11 results about "Survival analysis" patented technology

The application discloses a customer demand response prediction method and system based on double-model multi-source information fusion. The application integrates customer historical purchase, product price and regional market supply and demand multi-source data to construct a three-dimensional time series data set. Based on a sliding window, a customer daily sample sequence is generated to construct a survival analysis prediction framework. The core is to construct a double-model parallel prediction architecture. The first model is a survival analysis model based on an attention mechanism, which is used to predict the conditional risk sequence of customer future daily purchase. The second model is a double-head neural factor decomposition machine model, which is used to calculate the interest score of the customer for each product. The two are fused to generate the customer's daily purchase demand risk prediction results in the future multiple prediction days, and then the customers with purchase intention and their predicted purchase days are screened out. The application combines time dynamics and product preferences, and improves the accuracy and interpretability of demand prediction.

The present application belongs to the technical field of user operation, and specifically relates to a user operation method based on a survival analysis model, which comprises the following steps: S1, data definition and mathematical modeling; S2, descriptive analysis of user survival characteristics; S3, comparative analysis of the differences of different user groups; S4, modeling and prediction of user behavior influencing factors; and S5, application of analysis results in operation decision-making. The present application can realize the dynamicization and refinement of user insight, provide objective and quantitative user stratification standards, empower precise operation and resource optimization, and improve the effectiveness of user loss early warning.

This invention discloses a system for predicting the risk of recurrence after lumbar disc herniation surgery based on domain offset correction, relating to the fields of medical data processing and clinical disease risk prediction technology. The system acquires multimodal clinical and imaging feature vectors and follow-up outcome data from training and independent validation cohorts. It trains a center classifier to quantify and correct for differences in multi-center data distribution, calculates and normalizes the importance weights of each training case, trains a survival analysis base learner based on the processed weights, and performs stacking and ensemble integration. The final deployment model is determined by combining the independent validation case cohort. The final deployment model is then used to infer the risk of recurrence in target patients after lumbar disc herniation surgery, outputting individualized recurrence risk scores, risk stratification, and recurrence-free survival curves. This improves the robustness and reliability of the risk prediction system in cross-center scenarios and is suitable for follow-up management and clinical decision support after lumbar disc herniation surgery.

The present application relates to the cross field of intelligent transportation, human factors engineering, emergency management and artificial intelligence, and particularly relates to a fire driving fatigue monitoring system and method based on survival analysis, which fuses multi-modal data such as physiology, behavior, environment and task, constructs a time-dependent dynamic risk assessment model, introduces a survival analysis framework to quantitatively determine the probability of accidents caused by fatigue during task execution, and identifies fatigue critical points with intervention value in combination with an adaptive threshold mechanism, thereby providing scientific rest determination and takeover decision basis for the task management system. The present application solves the problems of existing fatigue recognition technology in special operation scenarios, such as hysteresis, limitations and lack of critical risk judgment ability; it takes into account real-time, predictability and deployability, and is suitable for high-risk driving environments such as fire fighting, police and dangerous chemical transportation.

The application provides a tea garden risk prediction method and system based on multi-modal data, comprising: based on risk correlation constraints, high-dimensional survival analysis and contribution degree filtering and noise removal are performed on a tea garden multi-modal data set, and space-time feature parallel extraction is performed, to obtain a multi-modal space-time feature set; based on a dynamic causal discovery algorithm, loop-free causal constraints are performed on the multi-modal space-time feature set, to obtain a modal causal graph; a time delay response kernel matrix is extracted from a lag causal matrix of the modal causal graph, the time delay response kernel matrix is taken as a prior constraint, cross-modal feature fusion is performed on the multi-modal space-time feature set based on a causal attention mechanism, to obtain a fused multi-modal feature; based on a hybrid expert architecture, multi-task risk analysis is performed on the fused multi-modal feature, to obtain a comprehensive risk vector; contribution degree attribution is performed on the comprehensive risk vector, to obtain a contribution heat map, cross verification is performed on the comprehensive risk vector in combination with the modal causal graph, and a tea garden risk report is generated.

The application discloses an analysis method, system and device based on cancer cross-modal data and a medium, and comprises the following steps: in response to receiving an analysis request input by a user, inputting the analysis request to a first large model, determining an analysis task and an analysis tool for executing the analysis task; in the case where the analysis task comprises a target analysis task, extracting information from clinical data input by the user by using an information extraction engine to obtain object basic data; obtaining a gene expression matrix formed by cancer transcriptome data after dimension reduction based on the object basic data by using a wisdom algorithm contraction engine; performing survival analysis on the gene expression matrix after dimension reduction by using a risk mapping engine to obtain a survival analysis result and a target gene set affecting survival; and checking the target gene set based on reference data by using an arbitration evaluation engine in the analysis tool to obtain a checking result.

The application relates to the technical field of digital pathology and artificial intelligence, in particular to an immunohistochemical efficacy prediction system based on dynamic analysis. The system comprises an image processing module, which is used for acquiring and preprocessing paired immunohistochemical digital pathology sections before and after treatment of the same patient; a dynamic registration and analysis module, which is used for high-precision image registration of the paired sections and extraction of dynamic quantitative parameters of biomarker expression changes with treatment; a spatial heterogeneity quantification module, which is used for dividing tumor functional subareas and quantifying spatial distribution heterogeneity parameters of biomarkers in different functional subareas; and a fusion prediction module, which is used for feature fusion of the dynamic quantitative parameters and the spatial distribution heterogeneity parameters and output of a clinical efficacy prediction result through a survival analysis model. The system realizes accurate prediction of tumor immunotherapy efficacy by performing dynamic and quantitative analysis on paired immunohistochemical sections before and after treatment and integrating multi-modal data.

This invention relates to a medical image classification method and system based on adaptive conformal training, belonging to the fields of medical image analysis and artificial intelligence technology. First, image data is grouped and multi-scale features are extracted. Second, a lightweight thresholding network is designed to generate adaptive thresholds for samples, and the difference between the predicted probability and the threshold is measured by the residual inconsistency score. Differentiable soft quantile estimation is used to calibrate the residual scores, and end-to-end optimization is performed using soft set loss, set size loss, and cross-entropy loss. During the inference phase, a prediction set is constructed based on the calibrated thresholds. This method effectively solves the problem that traditional conformal prediction with single thresholds and single-scale features cannot adapt to complex lesions. It significantly reduces the size of the prediction set and improves classification accuracy while maintaining coverage, and is widely applicable to clinical diagnostic scenarios such as pathological slide classification, tumor grading, and survival analysis.

The application relates to a customer life cycle prediction method and system fusing LSTM and survival analysis, which comprises the following steps: obtaining time sequence behavior data and user attribute data of a user; converting discrete features in the user attribute data into dense vectors through embedding coding, fusing the dense vectors with the time sequence behavior data, and generating a fusion feature sequence; inputting the fusion feature sequence into an LSTM neural network model to output behavior prediction features of the user in a preset future period; performing statistical calculation on the time sequence behavior data to construct a comprehensive feature set; inputting the comprehensive feature set into a Cox proportional hazards model to output user churn risk features; calculating a prediction consistency coefficient based on the behavior prediction features and the churn risk features; predicting a marketing conversion rate by using an XGBoost model, and calculating SHAP values of each feature; combining the prediction consistency coefficient, dynamically adjusting a marketing strategy weight related to the features according to the SHAP values, and generating an optimized precise marketing strategy. The application improves the precision of customer life cycle prediction.

The application discloses a sintering trolley overhaul cycle prediction method and system, and relates to the technical field of sintering, and comprises the following steps: acquiring trolley full-life cycle multi-source physical data, wherein the trolley full-life cycle multi-source physical data comprises static attribute data, dynamic operation data and historical maintenance data; extracting trolley exclusive feature vectors; taking historical trolley exclusive feature vectors as input, taking survival time labels and trolley event state labels as training labels, and training to obtain a converged survival analysis model; inputting the exclusive feature vectors of in-service trolleys into the trained survival analysis model for prediction, and outputting corresponding no-overhaul operation survival curves; and calculating residual no-overhaul operation time and a comprehensive health index based on the no-overhaul operation survival curves. The method realizes a fundamental change from regular maintenance to one-trolley-one-strategy and single-trolley predictive maintenance, and significantly improves equipment life prediction accuracy and operation and maintenance decision scientificity.