CN117524486A - TTE model establishment method for predicting non-progressive survival probability of postoperative patient - Google Patents
TTE model establishment method for predicting non-progressive survival probability of postoperative patient Download PDFInfo
- Publication number
- CN117524486A CN117524486A CN202410008018.XA CN202410008018A CN117524486A CN 117524486 A CN117524486 A CN 117524486A CN 202410008018 A CN202410008018 A CN 202410008018A CN 117524486 A CN117524486 A CN 117524486A
- Authority
- CN
- China
- Prior art keywords
- model
- variables
- prediction
- progression
- predicted
- 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.)
- Granted
Links
- 230000004083 survival effect Effects 0.000 title claims abstract description 90
- 238000000034 method Methods 0.000 title claims abstract description 63
- 230000002980 postoperative effect Effects 0.000 title claims abstract description 52
- 230000000750 progressive effect Effects 0.000 title claims description 8
- 230000000694 effects Effects 0.000 claims abstract description 30
- 210000000265 leukocyte Anatomy 0.000 claims description 51
- 206010028980 Neoplasm Diseases 0.000 claims description 37
- 230000006870 function Effects 0.000 claims description 27
- 238000004458 analytical method Methods 0.000 claims description 24
- 206010027476 Metastases Diseases 0.000 claims description 22
- 230000009401 metastasis Effects 0.000 claims description 22
- 238000004393 prognosis Methods 0.000 claims description 11
- 238000001356 surgical procedure Methods 0.000 claims description 10
- 238000012217 deletion Methods 0.000 claims description 7
- 230000037430 deletion Effects 0.000 claims description 7
- 102000008946 Fibrinogen Human genes 0.000 claims description 6
- 108010049003 Fibrinogen Proteins 0.000 claims description 6
- 229940012952 fibrinogen Drugs 0.000 claims description 6
- 230000002441 reversible effect Effects 0.000 claims description 6
- 238000010187 selection method Methods 0.000 claims description 6
- 241001115394 Reston ebolavirus Species 0.000 claims description 5
- 210000002966 serum Anatomy 0.000 claims description 5
- 238000004590 computer program Methods 0.000 claims description 4
- 108090000623 proteins and genes Proteins 0.000 claims description 4
- 102000004169 proteins and genes Human genes 0.000 claims description 4
- 201000004846 retroperitoneal sarcoma Diseases 0.000 abstract description 47
- 238000011156 evaluation Methods 0.000 abstract description 9
- 230000008569 process Effects 0.000 abstract description 8
- 201000010099 disease Diseases 0.000 abstract description 4
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 abstract description 4
- 238000013179 statistical model Methods 0.000 abstract description 4
- 238000010276 construction Methods 0.000 abstract description 3
- 238000007619 statistical method Methods 0.000 abstract description 3
- 238000012795 verification Methods 0.000 abstract description 2
- 206010061818 Disease progression Diseases 0.000 description 9
- 230000001186 cumulative effect Effects 0.000 description 9
- 230000005750 disease progression Effects 0.000 description 9
- 206010039491 Sarcoma Diseases 0.000 description 7
- 206010073135 Dedifferentiated liposarcoma Diseases 0.000 description 6
- 210000004027 cell Anatomy 0.000 description 6
- 206010024627 liposarcoma Diseases 0.000 description 6
- 238000000611 regression analysis Methods 0.000 description 6
- 238000002271 resection Methods 0.000 description 6
- 238000004088 simulation Methods 0.000 description 6
- 238000012216 screening Methods 0.000 description 5
- 230000000007 visual effect Effects 0.000 description 5
- 230000008901 benefit Effects 0.000 description 4
- 230000002829 reductive effect Effects 0.000 description 4
- 238000012360 testing method Methods 0.000 description 4
- 210000001519 tissue Anatomy 0.000 description 4
- 238000003384 imaging method Methods 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 239000000090 biomarker Substances 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000010219 correlation analysis Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000003902 lesion Effects 0.000 description 2
- 230000000670 limiting effect Effects 0.000 description 2
- 239000003550 marker Substances 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 230000037361 pathway Effects 0.000 description 2
- 230000001681 protective effect Effects 0.000 description 2
- 230000000306 recurrent effect Effects 0.000 description 2
- 238000011282 treatment Methods 0.000 description 2
- 238000007476 Maximum Likelihood Methods 0.000 description 1
- 238000012952 Resampling Methods 0.000 description 1
- 238000002679 ablation Methods 0.000 description 1
- 230000005856 abnormality Effects 0.000 description 1
- 230000002411 adverse Effects 0.000 description 1
- 230000000259 anti-tumor effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 201000011510 cancer Diseases 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 230000008030 elimination Effects 0.000 description 1
- 238000003379 elimination reaction Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 230000014759 maintenance of location Effects 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 229950008516 olaratumab Drugs 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- JMANVNJQNLATNU-UHFFFAOYSA-N oxalonitrile Chemical compound N#CC#N JMANVNJQNLATNU-UHFFFAOYSA-N 0.000 description 1
- 230000001575 pathological effect Effects 0.000 description 1
- 230000000737 periodic effect Effects 0.000 description 1
- 210000004303 peritoneum Anatomy 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 238000011470 radical surgery Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000002560 therapeutic procedure Methods 0.000 description 1
- 230000036962 time dependent Effects 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
- 230000001052 transient effect Effects 0.000 description 1
- 238000011277 treatment modality Methods 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
- 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
- G16H50/30—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for calculating health indices; for individual health risk assessment
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Medical Informatics (AREA)
- Health & Medical Sciences (AREA)
- Public Health (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Optimization (AREA)
- Mathematical Analysis (AREA)
- Databases & Information Systems (AREA)
- Theoretical Computer Science (AREA)
- Computational Mathematics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Probability & Statistics with Applications (AREA)
- Evolutionary Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Primary Health Care (AREA)
- General Health & Medical Sciences (AREA)
- Operations Research (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Epidemiology (AREA)
- Algebra (AREA)
- Pathology (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Biomedical Technology (AREA)
- Measuring And Recording Apparatus For Diagnosis (AREA)
Abstract
The invention provides a TTE model building method for predicting the progression-free survival probability of a postoperative patient. According to the TTE final model construction scheme, a time-varying prediction variable based on longitudinal model prediction, a base line characteristic variable and a cross-section variable related to a laboratory examination index before operation are introduced as covariates of a final model, and a TTE model for predicting postoperative progression-free survival probability of a patient with retroperitoneal sarcoma is constructed by applying various advanced statistical models and methods. The model can effectively estimate the postoperative PFS probability of the patient with the retroperitoneal sarcoma, can accurately identify the variable which has obvious influence on the PFS, and has great clinical significance for realizing individual prediction and early intervention in the postoperative disease progress process of the patient with the retroperitoneal sarcoma. The final model and Cox regression verification can be combined according to the invention, so that the key variable selection and effect evaluation are more accurate.
Description
Technical Field
The invention relates to the technical field of biomedical informatics and medical detection, in particular to a TTE model establishment method for predicting the progression-free survival probability of a postoperative patient.
Background
Postoperative recurrence is the leading cause of death in patients with retroperitoneal sarcoma (RPS). It is counted that the local recurrence rate is about 45% -50% in 5 years of RPS patients who first received radical surgery, whereas the distant metastasis rate is about 21% -34% in 5 years. Therefore, clinical guidelines such as the China medical Association, the national cancer Integrated network (NationalComprehensiveCancerNetwork, NCCN, miss) and the like all suggest lifelong follow-up and periodic imaging examinations of RPS patients. Relative to total survival (OS), progression-free survival (PFS) or disease-free survival (DFS) can intuitively reflect the progression and recurrence of RPS, and data can be acquired earlier in limited clinical practice time for rapid evaluation of disease prognosis, PFS (DFS) is one of the primary clinical endpoints of interest in RPS clinical studies. There is a need to explore models with predictive value for PFS, significant factors, etc. in order to achieve personalized prediction and early intervention for the post-operative disease progression process.
Disclosure of Invention
In view of the above, the present invention aims to provide a method for building a TTE model (time-to-event model, also referred to as event occurrence time model) for predicting the probability of progression-free survival of a patient after an operation, which can accurately estimate the probability of PFS of the patient after the operation and can effectively identify variables significantly affecting PFS.
According to an aspect of the present invention, there is provided a TTE model building method for predicting a probability of progression-free survival of a patient after an operation, the TTE model being used for predicting a probability of progression-free survival of a patient with a sarcoma after peritoneum, the method comprising:
constructing a patient data set comprising the collected post-operative progression free survival data of the patient;
establishing a basic risk model of the TTE model by fitting the constructed patient data set;
obtaining a plurality of candidate predicted variables, wherein the candidate predicted variables comprise time-varying predicted variables and constant predicted variables, the time-varying predicted variables comprise predicted values of a longitudinal model of statistics of a plurality of potential prognosis markers, the potential prognosis markers comprise postoperative body mass indexes BMI, postoperative serum total protein TP and postoperative white blood cells WBC, and the constant predicted variables comprise baseline characteristic variables and cross-section variables related to preoperative laboratory examination indexes, and the variables comprise tumor grading GRADE, baseline metastasis META, excision effect COMP, excision tumor volume RESTV and preoperative fibrinogen LNFIB;
and selecting a key prediction variable from the plurality of candidate prediction variables according to the significance degree of the influence of the candidate prediction variables on the progression-free survival prediction, and adding the key prediction variable into the basic risk model to establish a final model of the TTE model, wherein the final model is used for predicting the progression-free survival probability of the individual patient.
In some embodiments, constructing the patient data set includes:
examining an original patient data set to identify extremely deleted data, the extremely deleted data being data having individual deletion event times greater than a maximum of all of the progression event times in the patient population;
the identified extreme deleted data is filled in by tail-biting to the maximum of all progressive event times in the patient population.
In some embodiments, the underlying risk model employs an exponential distribution model.
In some embodiments, the final model is built based on the following formula:
，
wherein,represents the firstiRisk functions of TTE end models of individuals; />A function representing the TTE base risk model; />And->Respectively the firstiTime-varying predicted variables and constant predicted variables for individual individuals; />And->Is description of the firstjThe coefficient of each prediction variable effect is regular, the coefficient shows that the prediction variable is a risk factor, the coefficient is negative, the prediction variable is a protection factor, and the coefficient is 0, the prediction variable is an irrelevant factor; />A function form of the risk function affected by the time-varying prediction variable is referred;p、qandrthe number of continuous constant prediction variables, two kinds of constant prediction variables and time-varying prediction variables in the final model is sequentially determined.
In some embodiments, selecting a key prediction variable from the plurality of candidate prediction variables to add to the base risk model based on how significant the candidate prediction variable affects the progression-free survival prediction, comprising:
and combining a step forward selection method and a step reverse removal method, and selecting the key predicted variable from the candidate predicted variables according to the significance of the candidate predicted variables for influencing the progression-free survival prediction.
In some embodiments, the method further comprises:
performing multivariate Cox regression of individual parameters to analyze risk ratios of different individual parameter variables to progression-free survival, including empirical bayesian estimation of structural parameters of a longitudinal model of the key predicted variables in the final model;
based on the analysis results of the multivariate Cox regression of individual parameters, individual parameter variables that have significant impact on progression-free survival are identified.
In some embodiments, the method further comprises:
performing multivariate Cox regression of predicted constant variables based on a final model to analyze risk ratios of different predicted constant variables of the final model to progression-free survival, the predicted constant variables incorporated into the analysis comprising predicted results of a longitudinal model of the key predicted variables in the final model at specified times;
And identifying the final model prediction constant variable with obvious influence on the progression-free survival according to the analysis result of the multi-variable Cox regression based on the final model prediction constant variable.
In some embodiments, the method further comprises:
and respectively drawing an alignment chart for predicting the progression-free survival time of the specified period after the operation of the patient according to the analysis result of the multivariate Cox regression of the individual parameters and the analysis result of the multivariate Cox regression of the constant variables based on the final model.
According to another aspect of the present invention, there is also provided an electronic device including:
a memory storing executable instructions;
a processor executing the executable instructions in the memory to implement a TTE model building method for predicting a probability of progression free survival of a postoperative patient as described above.
According to another aspect of the present invention, there is also provided a computer-readable storage medium storing a computer program which, when executed by a processor, implements a TTE model building method for predicting a probability of progression-free survival of a patient after surgery as described above.
According to the technical scheme of the invention, firstly, a data set of a retroperitoneal sarcoma (RPS) patient is constructed, the data set of the patient comprises collected postoperative progression-free survival data of the RPS patient, and then a basic risk model of a TTE model is constructed by fitting the data set of the patient; obtaining a plurality of candidate predicted variables, wherein the candidate predicted variables comprise time-varying predicted variables and constant predicted variables, the time-varying predicted variables comprise predicted values of a longitudinal model of statistics of a plurality of potential prognosis markers, the potential prognosis markers comprise postoperative body mass index BMI, postoperative serum total protein TP and postoperative leucocytes WBC, the constant predicted variables comprise baseline characteristic variables and preoperative laboratory examination index related cross-section variables, and the baseline characteristic variables comprise tumor grading GRADE, baseline metastasis META, excision effect COMP, excision tumor volume RESTV and preoperative fibrinogen LNFIB; and then selecting a key prediction variable from the plurality of candidate prediction variables to add into the basic risk model according to the significance degree of the influence of the candidate prediction variables on the progression-free survival prediction so as to establish a final model of the TTE model, wherein the final model is used for predicting the progression-free survival probability of the individual patient. The technical scheme provided by the invention has at least the following beneficial effects:
1. Compared with a basic risk model, the built TTE final model greatly improves the prediction effect and individuation degree;
2. compared with the traditional survival analysis method, the method has the advantages that the time-varying prediction factors are added, so that the state of a postoperative patient can be dynamically estimated, and the individual risk can be adjusted;
3. advanced mixed model technology is applied, bias of retrospective data is reduced, and prediction reliability is improved;
4. the combination of the final model and Cox regression verification ensures that the key variable selection and effect evaluation are more accurate;
5. constructing a nomogram, and intuitively realizing the dynamic prediction of the progression-free survival of an individual;
6. and quantitative decision support is provided for enhancing postoperative monitoring and making accurate intervention.
In summary, according to the TTE final model construction scheme provided by the invention, a time-varying prediction variable based on longitudinal model prediction, a base line characteristic variable and a cross-section variable related to a preoperative laboratory examination index are introduced as the covariates of the final model, a TTE model for predicting the postoperative progression-free survival probability of a patient with a retroperitoneal sarcoma is constructed by using a plurality of advanced statistical models and methods, the postoperative PFS probability of the patient with the retroperitoneal sarcoma can be effectively estimated by using the model, the variable which has obvious influence on the PFS can be accurately identified, and the TTE model has great clinical significance for realizing individuation prediction and early intervention in the postoperative disease progression process of the patient with the retroperitoneal sarcoma.
The method and apparatus of the present invention have other features and advantages which will be apparent from or are set forth in detail in the accompanying drawings and the following detailed description, which are incorporated herein, and which together serve to explain certain principles of the present invention.
Drawings
The foregoing and other objects, features and advantages of the invention will be apparent from the following more particular descriptions of exemplary embodiments of the invention as illustrated in the accompanying drawings wherein like reference numbers generally represent like parts throughout the exemplary embodiments of the invention.
FIG. 1 illustrates a flow chart of a TTE model building method for predicting a probability of progression free survival of a post-operative patient according to one embodiment of the invention.
FIG. 2 is a table illustrating the progression free lifecycle TTE base risk model parameter results, established in accordance with an exemplary embodiment of the present invention.
Fig. 3 illustrates a visual predictive test diagram of a progression-free survival TTE base risk model for an RPS patient, according to an exemplary embodiment of the invention.
FIG. 4 is a table illustrating the results of a progression free lifecycle TTE end model parameter established in accordance with an exemplary embodiment of the present invention.
Fig. 5 shows a final model risk ratio map for post-operative progression-free survival TTE of RPS patients according to an exemplary embodiment of the invention.
Fig. 6 illustrates a visual predictive test chart of a post-operative progression-free survival TTE end model for an RPS patient, according to an exemplary embodiment of the invention.
Fig. 7 shows post-operative progression-free survival multivariate Cox regression for RPS patients based on longitudinal model individual parameters according to an example embodiment of the invention.
FIG. 8 illustrates an alignment chart of predicted progression free survival 1/3/5 years post-surgery for an RPS patient, according to an exemplary embodiment of the invention.
Fig. 9 shows post-operative progression-free survival multivariate Cox regression of RPS patients based on longitudinal model predicted constant variables according to an example embodiment of the invention.
Fig. 10 shows a nomogram of predicted progression free survival 1/3/5 years post-surgery for RPS patients, according to an exemplary embodiment of the present invention.
Detailed Description
Preferred embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. While the preferred embodiments of the present invention are shown in the drawings, it should be understood that the present invention may be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art.
FIG. 1 illustrates a flow chart of a TTE model building method for predicting a probability of progression free survival of a post-operative patient according to one embodiment of the invention. The TTE model is applied to postoperative progression-free survival prediction of patients with retroperitoneal sarcomas. As shown in the figure, the method comprises the steps 1 to 4.
Step 1, constructing a patient data set comprising the collected progression free lifetime data of the patient.
The data in the patient dataset may be from progression free survival information of the follow-up records, including event time and event type. For example, for individuals who have observed tumor recurrence (including local recurrence and distant metastasis), the event type may be 1, with recurrent event time noted as progressive event time; for individuals who did not observe tumor recurrence but observed a death event, the event type may be 1 and the death event time may be noted as the progression event time; for the rest of the individuals who did not observe either tumor recurrence or death event, the event type could be 0, with the last follow-up time noted as the time to delete event.
In some embodiments, constructing the patient data set includes:
examining an original patient data set to identify extremely deleted data, the extremely deleted data being data having individual deletion event times greater than a maximum of all of the progression event times in the patient population;
The identified extreme deleted data is filled in by tail-biting to the maximum of all progressive event times in the patient population.
According to the embodiment, the data of the data set acquired by the original follow-up visit can be further subjected to data inspection, and extreme deletion data can be identified, namely, the individual deletion event time is larger than the maximum value of all the progress event times in the group. The above extremely deleted data is filled into the maximum value of the time of the progression event in the dataset by tail-biting (wind) method, and the PFS dataset is constructed as the patient dataset adopted by the present invention.
And 2, building a basic risk model of the TTE model by fitting the constructed patient data set.
The TTE model estimates the risk function h (t), i.e. the conditional probability that an individual who has survived without progression to time t, will develop a progression event instantaneously at time t, to fit PFS (progression free survival) observations of each individual in the population.
Since the TTE model is a probabilistic model, each individual provides only one discrete PFS observation point (0 or 1) corresponding to the event time, and the model does not make any assumption about the probability density distribution of the observations, and therefore does not contain random effect terms.
The objective of constructing the underlying risk model is to provide a simple model that does not contain covariates, to fit the progression-free survival observations of each individual in the retroperitoneal sarcoma patient population, to describe the underlying risk functions of the whole population, and to provide a basis for the subsequent establishment of a more complex final model.
In some embodiments, the underlying risk model employs an exponential distribution model.
Typically, the underlying risk model may employ Weibull models, gompertz models, log-logistic models, exponential distribution models, and the like. Through a large number of model simulation researches, the inventor finds that the basic model established based on the exponential distribution model has the best effect on predicting the postoperative PFS probability of a patient with retroperitoneal sarcoma.
The parameters of the underlying risk model may be estimated using a maximum likelihood estimation method.
Step 3, obtaining a plurality of candidate predicted variables, wherein the candidate predicted variables comprise time-varying predicted variables and constant predicted variables, the time-varying predicted variables comprise predicted values of a longitudinal model of statistics of a plurality of potential prognosis markers, the potential prognosis markers comprise postoperative body mass indexes BMI, postoperative serum total protein TP and postoperative leucocytes WBC, and the constant predicted variables comprise baseline characteristic variables and relevant section variables of laboratory examination indexes before operation, and the variables comprise tumor grading GRADE, baseline metastasis META, excision effect COMP, excision tumor volume RESTV and preoperative fibrinogen LNFIB.
The time-varying predictive variables to be included in the progression free survival TTE model investigation may include patient BMI, TP, WBC longitudinal predictions from population longitudinal model simulations.
Longitudinal model (longitudinal model) is a very common model in biomedical statistical models for analyzing and modeling repeated measurement data over time, which usually come from observations at different time points in the same population.
One skilled in the art can construct a longitudinal model of the statistics of potential prognostic markers as desired. For each potential prognosis marker, various statistics can be respectively tried to be taken as the value of a time-varying prediction variable at the time t, wherein the values comprise a model prediction value at the time t, a cumulative average value of the model prediction values within the time range of 0-t, a prediction value at a specific time point, a cumulative average value within the specific time range, a prediction extremum, a cumulative time of index abnormality and the like.
Likewise, for an individual RPS patient, the prognostic marker-related statistic at a particular point in time or time range calculated based on the longitudinal model predictive value is a constant variable having a particular value, and is therefore referred to as a "longitudinal model predictive constant variable" in subsequent multivariate Cox regression analysis.
The often predicted variables to be included in the progression free survival TTE model survey may include the following baseline characteristic variables and preoperative laboratory examination index related cross-sectional variables.
1. Tumor grading: expressed as a dichotomous variable GRADE, grade=0 represents GRADE 1 or GRADE 2 of the france cancer center association sarcoma group (FrenchFederationofCancerCentersSarcomaGroup, FNCLCC), grade=1 represents GRADE fnlcc 3;
2. baseline transfer: expressed as a binary variable META, meta=0 represents no baseline shift present, meta=1 represents a baseline shift present;
3. excision effect: expressed as a binary variable COMP, comp=0 represents complete excision and comp=1 represents incomplete excision;
4. tumor volume was resected: expressed as a dichotomous variable RESTV, restv=0 represents that resected tumor volume is less than or equal to population median (1172.38 cm 3 ) Restv=1 represents that resected tumor volume is greater than population median;
5. preoperative fibrinogen: expressed as continuous variable LNFIB, which is the natural logarithm of the individual preoperative fibrinogen.
And 4, selecting a key prediction variable from the plurality of candidate prediction variables according to the significance degree of the influence of the candidate prediction variables on the progression-free survival prediction, and adding the key prediction variable into the basic risk model to establish a final model of the TTE model, wherein the final model is used for predicting the progression-free survival probability of the patient individual.
In some embodiments, the final model is built based on the following formula:
，
Wherein,represents the firstiRisk functions of TTE end models of individuals; />A function representing the TTE base risk model; />And->Respectively the firstiTime-varying predicted variables and constant predicted variables for individual individuals; />And->Is description of the firstjThe coefficient of each prediction variable effect is regular, the coefficient shows that the prediction variable is a risk factor, the coefficient is negative, the prediction variable is a protection factor, and the coefficient is 0, the prediction variable is an irrelevant factor; />A function form of the risk function affected by the time-varying prediction variable is referred;p、qandrthe number of continuous constant prediction variables, two kinds of constant prediction variables and time-varying prediction variables in the final model is sequentially determined.
Laplace algorithm may be used to estimate parameters related to fixed effects and random effects in the TTE model.
In some embodiments, the key prediction variable may be selected from the candidate prediction variables based on how significant the candidate prediction variable affects the progression-free survival prediction in combination with the progressive forward selection method and the progressive reverse culling method.
The stepwise forward selection method may start with a model without variables, adding one predictive variable at a time that maximizes the objective function value.
The gradual reverse elimination method can start from a full variable model, and gradually eliminates the variable with the least influence on the model prediction result. The change of the objective function value of the removed variable is smaller than a preset threshold value.
According to this embodiment, the forward selection method and the reverse removal method may be combined to make the selected key prediction variables and the resulting final model more reliable. In practice, a stepwise forward selection method can be used first to obtain a variable combination after preliminary screening; then the reverse removal method can be applied to removeVariables that do not contribute much to the model, avoiding the risk of overfitting with forward selection alone, in one example, the level of saliency of the inverse culling is set to a minimum objective function value (MVOF) difference of greater than 6.64 (based on). The adjustment, comparison and evaluation can be continued until an optimal combination of predicted variables is obtained.
The established final model can be diagnosed and evaluated. Model evaluation criteria may include a decrease in Objective Function Value (OFV), precision and rationality of parameter estimates, model predictive performance, robustness, etc. In one example, the model predictive performance may be evaluated using a Visual Predictive Check (VPC) based on 200 simulations. Model robustness can be examined using a Sampling Importance Resampling (SIR) method.
Further adjustments may be made to the model based on an evaluation, such as adjusting the key prediction variable combinations to account for their effects on model prediction performance, to ensure that variables that improve both model fit and prediction effect remain in the final model.
The probability of progression free survival of individuals with retroperitoneal sarcomas can be predicted using a final model constructed according to the methods described above.
According to the TTE model establishing method, a time-varying prediction variable based on longitudinal model prediction, a base line characteristic variable and a cross-section variable related to a preoperative laboratory examination index are introduced as covariates of a final model, a TTE model for predicting postoperative progression-free survival probability of a patient with a retroperitoneal sarcoma is established by using various advanced statistical models and methods, postoperative PFS probability of the patient with the retroperitoneal sarcoma can be effectively estimated by using the model, variables with obvious influence on PFS can be accurately identified, and the TTE model has great clinical significance for realizing individual prediction and early intervention in postoperative disease progression process of the patient with the retroperitoneal sarcoma.
In order to facilitate clinical preliminary assessment of disease progression risk in patients with retroperitoneal sarcomas, based on the above-mentioned TTE final model, the inventors further studied time-varying predicted variables to be obtained based on the longitudinal model, characterized in the form of individual parameter values of their respective models or longitudinal model predicted constant variables at specific time points/time ranges, further developed multivariate Cox regression analysis to more accurately identify key variables that have significant influence on PFS, and laid a foundation for subsequent nomographic drawing.
In some embodiments, the method further comprises:
performing multivariate Cox regression of individual parameters to analyze risk ratios of different individual parameter variables to progression-free survival, including empirical bayesian estimation of structural parameters of a longitudinal model of the key predicted variables in the TTE final model;
based on the analysis results of the multivariate Cox regression of individual parameters, individual parameter variables that have significant impact on progression-free survival are identified.
Multivariate Cox regression analysis can be performed on the patient's progression-free survival, and the included variables can include empirical bayesian estimates (empirical) of structural parameters derived from a BMI, TP, WBC population longitudinal model
Bayesian, EBEs). In some examples, individual parameter variables that are included in the analysis may also include other prognostic factors identified by other pathways that significantly affect PFS. All individual EBE values for each parameter can be normalized and converted to continuous variables with a population mean of 0 and variance of 1.
In some embodiments, the method further comprises:
performing multivariate Cox regression of predicted constant variables based on a final model to analyze risk ratios of different predicted constant variables of the final model to progression-free survival, the predicted constant variables incorporated into the analysis comprising predicted results of a longitudinal model of the key predicted variables in the final model at specified times;
And identifying the final model prediction constant variable with obvious influence on the progression-free survival according to the analysis result of the multi-variable Cox regression based on the final model prediction constant variable.
The BMI, TP, WBC model predictions at specific time points (ranges) can be included as constant variables into the PFS multivariate Cox regression analysis based on the screening results of the TTE model. In addition, variables incorporated into the analysis also include other prognostic factors identified by other pathways that significantly affect PFS.
In some embodiments, the method further comprises: and drawing an alignment chart for predicting the progression-free survival time of the patient in a specified postoperative period according to the analysis result of the multivariate Cox regression of the individual parameters and the analysis result of the multivariate Cox regression of the final model prediction constant variables.
The alignment chart for predicting the progression-free survival of the RPS patient 1/3/5 years after surgery can be drawn based on the analysis result of the multivariate Cox regression of the individual parameters obtained above and the analysis result of the multivariate Cox regression of the constant variables predicted based on the final model, respectively.
An exemplary application example according to the present embodiment is described below.
1. Raw data overview
In an RPS database containing a total of 174 patients, 111 events were recorded, 86 of which were tumor recurrence events and 25 of which were death events; the maximum observed value of the time of the progression event was 1213 days. Of the 63 individuals with the loss event, the original loss event time of 28 patients exceeded 1213 days, the median value of these 28 PFS original observations was 1753 days, and the distribution ranged from 1216 to 3132 days. After tail-biting, 1213 days was taken as the event time entry modeling dataset for those individuals with extremely deleted data. For the rest 35 individuals with deletion events and 111 individuals with progress events, corresponding PFS actual observed values are recorded. The dataset used to construct the progression free survival TTE model contained 174 data points for a total of 174 patients.
2. TTE base risk model of PFS
2.1 model parameter results
The progression free survival base risk model conforms to an exponential distribution model, and the TTE base risk model parameter estimation results are shown in the table of FIG. 2. The table includes an estimated value (estimate) and a phase of the base risk coefficient λ0The median value of the parameter distribution and 95% Confidence Interval (CI) obtained for the standard error (RSE), as well as SIR samples. The estimated value of λ0 is close to the SIR median, RSE is 10.8%, indicating good model stability and higher parameter accuracy. Basic risk function h 0 (t) is equal to 1.21×10 -3 Day, which indicates that the number of individuals who develop disease progression per day in the current RPS patient population is 0.121% of the total number of individuals who have not progressed.
2.2 model evaluation
The VPC results of 200 simulations of the underlying risk model are shown in fig. 3. Fig. 3 illustrates a visual predictive test diagram of a progression-free survival TTE base risk model for an RPS patient, according to an exemplary embodiment of the invention. The solid line curve in FIG. 3 is the observed Kaplan-Meier curve, and the black short vertical line on the solid line curve represents the deleted data; the hatched portion enclosed by the dashed line is the 95% confidence interval of the Kaplan-Meier curve based on 200 simulations of the model.
As can be seen from fig. 3, the Kaplan-Meier curve was observed to be generally consistent with the simulated 95% ci versus time distribution, indicating that the underlying risk model was able to predict the general trend of the progression/deletion event occurrence time distribution in the population. However, it can be clearly seen from the figure that the basic model, which assumes that the risk function of each individual in the population is the same, has poor effect on PFS prediction at early or late stages after surgery, and especially for the time phase that progresses faster at early stages after surgery, the Kaplan-Meier curve is observed to fall completely outside the prediction interval. It is therefore necessary to introduce predictive variables to describe the inter-individual differences in risk of progression and its variation over time.
3. TTE end model of PFS
3.1 model parameter results
The progression-free survival TTE end model incorporates 5 predicted variables, longitudinal TP and WBC, baseline Metastasis (META), ablation effect (COMP), tumor GRADE (GRADE), etc., wherein the functional form characterizing the quantitative relationship between 2 time-varying variables and progression risk is shown in the following formula:
，
wherein TP is DAY7 Predicted values of TP longitudinal models on the 7 th day after operation; WBC (WBC) AVE (t) is the cumulative average of the daily WBC counts predicted by the WBC longitudinal model over a time range of 0-t. The two formulas are non-positive and non-negative piecewise functions, respectively, describing two biomarker-related statistics versus the lower limit of the TP clinical reference range (60 g/L) and the upper limit of the WBC clinical reference range (10×10) 9 cells/L). Wherein the predictive variables characterizing WBC changes also comprise a time dependent conditional function, i.e. WBC within 8 weeks after surgery AVE (t) is continuously accumulated over time, and after 8 weeks the variable is constantly equal to the cumulative average of individual WBC predictions over 0-56 days. Based on some correlation analysis results, individual TP DAY7 With WBC AVE The Spearman correlation coefficient between them was-0.22, i.e. it exhibited a weak correlation.
The final model risk function can be described by the following formula:
，
wherein beta is TP 、β WBC 、γMETA、γ COMP 、γ GRADE The effect coefficients of TP, WBC, baseline metastasis, excision effect, tumor grade affecting PFS are in turn. The remaining parameters and symbol definitions are as described above. The final model parameter results are shown in the table of figure 4,the estimated value of (2) is reduced by an order of magnitude compared to the underlying risk model, in the final model 3.21×10 -4 /day。β WBC 、γMETA、γ COMP 、γ GRADE Positive numbers indicate higher WBC counts, higher risk of progression in patients with metastasis at baseline, incomplete resection, or tumor grade fnlcc 3; beta TP Negative numbers indicate that serum TP levels are a protective factor for PFS. None of the 95% cis for each effect coefficient contained 0, indicating reasonable inclusion of the predicted variable in the final model and selection of functional form. Removing gamma GRADE The RSE of the (C) is not more than 30.1%, the RSE of the other parameters is not more than 30%, The precision of the parameter basically meets the requirement. Each parameter estimate is close to the SIR median and falls within the corresponding 95% ci, indicating good model robustness.
Fig. 5 shows a graph of the risk ratio of the end model of the post-operative progression-free survival TTE of RPS patients, with black squares and black solid lines being the risk ratio of each predicted variable and its 95% confidence interval, respectively, according to an exemplary embodiment of the invention.
Based on the final model parameter estimation results, a corresponding risk ratio (HR) for each predicted variable may be calculated. As shown in fig. 5, for an individual in the dataset, TP DAY7 Every 5g/L increase (assuming the above formula is not exceededMedium piecewise function threshold 60 g/L), patient progression risk is reduced to 59.1% before change; WBC (WBC) AVE Every 5×10 increase 9 cell/L (assuming not lower than the above formula +.>Critical value of middle segment function 10 x 10 9 cells/L), the risk of patient progression increases 2.60-fold. The risk of progression of patients with metastasis, incomplete resection, tumor grade 3 of FNLCC at baseline was 2.86, 3.06, 2.01 times the risk of patients without metastasis, complete resection, tumor grade 1/2 of FNLCC at baseline.
3.2 model evaluation
The final model 200 simulated VPC results are shown in figure 6. Fig. 6 illustrates a visual predictive test chart of a post-operative progression-free survival TTE end model for an RPS patient, according to an exemplary embodiment of the invention. The solid line curve in fig. 6 is the observed Kaplan-Meier curve (the black short vertical line on it represents the deleted data); the hatched portion enclosed by the dashed line is the 95% confidence interval of the Kaplan-Meier curve based on 200 simulations of the model.
As shown in FIG. 6, the Kaplan-Meier curve was observed to fall almost completely within the simulated 95% CI, indicating that the model predictive performance was good.
4. Multivariable regression based on longitudinal model individual parameters
Sensitive parameters in the population longitudinal model can characterize the individual dynamics of the biomarker to some extent. In the BMI, TP, WBC population longitudinal model constructed according to this example, a total of 14 structural parameters with inter-individual variation were included. Based on the correlation analysis results, individual parameters with Spearman correlation coefficients not exceeding + -0.6 were included in the variable screening process of the multivariate regression.
The multivariate Cox regression results based on the normalized individual parameters are shown in fig. 7. Specifically, FIG. 7 illustrates a post-operative progression-free survival multivariate Cox regression of RPS patients based on longitudinal model individual parameters, where TP_TP is calculated according to an exemplary embodiment of the present invention POST TP model parameters—postoperative current day TP level; TP_TP SS TP model parameters—steady state TP level; wbc_γ, WBC model parameters—negative feedback shape factor; WBC_DEPOS 0 WBC model parameters-reservoir compartment initial values; DDlipo, dedifferentiated liposarcoma tissue subtype; fnlcc, france cancer center joint sarcomas group tumor grading system. The individual parameters of the longitudinal model are normalized to a population distribution with a mean value of 0 and a variance of 1. * P is p <0.05，**p<0.01，***p<0.001。
As shown in FIG. 7, the independent prognostic factors for PFS include TP POST (postoperative transient TP level), TP SS (steady state TP level), γ (WBC feedback Loop shape factor), DEPOS 0 (WBC reservoir compartment initial) 4 longitudinal model parameters, excision effect, tissue subtype, tumor grade and baseline metastasis. Wherein TP is POST And TP SS HR of (a) is significantly less than 1, i.e., is expressed as a protective factor; in contrast, gamma and DEPOS 0 Are risk factors for PFS. Cox model consistency index (C-index) of 0.78 and 95% CI of 0.74-0.82, indicating better model discrimination.
An alignment chart drawn based on the above Cox regression result is shown in fig. 8. Specifically, FIG. 8 shows a nomogram (based on longitudinal model individual parameters) of predicted 1/3/5 year progression-free survival of an RPS patient, according to one exemplary embodiment of the invention, where TP_TP POST TP model parameters—postoperative current day TP level; TP_TP SS TP model parameters-steady stateTP level; wbc_γ, WBC model parameters—negative feedback shape factor; WBC_DEPOS 0 WBC model parameters-reservoir compartment initial values; DDlipo, dedifferentiated liposarcoma tissue subtype; fnlcc, france cancer center joint sarcomas group tumor grading system.
The longitudinal model parameters in fig. 8 all retain the original dimensions. Progression free survival may be predicted based on patient individual parameters and baseline characteristic information. For example, for both patients ID158 and ID174 in the dataset, both received complete surgical resection, neither tissue subtype was in the dedifferentiated liposarcoma (DDlipo), both tumor grades were fnlcc grade 3, neither had baseline metastasis; TPPOST, TPSS, gamma, DEPOS0 individual parameter values of ID158 are 48.79g/L, 66.80g/L, 0.350, 1.393×10 in order 10 cells/L, and corresponding parameter values of ID174 are 53.02g/L, 70.92g/L, 0.191, 1.968X10, in order 10 cells/L. If the total risk scores of the IDs 158 and 174 are 176.3 and 130.5 respectively, as calculated according to fig. 8, the 1-year predicted progression-free survival probability of the ID158 is 48%, and the probability of disease progression in 3 years of the patient is more than 90%; the predicted progression-free survival probabilities for 1 year, 3 years, and 5 years for ID174 were 80%, 50%, and 46% in order.
5. Multivariable regression for predicting constant variables based on longitudinal model
WBC for time-varying predictive variable incorporated in the final model of the present example AVE (t) we tried to screen WBC longitudinal model predictive constant over a specific time point or time range during the TTE model construction process to describe the effect of WBC on PFS over time. The results show the cumulative average of daily WBC predictions (WBC AVE _ 8W ) There is some predictive power for PFS. Based on statistics, WBC AVE _ 8W The median of the population was 7.89×10 9 cells/L, patient ratio exceeding the upper limit of the clinical reference range was 21.3%; TP model prediction constant variable TP DAY7 The median of the population of (2) was 54.56g/L and the proportion of patients below the lower limit of the clinical reference range was 82.2%.
The results of incorporating the longitudinal model predictive constant variables described above into the Cox regression analysis are shown in fig. 9. Specifically, the figures9 illustrates longitudinal model-based predictive constant-variable, post-operative progression-free survival multivariate Cox regression of RPS patients in accordance with an exemplary embodiment of the invention, where TP DAY7 Model predicts day 7 TP levels; WBC (WBC) AVE _ 8W Model predicts 8 week cumulative average of WBC counts; fnlcc, france cancer center joint sarcomas group tumor grading system. * P<0.01，***p<0.001。
Wherein TP and WBC related prediction variable retention formulasAnd->In a functional form of (a). Independent prognostic factors for PFS include TP DAY7 、WBC AVE _ 8W Baseline metastasis, resecting effect and tumor grade. TP removal DAY7 WBC with HR less than 1 AVE _ 8W Baseline presence of metastasis, incomplete excision, fnlcc grade 3 all appear as risk factors for PFS. C-index was 0.79 (95% CI: 0.75-0.83), indicating better model discrimination.
Plotting an alignment chart as shown in fig. 10 based on the Cox regression results described above, the progression-free survival of the individual can be predicted based on the longitudinal model prediction results and the baseline characteristic information at the selected time points. Specifically, FIG. 10 shows an alignment chart (predicting constant volume based on longitudinal model) of predicted 1/3/5 year progression-free survival of an RPS patient obtained according to an exemplary embodiment of the invention, where TP DAY7 Model predicts day 7 TP levels; WBC (WBC) AVE _ 8W Model predicts 8 week cumulative average of WBC counts; fnlcc, france cancer center joint sarcomas group tumor grading system.
For example, for both patients ID158 and ID174 in the dataset, baseline metastasis did not exist in both, tumor resection was complete and tumor grading was fnlcc grade 3; TP of ID158 DAY7 And WBC (WBC) AVE _ 8W Predicted values are 52.78g/L and 1.125X10 respectively 10 The corresponding predicted values for ID174 are 59.43g/L and 5.07×109cells/L, respectively. ID158 and ID174 are calculated according to FIG. 10The total risk score is 63.5 and 27.5 respectively, the 1 year predicted no-progress survival probability of the ID158 is 54%, and the 3 year and 5 year predicted probabilities are reduced to 16% and 12% respectively; the predicted progression-free survival probabilities for 1 year, 3 years, and 5 years for ID174 were 78%, 48%, and 43% in this order.
RPS has the pathological property of easy recurrence after surgery. Although the clinical guidelines in Zhongmei recommend that patients regularly undergo imaging examinations, many patients fail to monitor the appearance of new lesions after tumor resection in time due to the limitations of many realistic factors. Tumor recurrence-related records in patient database of patients with retroperitoneal sarcoma obtained according to the present invention, the median of the first imaging tumor volume of new lesions after patient operation was 34.00cm 3 The method comprises the steps of carrying out a first treatment on the surface of the 38.5% of patients have tumor volumes exceeding 50cm when diagnosed with tumor recurrence 3 . The lack of timely monitoring of the course of recurrent disease has a potential adverse effect on the clinical benefit of patients from subsequent anti-tumor therapy. Therefore, based on the PFS index, the individual evaluation and prediction of the postoperative disease progression risk of the patient are carried out as early as possible, and the establishment of follow-up visit schemes is guided, so that the method is particularly important for improving the life quality of the patient, improving the treatment response rate and the like. According to the present invention, a quantitative tool is provided for personalized PFS prediction for RPS patients.
Through model screening, the TTE basic risk model according to the embodiment adopts an index distribution model, lambda 0 The estimated value is 1.90 times of the corresponding parameter of the OS basic model, namely, the probability of disease progression of the RPS patient is larger than the death probability of the patient. This is consistent with analysis of the raw survival observations of the population by other routes, with a median progression-free survival estimated by Kaplan-Meier of 475 days, well below the median total survival (1308 days), i.e., most patients developed tumor recurrence or metastasis prior to death. In a published model prognosis study of soft tissue sarcomas, the risk of disease progression in patients treated with olaratumab also met an exponential distribution model, λ 0 6.04X 10 -3 Day, far above the TTE model estimation according to the invention, the difference in the two parameters may be related to the different treatment modalities and patient characteristics. In addition, in the process of constructing data, event types corresponding to 28 individuals processed by adopting tail-biting methodThe event types of all individuals are consistent with the original data, and reasonable model fitting results can be obtained, so that the data processing method is considered to have good effects.
Based on the analysis results of the existing nonparametric method and the semi-parametric method, 3 time-varying prognostic markers BMI, TP and WBC are all examined in the variable screening process of the TTE model. Unlike the analysis conclusion of the prior art, in the present exemplary embodiment according to the present invention, statistics such as percentage BMI, cumulative average, etc. of the BMI longitudinal model predictions have no significant effect on the risk of progression; in contrast, the analysis results of the prior art showed no significant correlation of post-operative WBC counts with PFS, however, in the TTE model of the present exemplary embodiment, the WBC real-time running average was a significant predictive variable of PFS. It follows that non-parametric and semi-parametric methods according to the prior art have significant limitations in analyzing the correlation between time-varying variables and clinical endpoints; in contrast, the combined modeling method of the group longitudinal model and the parameterized TTE can effectively reduce bias introduced by retrospective longitudinal observation data and restore the real data characteristics to a certain extent. After comparison, the final model has good prediction performance on PFS, and the model is stable and reliable.
Multivariate Cox regression analysis according to the invention identified individual parameters of the longitudinal model related to PFS, including TP characterizing TP dynamics POST And TP SS And WBC-related gamma and DEPOS 0 . In the prior art, standardized individual parameters are included in a forest map, and the effect coefficients of variables are relatively close; while the alignment chart obtained according to the exemplary embodiment of the present invention shows original parameter values having physiological significance, which is convenient for clinical application.
According to another aspect of the present invention, there is also provided an electronic apparatus. The electronic device includes:
a memory storing executable instructions:
a processor executing the executable instructions in the memory to implement a TTE model building method for predicting a probability of progression free survival of a postoperative patient according to the present invention.
The TTE model is used to predict the probability of postoperative progression-free survival of a patient with a retroperitoneal sarcoma.
In particular, the memory may include one or more computer program products, which may include various forms of computer-readable storage media, such as volatile memory and/or non-volatile memory. The volatile memory may include, for example, random Access Memory (RAM) and/or cache memory (cache), and the like. The non-volatile memory may include, for example, read Only Memory (ROM), hard disk, flash memory, and the like.
The processor may be a Central Processing Unit (CPU) or other form of processing unit having data processing and/or instruction execution capabilities, and may control other components in the electronic device to perform the desired functions. In one embodiment of the invention, the processor is configured to execute the computer readable instructions stored in the memory.
According to another aspect of the present invention, there is also provided a computer-readable storage medium storing a computer program which, when executed by a processor, implements a TTE model building method for predicting a probability of progression-free survival of a patient after surgery according to the present invention. The TTE model is used to predict the probability of postoperative progression-free survival of a patient with a retroperitoneal sarcoma.
A computer-readable storage medium according to an embodiment of the present invention has stored thereon non-transitory computer-readable instructions. When executed by a processor, perform all or part of the steps of the methods of embodiments of the invention described above.
The computer-readable storage medium described above includes, but is not limited to: optical storage media (e.g., CD-ROM and DVD), magneto-optical storage media (e.g., MO), magnetic storage media (e.g., magnetic tape or removable hard disk), media with built-in rewritable non-volatile memory (e.g., memory card), and media with built-in ROM (e.g., ROM cartridge).
It should be understood by those skilled in the art that, in order to solve the technical problem of how to obtain a good user experience effect, the present embodiment may also include well-known structures such as a communication bus, an interface, and the like, and these well-known structures are also included in the protection scope of the present invention.
It will be appreciated that the foregoing embodiments and implementations of the present disclosure may be combined with each other to form combined embodiments and implementations without departing from the principle logic, and are not repeated herein for the sake of brevity. It will be appreciated by those skilled in the art that in the above-described methods of the embodiments, the particular order of execution of the steps should be determined by their function and possible inherent logic.
Note that all features disclosed in this specification (including any accompanying claims, abstract and drawings) may be replaced by alternative features serving the same, equivalent or similar purpose, unless expressly stated otherwise. Thus, unless expressly stated otherwise, each feature disclosed is one example only of a generic set of equivalent or similar features. Where used, further, preferably, still further and preferably, the brief description of the other embodiment is provided on the basis of the foregoing embodiment, and further, preferably, further or more preferably, the combination of the contents of the rear band with the foregoing embodiment is provided as a complete construct of the other embodiment. A further embodiment is composed of several further, preferably, still further or preferably arrangements of the strips after the same embodiment, which may be combined arbitrarily.
It will be appreciated by persons skilled in the art that the embodiments of the invention described above and shown in the drawings are by way of example only and are not limiting. The objects of the present invention have been fully and effectively achieved. The functional and structural principles of the present invention have been shown and described in the examples and embodiments of the invention may be modified or practiced without departing from the principles described.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present disclosure, and not for limiting the same; although the present disclosure has been described in detail with reference to the foregoing embodiments, it should be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some or all of the technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit of the corresponding technical solutions from the scope of the technical solutions of the embodiments of the present disclosure.
Claims (10)
1. A method of constructing a TTE model for predicting a probability of progression-free survival of a patient after surgery, the method comprising:
constructing a patient data set comprising the collected post-operative progression free survival data of the patient;
Establishing a basic risk model of the TTE model by fitting the constructed patient data set;
obtaining a plurality of candidate predicted variables, wherein the candidate predicted variables comprise time-varying predicted variables and constant predicted variables, the time-varying predicted variables comprise predicted values of a longitudinal model of statistics of a plurality of potential prognosis markers, the potential prognosis markers comprise postoperative body mass indexes BMI, postoperative serum total protein TP and postoperative white blood cells WBC, and the constant predicted variables comprise baseline characteristic variables and cross-section variables related to preoperative laboratory examination indexes, and the variables comprise tumor grading GRADE, baseline metastasis META, excision effect COMP, excision tumor volume RESTV and preoperative fibrinogen LNFIB;
and selecting a key prediction variable from the plurality of candidate prediction variables according to the significance degree of the influence of the candidate prediction variables on the progression-free survival prediction, and adding the key prediction variable into the basic risk model to establish a final model of the TTE model, wherein the final model is used for predicting the progression-free survival probability of the individual patient.
2. The method of claim 1, wherein constructing the patient data set comprises:
examining an original patient data set to identify extremely deleted data, the extremely deleted data being data having individual deletion event times greater than a maximum of all of the progression event times in the patient population;
The identified extreme deleted data is filled in by tail-biting to the maximum of all progressive event times in the patient population.
3. The method of claim 1, wherein the base risk model employs an exponential distribution model.
4. The method of claim 1, wherein the final model is built based on the following equation:
，
wherein,represents the firstiRisk functions of TTE end models of individuals; />A function representing the TTE base risk model; />And->Respectively the firstiTime-varying predicted variables and constant predicted variables for individual individuals; />And->Is description of the firstjThe coefficient of each prediction variable effect is regular, the coefficient shows that the prediction variable is a risk factor, the coefficient is negative, the prediction variable is a protection factor, and the coefficient is 0, the prediction variable is an irrelevant factor;/>a function form of the risk function affected by the time-varying prediction variable is referred;p、qandrthe number of continuous constant prediction variables, two kinds of constant prediction variables and time-varying prediction variables in the final model is sequentially determined.
5. The method of claim 1, wherein selecting a key prediction variable from the plurality of candidate prediction variables to add to the base risk model based on how significant the candidate prediction variable affects progression-free survival prediction, comprises:
And combining a step forward selection method and a step reverse removal method, and selecting the key predicted variable from the candidate predicted variables according to the significance of the candidate predicted variables for influencing the progression-free survival prediction.
6. The method according to claim 1, wherein the method further comprises:
performing multivariate Cox regression of individual parameters to analyze risk ratios of different individual parameter variables to progression-free survival, including empirical bayesian estimation of structural parameters of a longitudinal model of the key predicted variables in the final model;
based on the analysis results of the multivariate Cox regression of individual parameters, individual parameter variables that have significant impact on progression-free survival are identified.
7. The method of claim 6, wherein the method further comprises:
performing multivariate Cox regression of predicted constant variables based on a final model to analyze risk ratios of different predicted constant variables of the final model to progression-free survival, the predicted constant variables incorporated into the analysis comprising predicted results of a longitudinal model of the key predicted variables in the final model at specified times;
and identifying the final model prediction constant variable with obvious influence on the progression-free survival according to the analysis result of the multi-variable Cox regression based on the final model prediction constant variable.
8. The method of claim 7, wherein the method further comprises:
and respectively drawing an alignment chart for predicting the progression-free survival time of the specified period after the operation of the patient according to the analysis result of the multivariate Cox regression of the individual parameters and the analysis result of the multivariate Cox regression of the constant variables based on the final model.
9. An electronic device, the electronic device comprising:
a memory storing executable instructions;
a processor executing the executable instructions in the memory to implement the method of any of claims 1-8.
10. A computer readable storage medium, characterized in that the computer readable storage medium stores a computer program which, when executed by a processor, implements the method of any of claims 1-8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410008018.XA CN117524486B (en) | 2024-01-04 | 2024-01-04 | TTE model establishment method for predicting non-progressive survival probability of postoperative patient |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410008018.XA CN117524486B (en) | 2024-01-04 | 2024-01-04 | TTE model establishment method for predicting non-progressive survival probability of postoperative patient |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117524486A true CN117524486A (en) | 2024-02-06 |
| CN117524486B CN117524486B (en) | 2024-04-05 |
Family
ID=89749754
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410008018.XA Active CN117524486B (en) | 2024-01-04 | 2024-01-04 | TTE model establishment method for predicting non-progressive survival probability of postoperative patient |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117524486B (en) |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160342764A1 (en) * | 2015-05-19 | 2016-11-24 | Universidad De Vigo | System, computer-implemented method and computer program product for individualized multiple-disease quantitative risk assessment |
| CN112309576A (en) * | 2020-09-22 | 2021-02-02 | 江南大学 | Colorectal cancer survival period prediction method based on deep learning CT (computed tomography) image omics |
| CN112635056A (en) * | 2020-12-17 | 2021-04-09 | 郑州轻工业大学 | Lasso-based esophageal squamous carcinoma patient risk prediction nomogram model establishing method |
| CN113270188A (en) * | 2021-05-10 | 2021-08-17 | 北京市肿瘤防治研究所 | Method and device for constructing prognosis prediction model of patient after esophageal squamous carcinoma radical treatment |
| CN114613502A (en) * | 2022-02-25 | 2022-06-10 | 广西医科大学附属肿瘤医院 | Nasopharynx cancer patient survival prediction model, construction method, device and storage medium |
| CN114639482A (en) * | 2022-03-21 | 2022-06-17 | 郑州轻工业大学 | IDPC and LASSO-based esophageal squamous carcinoma prognosis survival risk assessment method |
| CN114898873A (en) * | 2022-03-31 | 2022-08-12 | 中山大学附属第一医院 | Method and system for predicting cardiovascular disease risk of diabetes mellitus pre-stage patient |
| CN115497562A (en) * | 2022-10-27 | 2022-12-20 | 中国医学科学院北京协和医院 | Pancreatic cancer prognosis prediction model construction method based on copper death-related gene |
| CN116313077A (en) * | 2023-02-20 | 2023-06-23 | 山东大学齐鲁医院 | Method and system for risk assessment after radical resection of bile duct cancer |
| CN117116493A (en) * | 2023-07-19 | 2023-11-24 | 重庆金域医学检验所有限公司 | Risk prediction method and system based on cox regression model |
| US20230383364A1 (en) * | 2022-05-05 | 2023-11-30 | The First Affiliated Hospital Of Sun Yat-Sen University | Prognostic model of hepatocellular carcinoma based on ddr and icd gene expression and construction method and application thereof |
-
2024
- 2024-01-04 CN CN202410008018.XA patent/CN117524486B/en active Active
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160342764A1 (en) * | 2015-05-19 | 2016-11-24 | Universidad De Vigo | System, computer-implemented method and computer program product for individualized multiple-disease quantitative risk assessment |
| CN112309576A (en) * | 2020-09-22 | 2021-02-02 | 江南大学 | Colorectal cancer survival period prediction method based on deep learning CT (computed tomography) image omics |
| CN112635056A (en) * | 2020-12-17 | 2021-04-09 | 郑州轻工业大学 | Lasso-based esophageal squamous carcinoma patient risk prediction nomogram model establishing method |
| CN113270188A (en) * | 2021-05-10 | 2021-08-17 | 北京市肿瘤防治研究所 | Method and device for constructing prognosis prediction model of patient after esophageal squamous carcinoma radical treatment |
| CN114613502A (en) * | 2022-02-25 | 2022-06-10 | 广西医科大学附属肿瘤医院 | Nasopharynx cancer patient survival prediction model, construction method, device and storage medium |
| CN114639482A (en) * | 2022-03-21 | 2022-06-17 | 郑州轻工业大学 | IDPC and LASSO-based esophageal squamous carcinoma prognosis survival risk assessment method |
| CN114898873A (en) * | 2022-03-31 | 2022-08-12 | 中山大学附属第一医院 | Method and system for predicting cardiovascular disease risk of diabetes mellitus pre-stage patient |
| US20230383364A1 (en) * | 2022-05-05 | 2023-11-30 | The First Affiliated Hospital Of Sun Yat-Sen University | Prognostic model of hepatocellular carcinoma based on ddr and icd gene expression and construction method and application thereof |
| CN115497562A (en) * | 2022-10-27 | 2022-12-20 | 中国医学科学院北京协和医院 | Pancreatic cancer prognosis prediction model construction method based on copper death-related gene |
| CN116313077A (en) * | 2023-02-20 | 2023-06-23 | 山东大学齐鲁医院 | Method and system for risk assessment after radical resection of bile duct cancer |
| CN117116493A (en) * | 2023-07-19 | 2023-11-24 | 重庆金域医学检验所有限公司 | Risk prediction method and system based on cox regression model |
Non-Patent Citations (3)
| Title |
|---|
| 李成鹏: "右半腹腔脏器切除术治疗腹膜后脂肪肉瘤的经验体会", 《中华普通外科杂志》, vol. 35, no. 6, 30 June 2020 (2020-06-30) * |
| 霍明科: "腹膜后脂肪肉瘤患者首次外科切除后 无进展生存期的影响因素及预后", 《中华普通外科杂志》, vol. 34, no. 5, 31 May 2019 (2019-05-31) * |
| 黎阳: "腹膜后脂肪肉瘤患者首次外科切除后无进展生存期的影响因素及预后分析", 《饮食保健》, vol. 7, no. 33, 31 August 2020 (2020-08-31), pages 64 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117524486B (en) | 2024-04-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP6063446B2 (en) | Analysis of biomarker expression in cells by product rate | |
| Hui et al. | Clinician prediction of survival versus the Palliative Prognostic Score: which approach is more accurate? | |
| CN110634563A (en) | Differential diagnosis device for diabetic nephropathy and non-diabetic nephropathy | |
| CN115699204A (en) | Clinical predictor based on multiple machine learning models | |
| CN111814868A (en) | Model based on image omics feature selection, construction method and application | |
| CN113782183B (en) | Device and method for predicting risk of pressure injury based on multi-algorithm fusion | |
| CN116705325B (en) | Wound infection risk assessment method and system | |
| CN112183557A (en) | MSI prediction model construction method based on gastric cancer histopathology image texture features | |
| Liu et al. | Evaluating the predictive value of biomarkers with stratified case-cohort design | |
| Kozminski et al. | Standardizing the definition of adverse pathology for lower risk men undergoing radical prostatectomy | |
| CN115295151A (en) | Sepsis prediction system, prediction model construction method, system and kit | |
| Gosselin et al. | Developing and using statistical tools to estimate observer effect for ordered class data: The case of the IBP (Index of Biodiversity Potential) | |
| Wu et al. | A prediction model to evaluate the pretest risk of malignancy in solitary pulmonary nodules: evidence from a large Chinese southwestern population | |
| CN117524486B (en) | TTE model establishment method for predicting non-progressive survival probability of postoperative patient | |
| CN111528918B (en) | Tumor volume change trend graph generation device after ablation, equipment and storage medium | |
| Zhao et al. | A machine learning method for improving liver cancer staging | |
| CN116483817A (en) | Data filling model selection and health evaluation method and device | |
| Janes et al. | The optimal ratio of cases to controls for estimating the classification accuracy of a biomarker | |
| Casey et al. | A machine learning approach to prostate cancer risk classification through use of RNA sequencing data | |
| CN114283937A (en) | Device for predicting kidney development risk of ANCA (acute coronary intervention) related small vasculitis and model training method | |
| WO2020111109A1 (en) | State visualization apparatus, state visualization method, and state visualization program | |
| Ravitch | Subjectivity in decision making: common problems and limitations | |
| CN116052889B (en) | sFLC prediction system based on blood routine index detection | |
| Murphy | Longitudinal studies 1: Determinants of risk | |
| Yördan et al. | Hybrid AI-Based Chronic Kidney Disease Risk Prediction |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |