30 results about "Survival modeling" patented technology

The present disclosure relates to a media attribution system that improves multi-channel media attribution by employing discrete-time survival modeling. In particular, the media attribution system uses event data (e.g., interactions and conversions) to generate positive and negative conversion paths, which the media attribution system uses to train an algorithmic attribution model. The media attribution system also uses the trained algorithmic attribution model to determine attribution scores for each interaction used in the conversion paths. Generally, the attribution score for an interaction indicates the effect the interaction has in influencing a user toward conversion.

The invention relates to a method for predicting an electric vehicle battery RUL (Remote Uplink) based on big data machine learning. The method is composed of a corresponding application architecture,a corresponding flow and a corresponding calculation model. The method comprises the following steps: firstly, acquiring battery real-time data in the operation process of an electric vehicle battery; and other operation data of the electric vehicle; performing data consolidation and cleaning, characteristic processing on the data; a model and a training verification algorithm are established through big data machine learning; wherein the modeling mainly uses a nonlinear hybrid algorithm model and a survival model, and the result is evaluated and optimized at different angles, so that a modelfor predicting the RUL of the electric vehicle battery is established, the maintenance and replacement of the battery are optimized, the safety index of a vehicle owner is improved, and the balance of system performance and economic benefits is achieved.

The invention relates to a method for predicting SOH of an electric vehicle battery based on big data machine learning. The method is composed of a corresponding application architecture, a corresponding flow and a corresponding calculation model. The method comprises the following steps: firstly, acquiring battery real-time data in the operation process of an electric vehicle battery; and other operation data of the electric vehicle; performing data consolidation and cleaning, characteristic processing is carried out on the data; a model and a training verification algorithm are established through big data machine learning; wherein the modeling mainly uses a nonlinear hybrid algorithm model and a survival model, and the result is evaluated and optimized at different angles, so that an SOH prediction model of the electric vehicle battery is established, the maintenance and replacement of the battery are optimized, the safety index of a vehicle owner is improved, and the balance of system performance and economic benefits is achieved.

The invention relates to an auxiliary assessment method for prognosis of nasopharynx cancer based on enhanced MRI radiomics. The auxiliary assessment method comprises the steps of: (1), performing MRIimage processing; (2), extracting imaging features; (3), screening the imaging features; (4), establishing a radiomics scoring formula; (5), screening clinical risk factors; and (6), establishing a prognostic survival model: establishing a prognostic observation model through combination of a radiomics score and the clinical risk factors of a patient with nasopharynx cancer, performing qualitative and quantitative prediction on the PFS (progression free survival) of the patient, and furthermore, assessing the performance of the prognostic survival model. The auxiliary assessment method in theinvention has little harm to the image examination of the patient; qualitative and quantitative analysis on the survival time of a specific patient is carried out; therefore, a doctor is assisted tomake an individualized treatment and follow-up visit scheme; furthermore, the doctor is assisted to assess the survival and recurrence time of the patient; simultaneously, the performance of the obtained prognostic survival model is verified; and thus, the accuracy of a prognostic prediction model is ensured.

The invention relates to a non-small cell lung cancer prognosis survival prediction method, a medium and electronic equipment. The method comprises the steps: obtaining a to-be-predicted CT image, carrying out the gray normalization of the to-be-predicted CT image, and extracting a region of interest; based on the region of interest, adopting a trained prognosis survival model based on deep learning to predict and obtain a corresponding prognosis lifetime classification result, wherein the prognosis survival model based on deep learning is a deep learning convolutional neural network model, and comprises five convolution blocks, a full connection layer and a classification layer, tumor abstract features are extracted layer by layer, a prognosis lifetime classification result is obtained, aBottleneck architecture is introduced into three convolution blocks in the middle of the five convolution blocks, and a fusion layer is added to the last convolution block on the basis of the Bttleneck architecture. Compared with the prior art, the method has the advantages of high prediction precision, convenience in implementation and the like.

The invention discloses a financial default risk prediction method and device based on GBST, electronic equipment and a computer readable medium. The financial default risk prediction method comprisesthe steps: initializing a basic survival tree of a GBST survival model based on a training data set; starting from the basic survival tree, carrying out optimization iteration by utilizing the survival probability predicted by the previous survival tree and the residual error of the real label so as to train to obtain the next survival tree until the total loss is smaller than a set threshold value; and for the input data of the new financial user, outputting a survival curve of the user by using a result of traversing the spanning tree by using the finally obtained survival tree, and predicting the default risk probability of each time period according to the survival curve. The financial default risk prediction method has a time dimension, can obtain the default time probability of eachcustomer, is high in prediction precision, and can process nonlinear heterogeneous data.

The invention provides an analytical method for prognosis survival case of non-small cell lung cancer. The analytical method for prognosis survival case of non-small cell lung cancer designs an experiment to perform prognosis survival analysis and research on a non-small cell lung cancer patient, constructs a prognostic analysis model for the non-small cell lung cancer based on CT image omics characteristics, according to a traditional imaging omics research framework, performing tumor segmentation, characteristic extraction, characteristic screening, and modeling of a correlation analysis andprognosis survival analysis model for image omics characteristics and the prognosis survival case on the data of the non-small cell lung cancer patient to obtain an image omics prognostic factor andprognosis survival analysis model correlated with prognosis survival significance of the patient with the non-small cell lung cancer, so as to provide the doctor with data information including the survival time of the patient and a series of late lesion development situations, and at the same time, to evaluate the performance of the prognosis survival model to ensure the accuracy of the prognosissurvival model.

The invention discloses a donation behavior and donor retention prediction method for a crowdfunding platform, comprising: crawling project data from the crowdfunding platform; dividing data intervalsaccording to time after performing feature extraction and preprocessing on the crawled project data; constructing a joint deep survival model by the donation behavior sequence and survival preservation sequence of the donor, and the features after the data interval division corresponding to the donor; and optimizing the joint deep survival model by combining the designed order constraint functionto predict the donation behavior and the survival maintenance state of the donor. By using the Joint Deep Survival Model (JDS) to model behaviors of the donors, the method may effectively predict whether the donors on the crowdfunding platform are lost, thereby reminding the platform to make measures to reduce the user lost rate, having certain practical application value, and bringing certain potential economic benefits to the platform.

The invention provides a financial risk management method based on GBST, and the method comprises the steps: obtaining a historical user data set, and building a training data set according to the historical user data set, wherein the training data set comprises user feature data, condition parameters and financial performance data; constructing a GBST survival model, and training the GBST survival model; inputting the user characteristic data and the condition parameters of the target user into a GBST survival model to output a prediction distribution curve of financial performance data of the target user; and based on the prediction distribution curve of the target user, customizing a risk management strategy corresponding to the target user according to a maximum profit judgment rule ofthe financial product. The financial risk management method takes profit maximization as a target, more personalized and more reasonable products can be formulated for users, existing risk managementstrategies of quota, pricing and periods can be optimized, and more profit is generated under the condition of controllable risk so as to realize profit maximization.

The invention provides a simulated survival ability judging method for a network information system. A work flow model for the system is established to describe the state of the structure and the performance of the system when, ideally, no failure occurs; a failure model and an error-tolerant model are established for resources with failure possibility to describe the failing process and reactions after the failure of a member; an information system survival model is established with combination of the two to describe the influence of certain failure resources on the whole system in a special network environment; a simulation method is adopted to inject safety events into the synthesized model, specifically including failure rate of the member, influence of invasive events on the member and survivability of the member; through statistics of accessible state information in SPN, survivability parameters of the information system service are computed to perform evaluation of the survival ability of the information system. The invention can analyze the completion quality of services of a network information system in a specific application environment according to the survival attributes of the construction members.

A survival model, for each of one or more pairs of genes, includes a function of a corresponding measure of the ratio of expression levels of the pairs of genes. For each pair of genes, there is a corresponding a cut-off value, such that patients are classified according to whether the corresponding measure is above or below the cut-off value. It is proposed (in an algorithm called “DDgR”) that the cut-off value should be selected so as to maximise the separation of the respective survival curves of the two groups of patients. It is further proposed that, for each of a number of genes or gene pairs, a selection is made from multiple survival models. The selection is according to whether a proportionality assumption is obeyed and/or according to a measure of data fit, such as the Baysian Information Criterion (BIC). Specific gene pairs identified by the methods are named.

The embodiment of the invention discloses a data processing method and a related device. The historical content push data of a target product having correlation with a to-be-promoted product is acquired for the to-be-promoted product, through the historical content push moment and the historical actual conversion moment in the historical content push data, information about how long the historical pushed object is converted after the content is pushed can be obtained, when a survival model is trained based on the object features of the historical pushed object, the data dimension of the conversion duration is introduced, the probability that the object is converted in a plurality of continuous time periods is concerned, the problem of conversion is converted into the problem of delayed conversion, learning of the survival model on delayed conversion in the time dimension is enhanced, decoupling of strong correlation between original historical content push data and a target product is achieved, and in a cold start stage, the first survival model obtained through training can temporarily replace a conversion rate model to provide prediction of the conversion rate for the product to be promoted.

The invention discloses a resource adjustment method, device and system based on a user survival model. The method comprises the steps: obtaining user behavior data which comprises financial behavior data, personnel behavior data and revenue behavior data; constructing a user risk judgment model and a user loss judgment model, and calculating a user risk index and a user loss index according to the user behavior data; establishing a coordinate system by taking the risk index as a horizontal coordinate and the loss index as a vertical coordinate, and determining the position of a user in the coordinate system according to the user risk index and the user loss index; and dividing quadrants, setting different resource adjustment strategies for different quadrants, and executing the corresponding resource adjustment strategy according to the quadrant to which the user belongs. By adopting the technical scheme, the behavior and the survival condition of the user can be pre-judged, the corresponding resource adjustment scheme can be selected according to the risk index and the loss index of the user, and intervention can be performed before the user is lost.

Methods and systems of fitting survival models to customer relationship management (CRM) data are presented. The models are used to compare productivity across business cohorts, optimize business sales processes, better plan for resource usage, and forecast events. Input to the systems include external CRM data along with other relevant business data, which are stored in a time-series database for processing.

The invention discloses a prediction method for clinical prognosis of a liver cancer patient based on machine learning. The method comprises the following steps: collecting corresponding sample data of hepatocellular carcinoma according to predetermined prediction characteristics; sorting the prediction features according to the sample data, and selecting part of the prediction features according to a sorting result; determining candidate prediction features related to the total survival rate from the selected prediction features by using a proportional risk regression model; and using a gradient enhanced survival classifier to select 20 optimal prediction features from the candidate prediction features as survival prediction factors to construct a liver cancer patient prognosis survival model. According to the invention, the HCC-related high-death-risk patients can be effectively identified.

The invention discloses a group of combined markers for predicting liver cancer prognosis and application thereof, and belongs to the technical field of biological medicines. Liver cancer is one of the most common cancer-related death reasons in the world and is extremely poor in prognosis, and recognition of effective prognosis biomarkers of liver cancer has important clinical significance. The invention provides a model for prognosis determination and risk assessment of a liver cancer patient based on combined biomarkers. The related combined biomarkers comprise CDCA3, CDCA8, SSRP1, HN1 and KIF4A. The prognosis risk score is quantitatively calculated according to the expression condition of the combined biomarker in a case sample, the patient is divided into a high-risk group and a low-risk group according to the median value of the risk score of the patient, and the result shows that the prognosis of the patient in the low-risk group is obviously superior to that of the patient in the high-risk group; and the accuracy and specificity of the prognosis survival model are verified through a K-Mplot survival curve, an ROC curve and the survival time and state of the patient. Therefore, the prediction model is of great significance to prognosis prediction and targeted therapy of patients with liver cancer.

A method for downsampling class-imbalanced sets with survival analysis comprising: acquiring a class-imbalanced data set, wherein the class-imbalanced data set comprises biological data from a plurality of subjects, wherein the biological data of each subject includes an observation, a time value, and a plurality of clinical measurements, and wherein the biological data is categorized as being part of a majority data class or a minority data class, wherein the majority data class has a greater number of observations than the minority data class; downsampling the class-imbalanced data set, wherein the downsampling results in the majority data class having an equivalent or substantially equivalent number of observations as the minority data class; and performing cross-validation on the downsampled data set with a survival analysis to generate a survival model, wherein the observation comprises an event or no event at a specific time value.

A method for performing, using a victim triage system, triage analysis of victims of an incident, comprising: (i) receiving a location of the incident, medical information, hospital capability information for hospitals in a predetermined vicinity of the location, and transport information relative to the location; (ii) determining, by a trained triage machine learning algorithm using the received information, a triage decision for the victims, wherein the triage decision for a victim comprises: (1) a probability of the victim's survival over time; (2) a recommendation to transport or not transport the victim to a hospital; and (3) to which of the two or more hospitals the victim should be transported; (iii) generating (140) a triage report comprising the determined triage decision for each of the plurality of victims; and (iv) displaying the triage report on a user display of the victim triage system.

A method of fraud prevention using deep learning and survival models is provided. The method may include receiving, with at least one processor, transaction data associated with a plurality of transactions of at least one payment account; at least one attempted attack may be detected based on the transaction data. A fraud risk score for each of a plurality of sub-periods in a time period following the at least one attempted attack may be generated based on the transaction data using a deep learning model and a survival model. The fraud risk score for each respective sub-period may be associated with a probability that the respective sub-period does not have a fraudulent transaction. A system and a computer program product are also disclosed.