment of the present application further provides a prediction system for predicting a clinically significant prostate cancer, where the system includes a data acquisition module M1 and a risk prediction module M2, and the data processing module M1 is configured to acquire clinical examination and test data of a user. The risk prediction module M2 is configured to calculate a risk of the user for developing prostate cancer with clinical significance according to a pre-trained logistic regression model.
In a specific embodiment, the clinical examination data in the data processing module M1 are specifically: including PSAd value, SUVmax value, PI-RADS score.
In a specific embodiment, the risk prediction module M2 calculates the risk of the user for developing the prostate cancer with clinical significance according to a pre-trained logistic regression model, specifically: clinical examination data provided by a user are obtained and substituted into a pre-trained logistic regression model, such as the technology in the modeling method of the model, to calculate the risk of the user for having the prostate cancer with clinical significance.
If patient Li Mou is an examination subject, the PSAD value is 0.67, SUVmax value is 13.5, and the PIRADS score is 5. The user inputs the PSAD, SUVmax, P I-RADS scores into the M1 module of the prediction system. The prediction system obtains user input data through the data processing module M1, and transmits the user input data to the risk prediction module M2. The risk prediction module M2 substitutes the data into a risk value formula obtained by a pre-trained logistic regression model, calculates a risk value as 0.9157143, and outputs a result.
Those skilled in the art will appreciate that implementing all or part of the above-described methods in accordance with the embodiments may be accomplished by way of a computer program stored on a computer-readable storage medium, which when executed, may comprise the steps of the embodiments of the methods described above.
The application provides a prediction model modeling method and a prediction system for predicting prostate cancer with clinical significance, which are used for constructing a logistic regression model nomogram by using clinical examination and inspection data of patients, wherein the nomogram can well detect high risk groups of the prostate cancer, so that early detection and early treatment of the prostate cancer by early prostate puncture of the high risk groups are realized; the low risk group of the prostate cancer is eliminated, so that unnecessary prostate puncture is avoided, the pain of patients is reduced, and a large amount of medical resources are saved. The risk prediction is carried out by the ordinary people through the device, so that the risk of the prostate cancer can be queried quickly and conveniently at any time and any place according to the actual situation of the people, and the cancer prevention consciousness is improved. In addition, the device is beneficial to medical institutions for targeted treatment according to results, and has important public health significance.
The foregoing description is only of the preferred embodiments of the present application, and is not intended to limit the scope of the application, but rather is intended to cover any equivalents of the structures or equivalent processes disclosed herein or in the alternative, which may be employed directly or indirectly in other related arts.
Claims (4)
1. A method of modeling a predictive model of prostate cancer, the method comprising the steps of:
s1, acquiring clinical examination and inspection data of a study object;
s2, screening out relevant clinical examination and inspection data and performing single-factor logistic regression analysis to obtain prostate cancer prediction factors with clinical significance;
s3, performing multi-factor logistic regression analysis on the predicted factors to obtain independent predicted factors of the prostate cancer with clinical significance, wherein the independent predicted factors are as follows: prostate specific antigen density PSAd, maximum standard uptake value SUVmax, and prostate imaging report and data system PI-RADS score, constructing a predictive model logitp=ln [ P/(1-P) ]= -4.0359+1.6926×a+0.2341×b+c;
where P is the likelihood of developing a clinically significant prostate cancer, a is the PSAD value, b is the SUVmax value, c is the value corresponding to the PI-RADS score,and calculating a risk value of the prostate cancer with clinical significance, wherein the risk value is calculated according to the following formula:
s4, constructing an alignment chart according to independent prediction factors;
and S5, performing performance evaluation and verification on the constructed prediction model.
2. The method of modeling a predictive model of prostate cancer according to claim 1, wherein the clinical examination and test data in step S1 includes: basic data, hematology index, serum PSAD, imaging examination report, pathology examination report.
3. The modeling method of a prostate cancer predictive model according to claim 2, wherein in the alignment chart, the first behavior score scale has a score range of 0-100 points; the second behavior PSAD ranges from 0 to 2.4, and the corresponding score ranges from 0 to 28.9; the third behavior PI-RADS score ranges from 1,2,3,4,5, 11.8 points if 4, 15.1 points if 5, or 0 points if not; the fourth behavior SUVmax ranges from 0 to 60, and the corresponding score ranges from 0 to 100 minutes; the fifth action total score ranges from 0 to 160 minutes; the sixth behavior has a clinically significant probability of prostate cancer ranging from 0.01 to 0.999; in the alignment chart, the second row to the fourth row are related to factors, and different factor states correspond to different scores of the scale; the total score of the fifth row is divided into the total score of each factor, the risk of the sixth row has a corresponding relation with the total score of the fifth row, and the risk is projected to the corresponding position according to different scores, namely the corresponding risk.
4. A prostate cancer prediction system, characterized in that a modeling method of a prostate cancer prediction model according to any one of claims 1-3 is used for predicting a risk of a user suffering from a clinically significant prostate cancer, the system comprising a data acquisition module, a risk prediction module, wherein:
the data processing module is used for acquiring clinical examination and checking data of a user;
the risk prediction module is used for calculating the risk of the user suffering from the prostate cancer with clinical significance according to a pre-trained logistic regression model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310750568.4A CN116978582A (en) | 2023-06-25 | 2023-06-25 | Modeling method and prediction system of prostate cancer prediction model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310750568.4A CN116978582A (en) | 2023-06-25 | 2023-06-25 | Modeling method and prediction system of prostate cancer prediction model |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116978582A true CN116978582A (en) | 2023-10-31 |
Family
ID=88472095
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310750568.4A Pending CN116978582A (en) | 2023-06-25 | 2023-06-25 | Modeling method and prediction system of prostate cancer prediction model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116978582A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117727443A (en) * | 2023-12-13 | 2024-03-19 | 南方医科大学珠江医院 | Prediction system and prediction model for prognosis of prostate cancer patient |
-
2023
- 2023-06-25 CN CN202310750568.4A patent/CN116978582A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117727443A (en) * | 2023-12-13 | 2024-03-19 | 南方医科大学珠江医院 | Prediction system and prediction model for prognosis of prostate cancer patient |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| RU2640568C2 (en) | Methods and systems for gastrointestinal tract cancer risk assessment | |
| CN106202968B (en) | Cancer data analysis method and device | |
| US9996671B2 (en) | Method for providing reliable non-invasive diagnostic tests | |
| CN116978582A (en) | Modeling method and prediction system of prostate cancer prediction model | |
| CN113223722A (en) | Method and system for constructing pulmonary nodule database and prediction model based on nomogram | |
| CN115099331A (en) | Auxiliary diagnosis system for malignant pleural effusion based on interpretable machine learning algorithm | |
| JP2010243406A (en) | Method for detecting extent of clinical condition of liver cancer and chronic liver disease, using discriminant function taking measurement values of afp and pivka-ii as characteristic values | |
| CN114864080A (en) | Method, system, equipment and medium for establishing liver cancer diagnosis model C-GALAD II | |
| US20180173847A1 (en) | Establishing a machine learning model for cancer anticipation and a method of detecting cancer by using multiple tumor markers in the machine learning model for cancer anticipation | |
| CN114023448A (en) | Construction method of endometrial cancer prediction diagnosis model, diagnosis model and diagnosis device | |
| CN104720804A (en) | Method for predicting sensitivity of cervical cancer patient subjected to paclitaxel and cis-platinum adopted chemotherapy | |
| Rodriguez-Lopez et al. | Impaired immune reaction and increased lactate and C-reactive protein for early prediction of severe morbidity and pancreatic fistula after pancreatoduodenectomy | |
| CN117253625A (en) | Construction device of lung cancer screening model, lung cancer screening device, equipment and medium | |
| CN111430030A (en) | Application method and system of biomarker in ovarian cancer assessment | |
| Sottile et al. | A pilot study evaluating serum pro-prostate-specific antigen in patients with rising PSA following radical prostatectomy | |
| CN115602327A (en) | Construction method of prediction model for lung nodule lung cancer risk | |
| CN111583994A (en) | Tumor marker cutoff value combined model and application thereof | |
| CN110082528B (en) | System and application for diagnosing or prognosing human oral cancer | |
| Chen et al. | The added value of S-detect in the diagnostic accuracy of breast masses by senior and junior radiologist groups: a systematic review and meta-analysis | |
| Huang et al. | Prediction Model of Bone Marrow Infiltration in Patients with Malignant Lymphoma Based on Logistic Regression and XGBoost Algorithm | |
| CN111537424B (en) | System for assessing prognosis of patient with glioma based on peripheral blood cells | |
| CN114898878A (en) | Model for predicting purpuric nephritis occurrence risk and Nomogram graph construction method thereof | |
| Morote Robles et al. | Who with suspected prostate cancer can benefit from Proclarix after multiparametric magnetic resonance imaging? | |
| CN117741150A (en) | Combined marker for detecting lung cancer, kit and application thereof | |
| CN117831750A (en) | Evaluation system of triple-negative breast cancer immune combined new adjuvant therapy scheme |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |