Cloud computing-based tumor care data processing system

Abstract

The application relates to the technical field of medical data processing, in particular to a tumor nursing data processing system based on cloud computing, which comprises the following steps: a multi-scale feature analysis module acquires tumor nursing data, the tumor nursing data are disassembled into multiple time dimensions, multi-scale features of the tumor nursing data are extracted, the multi-scale features of the tumor nursing data are sorted according to importance, and a preliminary tumor nursing data feature alignment result is acquired. In the application, feature decomposition and analysis of multiple time scales are carried out on the tumor nursing data, the feature changes in different time dimensions are effectively captured, and the comprehensiveness of feature recognition is enhanced. The important features are sorted and dynamically adjusted and matched, the features with high correlation are combined and weightedly fused, meanwhile, distributed optimization is utilized to improve the large-scale data processing efficiency, and the accuracy and adaptability of data integration are significantly improved.

Description

Technical Field

[0001] This invention relates to the field of medical data processing technology, and in particular to a cloud-based tumor care data processing system. Background Technology

[0002] Medical data processing technology encompasses methods for collecting, storing, analyzing, and visualizing large amounts of healthcare-related data. This field primarily includes electronic health records (EHRs), medical image processing, clinical decision support systems, remote monitoring, AI-assisted diagnosis, and personalized treatment planning. Its core objective is to improve the efficiency, accuracy, and personalization of healthcare services through data-driven approaches, ultimately enhancing patient health management.

[0003] Among them, the oncology nursing data processing system is a data processing and management system specifically designed for the field of oncology nursing, optimizing the flow, analysis, and application of data related to the care of oncology patients. Its main uses include collecting patients' treatment and nursing records, analyzing nursing outcomes, and providing data support to assist in the development of personalized care plans.

[0004] Existing technologies struggle to comprehensively capture the characteristics of nursing data across different time scales, neglecting the changing patterns of short-term and long-term features. For example, when analyzing patient nursing responses, short-term fluctuations may be perceived as noise, while long-term trends may not be accurately reflected, limiting the precise assessment of nursing outcomes. The integration of cross-patient data fails to adequately consider the correlations and importance rankings between features, easily leading to feature fusion results that cannot support the accurate analysis of complex nursing data. For instance, the correlations between key indicators across different patients may be diluted, affecting the scientific validity of feature combinations. Regarding storage, existing technologies do not rationally stratify data according to data type and feature importance, often resulting in uneven resource allocation, inefficient storage of critical data, and the consumption of more storage space for non-critical data, leading to inefficient data management. The failure to update dynamic changes in nursing data in real time makes it difficult to capture new response patterns emerging during the nursing process, ultimately affecting the timeliness and accuracy of nursing decisions. Summary of the Invention

[0005] The purpose of this invention is to address the shortcomings of existing technologies by proposing a cloud computing-based tumor care data processing system.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: A cloud computing-based tumor nursing data processing system includes:

[0007] The multi-scale feature analysis module acquires tumor nursing data, decomposes the tumor nursing data into multiple time dimensions, extracts multi-scale features of the tumor nursing data, ranks the importance of the multi-scale features of the tumor nursing data, and obtains preliminary tumor nursing data feature alignment results.

[0008] The dynamic feature alignment adjustment module, based on the preliminary tumor nursing data feature alignment results, filters matching feature combinations according to a preset correlation threshold, dynamically updates the matching relationship between features, and obtains cross-patient feature fusion results.

[0009] Based on the cross-patient feature fusion results, the non-uniform gridded storage module analyzes the multi-scale features of the fused tumor nursing data, maps inconsistent data to matching storage layers, and obtains the data storage hierarchy division results.

[0010] The adaptive data partitioning and optimization module monitors real-time changes in tumor nursing data based on the data storage hierarchy partitioning results, identifies the patient's response patterns during the nursing process, dynamically updates the data storage hierarchy, and generates tumor nursing data processing results.

[0011] As a further aspect of the present invention, the steps for obtaining the multi-scale features of the tumor nursing data are specifically as follows:

[0012] Based on patient monitoring data and nursing records, the data is broken down according to the time dimension, and the range of changes in monitoring data, the frequency of nursing interventions and the duration of treatment response are extracted to generate corresponding feature data tables.

[0013] Based on the corresponding feature data table, the range of each feature value is organized, and the features are normalized to adjust the magnitude of the feature values ​​and obtain a standardized feature dataset.

[0014] As a further aspect of the present invention, the steps for obtaining the preliminary tumor nursing data feature alignment results are specifically as follows:

[0015] For the standardized feature dataset, the formula is as follows:

[0016]

[0017] Calculate the importance score I for a single feature i i The results of the feature importance analysis were obtained.

[0018] Among them, f i It is the value of feature i after normalization, r i It is a correlation factor, representing the strength of the correlation between feature i and the nursing goal, d i It is the sampling interval time of feature i;

[0019] Based on the feature importance analysis results, the features are sorted from high to low scores, and the order of the corresponding features in the data table is adjusted simultaneously to obtain preliminary feature alignment results for tumor nursing data.

[0020] As a further aspect of the present invention, the step of obtaining the feature combination for screening and matching specifically includes:

[0021] Based on the preliminary tumor nursing data feature alignment results, features such as trend change rate, nursing intervention frequency change rate, and treatment response duration across patient data are extracted according to importance ranking. Missing feature values ​​are imputed and data alignment is performed to generate a feature integration set across patients.

[0022] Based on the aforementioned cross-patient feature integration set, the following formula is used:

[0023]

[0024] Calculate the Pearson correlation coefficient between features i1 and i2 in the feature pair. Obtain the matching feature combination;

[0025] Where i 1j i 2j These are the values ​​of features i1 and i2 in the j-th sample. is the mean of features i1 and i2, and n is the total number of patients or data points.

[0026] As a further aspect of the present invention, the step of obtaining the cross-patient feature fusion result specifically includes:

[0027] Based on the highly relevant feature pairs selected from the matched feature combinations, feature data from multiple periods are extracted, and the features are classified and weighted according to time characteristics. The feature matching relationship is dynamically adjusted to generate dynamically updated cross-time scale feature matching results.

[0028] Based on the dynamically updated cross-timescale feature matching results, and combined with the cloud computing environment, the feature data is distributed to multiple nodes for parallel processing, local optimization and global integration are performed, and feature matching verification and result optimization are conducted to obtain cross-patient feature fusion results.

[0029] As a further aspect of the present invention, the step of obtaining the data storage hierarchy partitioning result specifically includes:

[0030] Based on the cross-patient feature fusion results, physiological indicators, treatment plans, monitoring data and nursing records of tumor nursing data are classified and extracted. The data are mapped to inconsistent storage layers according to data characteristics and storage requirements, and the mapping is verified to obtain matching storage layer mapping information.

[0031] Based on the matched storage layer mapping information, a corresponding storage mapping relationship is constructed, the storage rules and resource allocation of the storage layer are optimized, abnormal allocations are adjusted, and data storage hierarchy division results are generated.

[0032] As a further aspect of the present invention, the step of obtaining the identification of the patient's response pattern during the nursing process specifically includes:

[0033] Based on the data storage hierarchy division results, real-time tumor nursing data, including physiological indicators, monitoring data and nursing records, are extracted. By real-time monitoring and analysis of the dynamic changes in the data, the real-time tumor nursing data change analysis results are obtained.

[0034] Based on the analysis results of the real-time tumor care data changes, the following formula is used:

[0035]

[0036] The magnitude of change (B) of the patient's physiological indicators and the rate of change (F) of the frequency of nursing interventions were calculated respectively, and the analysis results of the patient's response pattern were generated.

[0037] Where T is the total number of data points within the time period, and V t It is the physiological indicator value at time point t. ΔN represents the average physiological index within the monitoring period, ΔN represents the change in the number of nursing interventions within the period, and ΔT represents the time difference between two adjacent records.

[0038] As a further aspect of the present invention, the steps for obtaining the tumor nursing data processing results are specifically as follows:

[0039] Based on the analysis results of the patient response patterns, the patient data that has been divided into multiple storage levels is extracted and reanalyzed. In view of the latest situation of changes in heart rate fluctuation amplitude and nursing frequency, the patient data is dynamically adjusted and the storage levels are re-divided to obtain updated storage level data.

[0040] Based on the updated storage hierarchy data, real-time patient data from multiple priority categories are integrated to analyze individual and group nursing status and generate tumor nursing data processing results.

[0041] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0042] This invention effectively captures feature changes across different time dimensions by performing multi-timescale feature decomposition and analysis on tumor nursing data, enhancing the comprehensiveness of feature recognition. Normalization of features from different patients improves the comparability and consistency of cross-patient data, eliminating interference from differences in feature magnitude on the analysis results. Ranking important features and dynamically adjusting matching relationships, weighted fusion of highly correlated feature combinations, along with distributed optimization, significantly improves the efficiency of large-scale data processing, resulting in enhanced accuracy and adaptability of data integration. A storage mapping relationship is constructed based on data type and importance, optimizing storage efficiency and resource utilization, ensuring that nursing data storage is highly consistent with actual application needs. By identifying response patterns in the nursing process in real time and dynamically adjusting the data storage hierarchy, data processing and updates can respond promptly to nursing needs, providing data support for tumor nursing. Attached Figure Description

[0043] Figure 1 This is a system flowchart of the present invention;

[0044] Figure 2 This is a flowchart illustrating the multi-scale features of tumor nursing data acquired in this invention.

[0045] Figure 3 This is a flowchart illustrating the process of obtaining preliminary tumor nursing data feature alignment results according to the present invention;

[0046] Figure 4 This is a flowchart illustrating the process of filtering and matching feature combinations according to the present invention;

[0047] Figure 5 This is a flowchart illustrating the process of obtaining cross-patient feature fusion results according to the present invention;

[0048] Figure 6 This is a flowchart illustrating the process of obtaining the data storage hierarchy division results for this invention.

[0049] Figure 7 A flowchart illustrating the patient response patterns observed during the nursing process, as described in this invention;

[0050] Figure 8 This is a flowchart illustrating the process of obtaining tumor nursing data processing results according to the present invention. Detailed Implementation

[0051] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0052] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0053] Please see Figure 1 The cloud-based tumor care data processing system includes:

[0054] The multi-scale feature analysis module acquires tumor nursing data, decomposes the tumor nursing data into multiple time dimensions, extracts local features for different time scales, including the changing trend of patient monitoring data, the frequency of nursing intervention, and the duration of treatment response. The extracted features are normalized, the feature magnitude of nursing events among patients is adjusted, multi-scale features of tumor nursing data are extracted, the importance of multi-scale features of tumor nursing data is ranked, and preliminary tumor nursing data feature alignment results are obtained.

[0055] The dynamic feature alignment adjustment module integrates the features of tumor nursing data across patients based on the preliminary feature alignment results of tumor nursing data and the multi-scale features of tumor nursing data after importance ranking. It selects matching feature combinations according to the preset relevance threshold, dynamically updates the matching relationship between features, and performs weighted fusion of the matching relationship at different time scales. Combined with the cloud computing environment, it performs distributed optimization of the weighted fusion process to obtain the feature fusion results across patients.

[0056] The non-uniform gridded storage module analyzes the multi-scale features of the fused tumor nursing data based on the cross-patient feature fusion results. It considers the importance of the multi-scale features of tumor nursing data in the nursing process, as well as the types of tumor nursing data, including patients' physiological indicators, treatment plans, monitoring data, and nursing records. It maps inconsistent data to matching storage layers, optimizes data storage efficiency and resource utilization by constructing corresponding storage mapping relationships, and obtains the data storage hierarchy division results.

[0057] The adaptive data partitioning and optimization module monitors real-time changes in tumor nursing data based on the data storage hierarchy division results, identifies the response patterns exhibited by patients during the nursing process, including fluctuations in physiological indicators and changes in the frequency of nursing interventions. Through continuous analysis of patient data, the data storage hierarchy is dynamically updated to reflect the latest nursing information and generate tumor nursing data processing results.

[0058] Preliminary results of tumor nursing data feature alignment include normalized features of patient monitoring data trends, time-scale features of nursing intervention frequency, and distribution features of treatment response duration. Cross-patient feature fusion results include matching feature groups of different patients in treatment response, integrated distribution after feature weight adjustment, and time-scale adaptation results based on weighted processing. Data storage hierarchy partitioning results include high-priority data storage layer, secondary-priority data cache layer, and low-frequency data storage layer. Tumor nursing data processing results include data partitioning adjustment results reflecting the latest patient nursing status, real-time updated hierarchical storage distribution, and dynamically identified patient group feature patterns.

[0059] Please see Figure 2 The specific steps for obtaining multi-scale features of tumor nursing data are as follows:

[0060] Based on patient monitoring data and nursing records, the data is broken down according to the time dimension, and the range of changes in monitoring data, the frequency of nursing interventions and the duration of treatment response are extracted to generate corresponding feature data tables.

[0061] To acquire oncology nursing data, the data is broken down into multiple time dimensions. The data is then processed using a cloud computing platform. First, the patient's monitoring data is uploaded to cloud storage. For example, a patient's vital signs data over one week include heart rate, blood pressure, and blood oxygen concentration, recorded 24 times a day, once per hour, for a total of 168 data points. Nursing intervention data is recorded as 1-3 interventions per day, for a total of 15 interventions. Treatment response data is recorded as the duration of the treatment response, with the response duration after each intervention ranging from 2 to 4 hours. The data is broken down by time scale using the layered processing module of the cloud computing platform: Hourly: Calculate the hourly fluctuation range of heart rate, blood pressure, and blood oxygen saturation. For example, the heart rate fluctuates between 72-85 bpm in the first hour, with an amplitude of 13 bpm. Daily: Calculate the average of vital signs data daily. For example, the daily average heart rate is 78 bpm, the average blood pressure is 120 / 80 mmHg, and the average blood oxygen saturation is 96%. The frequency of nursing interventions is also statistically analyzed. For example, the intervention frequency is 2 times on a certain day. Weekly: Summarize the duration of treatment responses and calculate the average duration of treatment responses per week. For example, the total treatment response time is 45 hours, and the average duration of each response is 3 hours. The total frequency of nursing interventions and their changing trends within a week are also recorded. Through these steps, a data table containing hourly, daily, and weekly scale features is finally generated and stored in the cloud platform's analysis database for subsequent processing and analysis.

[0062] Based on the corresponding feature data table, the range of each feature value is organized, and the features are normalized to adjust the magnitude of the feature values ​​and obtain a standardized feature dataset.

[0063] The extracted multi-scale feature data were processed. For example, hourly data showed heart rate fluctuations of 13 bpm, 10 bpm, and 15 bpm, blood pressure fluctuations of 20 mmHg, 25 mmHg, and 22 mmHg, and blood oxygen concentration fluctuations of 2%, 3%, and 4%. Daily data showed average heart rate ranging from 70 to 85 bpm, average blood pressure from 110 to 130 mmHg, and average blood oxygen concentration from 95% to 98%. Weekly data showed the average duration of treatment response from 2.5 to 3.5 hours. To adjust the magnitude of each feature, the hourly, daily, and weekly data were normalized within each dimension. The heart rate fluctuation was adjusted to a relative range (e.g., between 0 and 1), and the blood pressure and blood oxygen concentration fluctuations were similarly scaled proportionally to their relative ranges. This process aligned the feature magnitudes of different patients, forming a unified multi-scale feature set for subsequent analysis. Finally, the processed feature set is saved as a standardized format file in the cloud platform.

[0064] Please see Figure 3The specific steps for obtaining the preliminary tumor nursing data feature alignment results are as follows:

[0065] For standardized feature datasets, the formula is used:

[0066]

[0067] Calculate the importance score I for a single feature i i The results of the feature importance analysis were obtained.

[0068] Among them, f i It is the normalized value of feature i, representing the dimensionless numerical value of a specific feature after standardization. i This is a correlation factor, representing the strength of the correlation between feature i and the nursing goal, ranging from 0 to 1. It is determined through data analysis; the larger the value, the stronger the correlation between the feature and the goal. It is determined by analyzing the correlation between changes in the feature and key patient indicators (such as treatment effectiveness). i It is the sampling interval time of feature i, representing the collection period of specific feature data, which is directly determined by the collection plan or data recording frequency.

[0069] For example, the following characteristic data are available: trend change rate f1 = 0.6, correlation factor r1 = 0.9, sampling interval d1 = 1 hour, nursing intervention frequency change rate f2 = 0.8, correlation factor r2 = 0.7, sampling interval d2 = 24 hours, treatment response duration f3 = 0.4, correlation factor r3 = 0.5, sampling interval d3 = 168 hours.

[0070] Calculate the importance score of the rate of change of trend:

[0071]

[0072] Calculate the importance score of the rate of change in nursing intervention frequency:

[0073]

[0074] The importance score for calculating the duration of treatment response:

[0075]

[0076] The results showed that the importance score of the trend change rate was 0.54, the importance score of the nursing intervention frequency change rate was 0.0233, and the importance score of the duration of treatment response was 0.0012.

[0077] Based on the feature importance analysis results, the features are sorted from high to low scores, and the order of the corresponding features in the data table is adjusted accordingly to obtain preliminary feature alignment results for tumor nursing data.

[0078] The importance of multi-scale features in tumor nursing data was ranked. Based on the calculated importance scores of individual features, the trend change rate I1 = 0.54, the nursing intervention frequency change rate I2 = 0.0233, and the treatment response duration I3 = 0.0012 were arranged in descending order. The results were: trend change rate (0.54) > nursing intervention frequency change rate (0.0233) > treatment response duration (0.0012). According to the ranking results, the original feature set was rearranged in the above order, the index information corresponding to each feature was recorded, and the column order of the data table was adjusted accordingly.

[0079] Please see Figure 4 The specific steps for obtaining the matching feature combinations are as follows:

[0080] Based on the preliminary feature alignment results of tumor nursing data, features such as trend change rate, nursing intervention frequency change rate, and treatment response duration were extracted across patient data according to importance ranking. Missing feature values ​​were imputed and data alignment was performed to generate a feature integration set across patients.

[0081] First, extract the features with the highest importance, such as the trend change rate, the change rate of nursing intervention frequency, and the duration of treatment response. For each feature, extract the corresponding feature value from the data of different patients one by one. For example, for patients A, B, and C, the extracted trend change rates are 0.6, 0.8, and 0.7, the change rates of nursing intervention frequency are 0.3, 0.4, and 0.2, and the durations of treatment response are 5 hours, 4 hours, and 6 hours, respectively. Arrange these feature values ​​in order of trend change rate, change rate of nursing intervention frequency, and duration of treatment response. 0.6, 0.3, 5 (Patient A), 0.8, 0.4, 4 (Patient B), 0.7, 0.2, 6 (Patient C); During the integration process, for patients with missing feature values ​​(such as Patient D missing the nursing intervention frequency change rate), imputation is performed using the mean of adjacent patients (Patients A and B). For example, if the mean of the nursing intervention frequency change rate is calculated to be 0.35, this value is added to the nursing intervention frequency change rate field of Patient D to ensure the integrity of all patient feature sets. Finally, the integrated patient feature set is saved as aligned multi-patient feature data.

[0082] Based on a cross-patient feature integration set, the formula is used:

[0083]

[0084] Calculate the Pearson correlation coefficient between features i1 and i2 in the feature pair. Obtain the matching feature combination;

[0085] in, The range of values ​​is -1 to 1, with positive values ​​indicating positive correlation, negative values ​​indicating negative correlation, and absolute values ​​close to 1 indicating strong correlation. 1j i 2j These are the values ​​of features i1 and i2 in the j-th sample (e.g., a patient). These are the means of features i1 and i2, representing the average levels of features i1 and i2 across all samples. The calculation formula is: n is the total number of patients or data points. The deviation of an eigenvalue from the mean measures the magnitude of the eigenvalue's fluctuation relative to the overall level. The square of the deviation of the eigenvalue from the mean is used to measure the squared contribution of the deviation.

[0086] For example, the characteristic data of 5 patients are as follows: trend change rate (characteristic x): 0.6, 0.7, 0.8, 0.5, 0.9; nursing intervention frequency change rate (characteristic y): 0.3, 0.4, 0.5, 0.2, 0.6.

[0087] Calculate the mean rate of change of the trend:

[0088]

[0089] Calculate the mean rate of change in the frequency of nursing interventions:

[0090]

[0091] Substitute into the formula:

[0092]

[0093]

[0094] Calculate the correlation coefficient:

[0095]

[0096] The results showed that there was a perfect positive correlation (correlation coefficient of 1) between the rate of change of the trend and the rate of change of the frequency of nursing interventions. A positive correlation indicates that features i1 and i2 change in the same direction. A negative correlation indicates that features i1 and i2 change in opposite directions. A strong correlation indicates a close linear relationship between the features. There is no obvious linear correlation. For high correlation (strong linear relationship): a threshold is usually set. This indicates that the feature pairs are strongly correlated and have analytical value. For weak correlations (weak linear relationships): when When the correlation is low, the contribution to the analysis may be limited, and it can be considered for exclusion. For example, suppose the correlation coefficients of the three feature pairs are as follows: (Rate of change in trend and rate of change in frequency of nursing interventions), (Trend rate of change and duration of treatment response), (Change rate of nursing intervention frequency and duration of treatment response), set a threshold. The filtered results are as follows: Keep: It meets the threshold condition (positive correlation and strong linear relationship). Eligible items will be removed if they meet the threshold criteria (negative correlation but strong linear relationship). Features below a threshold (weak linear relationship). Feature pairs retained through threshold filtering can be used as key data combinations for subsequent fusion and optimization, while feature pairs removed due to insufficient correlation will not participate in further analysis, thereby improving the efficiency and accuracy of feature analysis.

[0097] Please see Figure 5 The specific steps for obtaining cross-patient feature fusion results are as follows:

[0098] Based on the highly relevant feature pairs selected from the matched feature combinations, feature data from multiple periods are extracted, and the features are classified and weighted according to time characteristics. The feature matching relationship is dynamically adjusted to generate dynamically updated cross-time scale feature matching results.

[0099] Based on the calculated correlation results, the matching relationship between features in tumor nursing data is dynamically updated. Assuming the goal is to monitor the matching between treatment response and nursing intervention effect in different patients, the trend change rate (short-term feature) and treatment response duration (long-term feature) are first selected from the highly correlated feature pairs. Feature values ​​are extracted for different patients. For example, patient A has a short-term trend change rate of 0.6 and a treatment response duration of 5 hours; patient B has a short-term trend change rate of 0.8 and a treatment response duration of 6 hours; patient C has a short-term trend change rate of 0.7 and a treatment response duration of 4 hours. These data are classified into short-term feature group and long-term feature group, respectively, and the feature values ​​are normalized to make the numerical distribution range of different features consistent. For example, the short-term trend change rate is normalized to 0.75, 1, and 0.88. After feature normalization, weighted values ​​are assigned to each pair of features. Based on the temporal characteristics of the features, short-term features are weighted at 0.7, and long-term features at 0.3. The weighting is dynamically adjusted according to the real-time needs of nursing care. For example, short-term features, reflecting the timeliness of nursing care, have higher weights, while long-term features, used to observe overall trends, have relatively lower weights. Then, a weighted fusion value is calculated for each patient. For example, patient A's fusion value is calculated as 0.75×0.7+5×0.3=1.575, patient B's is 1×0.7+6×0.3=1.90, and patient C's is 0.88×0.7+4×0.3=1.516. After calculating the weighted fusion values ​​for all patients, the weighted fusion values ​​are sorted, and matching relationships are generated according to the feature matching results. For example, patient B has the highest feature combination score and the highest matching priority, followed by patients A and C. The dynamically updated feature matching relationship results provide a quantitative basis for the comprehensive analysis of short-term and long-term features, used for subsequent data integration and optimization processing.

[0100] Based on dynamically updated cross-timescale feature matching results, and combined with a cloud computing environment, feature data is distributed to multiple nodes for parallel processing, local optimization and global integration are performed, and feature matching verification and result optimization are conducted to obtain cross-patient feature fusion results.

[0101] The weighted fusion process is optimized in a distributed manner using a cloud computing environment. Following the dynamic matching results completed in paragraph 1, the weighted fusion values ​​and corresponding feature data of patients are uploaded to the cloud computing environment. To optimize the efficiency of distributed processing, the data is first divided into different computing tasks according to patient feature groups. For example, the short-term and long-term features of patients A, B, and C are assigned to different nodes for processing. Each node performs local optimization on the assigned patient data, including verifying the accuracy of the weighted fusion results, checking for abnormal data points, and further adjusting the weighting values ​​to adapt to the global optimization needs. For example, the weights are increased for patients with large fluctuations in short-term features. After the distributed computing nodes complete their local tasks, the system aggregates the intermediate results from each node and integrates them uniformly through the central node. During the integration process, the fusion rules are dynamically adjusted according to the weight distribution of patient feature values. For example, priority fusion is given to patient groups with higher short-term feature importance, and the priority list of feature matching is recalculated based on the aggregated data. After the integration is completed, the cloud computing environment verifies the overall fusion results. For example, historical data is used to backtest the matching results and evaluate the effectiveness of the optimized fusion results in prediction and analysis. Finally, the optimized cross-patient feature fusion results are output for downstream analysis tasks, such as treatment effect evaluation and nursing strategy adjustment.

[0102] Please see Figure 6 The specific steps for obtaining the data storage hierarchy partitioning results are as follows:

[0103] Based on the cross-patient feature fusion results, physiological indicators, treatment plans, monitoring data and nursing records of tumor nursing data are classified and extracted. The data are mapped to different storage layers according to data characteristics and storage requirements, and the mapping is verified to obtain matching storage layer mapping information.

[0104] In a cloud computing environment, the specific process of mapping oncology nursing data to different storage layers is as follows: By analyzing the fused oncology nursing data, the data is classified according to feature types, such as: Monitoring data: real-time heart rate (80 bpm), blood pressure (120 / 80 mmHg), body temperature (36.5℃ to 37.2℃). Treatment plans: drug dosage (50 mg), administration time (8:00 AM). Nursing records: nursing intervention time (2024-11-19 15:00) and nursing type (routine examination). Matching rules are defined according to storage needs and data characteristics: High-speed storage layer: used to store real-time data, supporting fast access and high-frequency reading, such as monitoring data. Medium-speed storage layer: used to store periodically updated data, such as treatment plans. Low-speed storage layer: used to store archived or long-term stored data, such as nursing records. Data mapping process: Monitoring data: heart rate (80 bpm) and body temperature (36.5℃ to 37.2℃), as real-time data, are directly mapped to the high-speed storage layer for rapid response to nursing needs. Treatment plan: Drug dosage (50mg) and administration time (8:00 AM) are periodically updated data, mapped to the medium-speed storage layer to meet the moderate requirement of data modification frequency. Nursing records: Nursing intervention time and type are historical data, mapped to the low-speed storage layer for subsequent long-term archiving and analysis. The system performs consistency verification on the storage mapping, such as checking whether monitoring data is completely mapped to the high-speed layer, whether the treatment plan is updated on time, and readjusting any data omissions or incorrect storage allocations. During verification, if nursing records are found to be incorrectly allocated to the high-speed layer, they are remapped to the low-speed layer according to rules. For example, patient A's real-time heart rate and temperature data are stored in the high-speed storage layer and subsequently accessed multiple times in nursing operations to support real-time nursing decisions; patient B's nursing records are stored in the low-speed storage layer to assess the overall trend of historical nursing interventions. Through reasonable allocation of storage layers, the system ensures efficient data access and optimized utilization of storage resources, providing support for nursing decision-making and data management.

[0105] Based on the matched storage layer mapping information, the corresponding storage mapping relationship is constructed, the storage rules and resource allocation of each storage layer are optimized, abnormal allocation is adjusted, and the data storage layer division result is generated.

[0106] First, clear storage rules are defined for each storage layer. For example, high-speed storage layers are allocated greater bandwidth and faster access speeds, medium-speed storage layers support periodic data updates, and low-speed storage layers focus on data archiving and long-term preservation. Next, data of each category is allocated according to the defined rules. For instance, monitoring data is mapped to a high-speed storage layer supporting high-frequency reads, treatment plans are mapped to a medium-speed storage layer supporting periodic updates, and nursing records are mapped to a low-speed storage layer to reduce resource consumption. After the mapping relationship is established, the storage allocation is verified and adjusted. For example, simulated nursing scenarios are used to test whether the real-time access speed of the high-speed storage layer meets the requirements, and the accuracy of update operations in the medium-speed storage layer is checked. Any abnormal allocations or inefficient storage problems are corrected. Then, the resource utilization of the overall storage system is analyzed. For example, the capacity utilization and read / write efficiency of each storage layer are evaluated. Based on the analysis results, the storage layer capacity or data allocation rules are dynamically adjusted to optimize the resource configuration of the storage layers. Finally, accurate storage layer partitioning results are output, providing an efficient storage solution for the long-term management of oncology nursing data.

[0107] Please see Figure 7 The specific steps for identifying patient response patterns during care are as follows:

[0108] Based on the data storage hierarchy division results, real-time tumor nursing data, including physiological indicators, monitoring data and nursing records, are extracted. By real-time monitoring and analysis of the dynamic changes of these data, the real-time tumor nursing data change analysis results are obtained.

[0109] First, the system extracts the patient's physiological indicators, monitoring data, and nursing intervention records from the high-speed storage layer in real time. For example, the heart rate data extracted from Patient A's high-speed storage layer is: 80 bpm (0 minutes), 85 bpm (5 minutes), and 90 bpm (10 minutes). The nursing record shows that 5 nursing interventions were completed within 1 hour. Subsequently, during the monitoring process, the system automatically identifies the dynamic changes in these data and marks abnormalities. For example, if the heart rate rises from 85 bpm to 90 bpm at a rate of 1 bpm / min, it is within the normal range. However, if the heart rate fluctuation rate increases to 3 bpm / min, the system marks it as a high-risk state and records the time of the abnormal change. In addition, the system monitors the frequency of nursing interventions. For example, if Patient B's nursing frequency increases from 2 times / hour to 5 times / hour, it indicates an increase in the density of nursing operations. During the monitoring process, the system compares the changing data in real time, analyzes trends based on historical data, and generates a monitoring report. At the same time, it records relevant logs for subsequent analysis of the source and dynamic change patterns of abnormal data, providing support for subsequent nursing intervention optimization and data adjustment.

[0110] Based on the analysis results of real-time tumor nursing data changes, the following formula was used:

[0111]

[0112] The magnitude of change (B) of the patient's physiological indicators and the rate of change (F) of the frequency of nursing interventions were calculated respectively, and the analysis results of the patient's response pattern were generated.

[0113] Where T is the total number of data points within the time period, representing the number of samples for monitoring data, and V... t These are physiological indicator values ​​at time point t, recorded in real time by physiological monitoring equipment, such as monitoring changes in heart rate or blood pressure. It is the average physiological indicator during the monitoring period, calculated using the following formula: ΔN is the change in the number of nursing interventions within a time period. It is obtained by extracting the number of interventions for each time period from the nursing log and calculating the change in adjacent time periods. ΔT represents the time difference between two adjacent records.

[0114] For calculating the range of change B of a patient's physiological indicators, for example, patient A's heart rate data is collected by monitoring equipment and is 80, 85, 90, 88, and 92 bpm within 10 minutes.

[0115] Calculate the mean:

[0116] Calculate the sum of squared deviations:

[0117] Fluctuation range:

[0118] The results showed that the range of change B in the patient's physiological indicators was 4.19 bpm.

[0119] For the rate of change F of nursing intervention frequency, for example, if the nursing log shows that patient B completed 5 nursing interventions in the first hour and 3 nursing interventions in the second hour.

[0120] The change in the number of nursing care sessions ΔN = 3 - 5 = -2 times;

[0121] Time period length ΔT = 1 hour;

[0122] Changes in nursing frequency times / hour.

[0123] The results showed that the nursing frequency decreased by 2 times per hour.

[0124] Please see Figure 8 The specific steps for obtaining the results of tumor nursing data processing are as follows:

[0125] Based on the analysis results of patient response patterns, patient data that has been divided into multiple storage levels is extracted and re-analyzed. In response to the latest changes in heart rate fluctuation amplitude and nursing frequency, the patient data is dynamically adjusted and the storage levels are re-divided to obtain updated storage level data.

[0126] By continuously analyzing patient data and dynamically updating data storage layers, the system first re-extracts and re-analyzes patient data already assigned to different storage layers. For example, patient A's heart rate fluctuation range is 4.19 bpm, initially classified into the medium-speed storage layer. However, during continuous monitoring, it was found that the heart rate fluctuation range increased from 4.19 bpm to 6 bpm, and the nursing frequency decreased at a rate of -2 times / hour. Based on the latest changes, the system adjusts patient A's data to the high-speed storage layer for further real-time analysis. Similarly, for patient B, initially assigned to the high-speed storage layer due to a large heart rate fluctuation range (6 bpm) and a high rate of increase in nursing frequency (+3 times / hour), continuous monitoring showed that the heart rate fluctuation range decreased to 3 bpm, and the nursing frequency stabilized. The system then adjusts patient B's data to the medium-speed storage layer for periodic data. The entire dynamic update process involves data monitoring, classification rule comparison, storage layer transfer execution, and updating logs for adjusted data to ensure that dynamic data changes reflect the latest status and meet the needs of hierarchical storage.

[0127] Based on the updated storage hierarchy data, real-time patient data from multiple priority categories are integrated to analyze individual and group nursing status and generate tumor nursing data processing results.

[0128] The system integrates high-priority patient data that is adjusted in real time with periodically updated medium- and low-priority data. For example, for patient B, whose heart rate fluctuations are continuously increasing and whose nursing frequency is rising, personalized nursing strategy optimization suggestions are generated. At the same time, combined with patient A's stable nursing data, a comparative analysis report on the group's nursing status is generated. During the data integration process, the system dynamically analyzes the impact of new data on the overall trend, and uses changes in nursing records and physiological indicator fluctuations to display the latest nursing status of each patient. Finally, it outputs nursing data processing results covering multi-level patient groups, providing support for real-time decision-making and strategy adjustment.

[0129] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A cloud-based tumor nursing data processing system, characterized in that, The system includes: The multi-scale feature analysis module acquires tumor nursing data, decomposes the tumor nursing data into multiple time dimensions, extracts multi-scale features of the tumor nursing data, ranks the importance of the multi-scale features of the tumor nursing data, and obtains preliminary tumor nursing data feature alignment results. The dynamic feature alignment adjustment module, based on the preliminary tumor nursing data feature alignment results, filters matching feature combinations according to a preset correlation threshold, dynamically updates the matching relationship between features, and obtains cross-patient feature fusion results. The specific steps for obtaining the feature combinations for filtering and matching are as follows: Based on the preliminary tumor nursing data feature alignment results, features such as trend change rate, nursing intervention frequency change rate, and treatment response duration across patient data are extracted according to importance ranking. Missing feature values ​​are imputed and data alignment is performed to generate a feature integration set across patients. Based on the aforementioned cross-patient feature integration set, the following formula is used: ； Calculate feature alignment features and Pearson correlation coefficient between This yields the matching feature combinations; in It is a feature and In the The values ​​in each sample It is a feature and The mean, It is the total number of patients; The specific steps for obtaining the cross-patient feature fusion results are as follows: Based on the highly relevant feature pairs selected from the matched feature combinations, feature data from multiple periods are extracted, and the features are classified and weighted according to time characteristics. The feature matching relationship is dynamically adjusted to generate dynamically updated cross-time scale feature matching results. Based on the dynamically updated cross-timescale feature matching results, and combined with the cloud computing environment, the feature data is distributed to multiple nodes for parallel processing, local optimization and global integration are performed, and feature matching verification and result optimization are conducted to obtain cross-patient feature fusion results. Based on the cross-patient feature fusion results, the non-uniform gridded storage module analyzes the multi-scale features of the fused tumor nursing data, maps inconsistent data to matching storage layers, and obtains the data storage hierarchy division results. The adaptive data partitioning and optimization module monitors real-time changes in tumor nursing data based on the data storage hierarchy partitioning results, identifies the patient's response patterns during the nursing process, dynamically updates the data storage hierarchy, and generates tumor nursing data processing results. The specific steps for identifying the patient's response patterns during the nursing process are as follows: Based on the data storage hierarchy division results, real-time tumor nursing data, including physiological indicators, monitoring data and nursing records, are extracted. By real-time monitoring and analysis of the dynamic changes in the data, the real-time tumor nursing data change analysis results are obtained. Based on the analysis results of the real-time tumor care data changes, the following formula is used: ； Calculate the range of change in the patient's physiological indicators respectively and the rate of change in the frequency of nursing interventions The analysis results of generating patient response patterns are generated. in, It is the total number of data points within the time period. It is the first Physiological index values ​​at each time point It is the average value of physiological indicators during the monitoring period. It is the change in the number of nursing interventions within a time period. It represents the time difference between two adjacent records.

2. The cloud-based tumor nursing data processing system according to claim 1, characterized in that, The specific steps for obtaining the multi-scale features of the tumor nursing data are as follows: Based on patient monitoring data and nursing records, the data is broken down according to the time dimension, and the range of changes in monitoring data, the frequency of nursing interventions and the duration of treatment response are extracted to generate corresponding feature data tables. Based on the corresponding feature data table, the range of each feature value is organized, and the features are normalized to adjust the magnitude of the feature values ​​and obtain a standardized feature dataset.

3. The cloud-based tumor nursing data processing system according to claim 2, characterized in that, The specific steps for obtaining the preliminary tumor nursing data feature alignment results are as follows: For the standardized feature dataset, the formula is as follows: ； Calculate a single feature Importance score The results of the feature importance analysis were obtained. in, These are the normalized features. The value, It is a correlation factor, representing a feature. The degree of relevance to nursing goals, It is a feature The sampling interval time; Based on the feature importance analysis results, the features are sorted from high to low scores, and the order of the corresponding features in the data table is adjusted simultaneously to obtain preliminary feature alignment results for tumor nursing data.

4. The cloud computing-based tumor nursing data processing system according to claim 1, characterized in that, The specific steps for obtaining the data storage hierarchy division results are as follows: Based on the cross-patient feature fusion results, physiological indicators, treatment plans, monitoring data and nursing records of tumor nursing data are classified and extracted. The data are mapped to inconsistent storage layers according to data characteristics and storage requirements, and the mapping is verified to obtain matching storage layer mapping information. Based on the matched storage layer mapping information, a corresponding storage mapping relationship is constructed, the storage rules and resource allocation of the storage layer are optimized, abnormal allocations are adjusted, and data storage hierarchy division results are generated.

5. The cloud-based tumor nursing data processing system according to claim 1, characterized in that, The specific steps for obtaining the tumor nursing data processing results are as follows: Based on the analysis results of the patient response patterns, the patient data that has been divided into multiple storage levels is extracted and reanalyzed. In view of the latest situation of changes in heart rate fluctuation amplitude and nursing frequency, the patient data is dynamically adjusted and the storage levels are re-divided to obtain updated storage level data. Based on the updated storage hierarchy data, real-time patient data from multiple priority categories are integrated to analyze individual and group nursing status and generate tumor nursing data processing results.

Images