Lung cancer caregiver demand data mining and resource recommendation method

Abstract

The invention provides a demand data mining and resource recommendation method for a lung cancer caregiver, and relates to the technical field of resource recommendation, and the method comprises the steps: receiving the target disease course stage information of a target cared person and the target care information of a target caregiver; performing demand mining and resource recommendation on the target disease course stage information and the target nursing information by a resource recommendation device, and outputting a target recommendation resource; the resource recommendation device comprises a demand mining channel and a resource recommendation channel; and anonymously updating the resource recommender according to the target recommendation resource. According to the invention, the technical problem that the improvement of the nursing quality is restricted due to the deviation between recommended resources and actual demands in the prior art can be solved, and the technical goal of constructing an intelligent nursing support system based on multi-dimensional demand perception and dynamic resource mapping is achieved. The technical effects that the recommendation accuracy is improved, the self-adaptive capability of the system is continuously enhanced, and the user satisfaction is stably increased are achieved.

Description

Technical Field

[0001] This application relates to the field of resource recommendation technology, and in particular to a method for data mining and resource recommendation of lung cancer caregivers' needs. Background Technology

[0002] With increasing societal emphasis on lung cancer patient care systems, AI-based care support technologies have seen rapid development in recent years.

[0003] Currently, existing care resource recommendation systems mainly employ two technical approaches: one is a static recommendation model based on keyword matching, which distributes resources by establishing a correspondence between disease stages and nursing knowledge; the other is a personalized recommendation algorithm based on collaborative filtering, which achieves resource matching by analyzing user group behavior patterns. While these existing technologies have addressed the information asymmetry problem to some extent, several structural defects have gradually emerged in practical applications. First, at the level of demand perception, existing technologies often only capture explicit demands expressed by users, failing to effectively identify potential demands implied through behavioral data. This single-channel demand identification mechanism leads to serious biases in judging the user's true state. Second, at the level of resource adaptation, the mapping relationships established by most systems lack dynamic evolution capabilities. They cannot adapt to the non-linear characteristics of patient disease progression, nor can they respond to the dynamic changes in caregiver needs over time. This rigid matching model demonstrates significant inadequacy in the face of complex and ever-changing clinical scenarios.

[0004] In summary, existing technologies suffer from inherent defects such as a single dimension of demand perception and a rigid resource mapping mechanism, which leads to a significant deviation between recommended resources and actual needs. This further affects the actual effectiveness and large-scale application of care support systems, ultimately hindering the improvement of care quality. Summary of the Invention

[0005] The purpose of this application is to provide a method for data mining and resource recommendation of lung cancer caregivers' needs, in order to solve the technical problem that the existing technology has inherent defects such as a single dimension of demand perception and a rigid resource mapping mechanism, which leads to a significant deviation between recommended resources and actual needs, further affecting the actual effectiveness and large-scale promotion and application of care support systems, and ultimately restricting the improvement of care quality.

[0006] In view of the above problems, this application provides a method for data mining and resource recommendation of lung cancer caregivers' needs, comprising: receiving target disease stage information of target caregivers and target care information of target caregivers; using a resource recommender to perform needs mining and resource recommendation on the target disease stage information and target care information, and outputting target recommended resources; the resource recommender includes a needs mining channel and a resource recommendation channel; and updating the resource recommender anonymized according to the target recommended resources.

[0007] Preferably, the method for data mining and resource recommendation of lung cancer caregiver needs further includes: introducing a lung cancer database, collecting disease progression records of lung cancer patients and corresponding care records of lung cancer caregivers; mapping the disease progression records and care records to obtain a disease progression-needs mapping framework, and embedding it into the needs mining channel; matching care needs according to a resource engine, obtaining a needs-resource mapping framework based on the resource matching results, and embedding it into the resource recommendation channel; connecting the needs mining channel and the resource recommendation channel to generate the resource recommender.

[0008] Preferably, the method for data mining and resource recommendation of lung cancer caregiver needs further includes: extracting disease stage records and treatment stage records based on the disease stage records; using disease stage and treatment stage as matrix indicators; filling the matrix indicator values ​​with the disease stage records and treatment stage records to obtain a disease stage matrix; wherein the disease stage includes at least early, middle, and late stages, and the treatment stage includes at least diagnosis, treatment, and outcome; extracting active care records and passive care records from the care records; performing care need analysis on the active care records and passive care records to generate a care need vector; concatenating the care need vector according to the disease stage matrix to generate a full-cycle care need matrix; and combining the disease stage matrix and the full-cycle care need matrix to obtain the disease-need mapping framework.

[0009] Preferably, the method for data mining and resource recommendation of lung cancer caregiver needs further includes: using the active care records as the table demand script, performing demand script transformation based on information needs, psychological needs, and support needs to obtain the table demand script; using the passive care records as the latent demand script, performing demand script transformation based on information needs, psychological needs, and support needs to obtain the latent demand script; performing structured feature transformation on the table demand script and the latent demand script respectively to obtain table demand features and latent demand features; and concatenating the table demand features and latent demand features to generate the care demand vector.

[0010] Preferably, the method for data mining and resource recommendation of lung cancer caregivers' needs further includes: obtaining care gap records based on the disease stage records; parsing care needs based on the care gap records to generate demand gaps; triggering resource creation tasks through the demand gaps; reviewing the safety and effectiveness of the created resources to obtain qualified resources; assigning resource profile tags to the qualified resources and adding them to the resource engine.

[0011] Preferably, the method for data mining and resource recommendation of lung cancer caregivers' needs further includes: parsing care needs based on the care records to generate a care need vector; assigning a need profile label to the care need vector; matching the need profile label with the resource profile label in the resource engine to output a resource matching result; and mapping the care need vector and the resource matching result to obtain the need-resource mapping framework.

[0012] Preferably, the method for data mining and resource recommendation of lung cancer caregivers' needs further includes: matching the demand profile tags with the resource profile tags to obtain a preliminary matching result set; sorting the preliminary matching result set based on preference, quality, and popularity to obtain a preferred resource sequence, a quality resource sequence, and a popular resource sequence; selecting resources from the preferred resource sequence, the quality resource sequence, and the popular resource sequence based on the preferred resource weight, the quality resource weight, and the popular resource weight, and outputting the resource matching result.

[0013] Preferably, the method for data mining and resource recommendation of lung cancer caregivers' needs further includes: retrieving a target matching result set based on the resource recommender; if the active selection of the target caregiver is empty, automatically selecting and generating the target recommended resource; if the active selection of the target caregiver is not empty, receiving the active selection of the target caregiver as the target recommended resource; updating the preferred resource sequence according to the target recommended resource, and updating the popular resource sequence based on the updated preferred resource sequence.

[0014] Preferably, the method for data mining and resource recommendation of lung cancer caregivers' needs further includes: receiving a target disease inflection point and triggering a high-risk support mode; simplifying the interactive interface of the resource recommender through the high-risk support mode, increasing the query frequency of the target disease stage information and target care information, and outputting inflection point recommended resources.

[0015] Preferably, the method for data mining and resource recommendation of lung cancer caregivers' needs further includes: collecting reference suggestions from the resource engine based on the target recommended resource; if the reference suggestion is answered with "agree to withdraw", deleting the target recommended resource, the target disease stage information, and the target care information; if the reference suggestion is answered with "agree to reference", using the anonymized target recommended resource, target disease stage information, and target care information as the update input for the resource recommender.

[0016] The technical solution provided in this application has at least the following technical effects or advantages: by achieving the technical goal of constructing an intelligent care support system with multi-dimensional demand perception and dynamic resource mapping, it achieves the technical effects of improved recommendation accuracy, continuously enhanced system adaptability, and stable growth in user satisfaction.

[0017] The above description is merely an overview of the technical solution of this application. To enable a clearer understanding of the technical means of this application and to facilitate its implementation according to the description, and to make the above and other objects, features, and advantages of this application more apparent, specific embodiments of this application are described below. It should be understood that the content described in this section is not intended to identify key or important features of the embodiments of this application, nor is it intended to limit the scope of this application. Other features of this application will become readily apparent through the following description. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely exemplary. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0019] Figure 1 This is a flowchart illustrating a method for data mining and resource recommendation of lung cancer caregivers' needs, as proposed in this application.

[0020] Figure 2 This is a schematic diagram illustrating the process of constructing a resource recommender in the "Data Mining and Resource Recommendation Method for Lung Cancer Caregivers" of this application. Detailed Implementation

[0021] This application provides a method for data mining and resource recommendation of lung cancer caregivers' needs. It addresses the inherent flaws in existing technologies, such as a single dimension of need perception and a rigid resource mapping mechanism, which leads to a significant discrepancy between recommended resources and actual needs. This discrepancy further affects the actual effectiveness and large-scale application of care support systems, ultimately hindering the improvement of care quality. The method aims to construct an intelligent care support system with multi-dimensional need perception and dynamic resource mapping, achieving improved recommendation accuracy, continuously enhanced system adaptability, and stable growth in user satisfaction.

[0022] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. It should be understood that this application is not limited to the exemplary embodiments described herein. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application. It should also be noted that, for ease of description, only the parts related to this application are shown in the accompanying drawings, not all of them.

[0023] Please see Figure 1 and Figure 2 This application provides a method for data mining and resource recommendation of lung cancer caregivers' needs, which specifically includes the following steps:

[0024] S1: Receive information on the target disease stage of the target care recipient and the target care information of the target caregiver.

[0025] Specifically, the system receives information on the target disease stage of the target recipient and the target care information of the target caregiver. The target recipient specifically refers to an individual patient with lung cancer who is receiving care. The target disease stage information is a classification description of the patient's lung cancer progression, divided according to medical standards into stages such as early, intermediate, and late stages, and can be further refined in conjunction with specific diagnostic, treatment, and prognosis stages. The target caregiver refers to the individual or group providing daily care and support to the patient. The target care information encompasses various data generated during the care process, including both proactive statements from caregivers, such as questions raised in questionnaires or requests for help posted on forums, and unconscious behavioral traces, such as browsing history or search keywords in resource libraries.

[0026] S2: The resource recommender performs demand mining and resource recommendation on the target disease stage information and target care information, and outputs target recommended resources; the resource recommender includes a demand mining channel and a resource recommendation channel.

[0027] Specifically, the resource recommender comprehensively analyzes and processes the input target disease stage information and target care information. The resource recommender is a specially designed computational module containing two collaborative functional units: a needs mining channel and a resource recommendation channel. The needs mining channel performs needs mining operations, that is, using an algorithmic model to identify and extract the core needs and their intensity from the raw input information. The resource recommendation channel is responsible for resource recommendation, retrieving and filtering the most relevant supporting materials from a pre-built resource library based on the identified needs characteristics. After collaborative processing by the two channels, the final output is the target recommended resources, which refer to the set of optimal support solutions customized for the current specific care scenario. The needs mining channel receives the target disease stage information, obtaining the disease development stage of the target caregiver and its corresponding treatment steps. It also receives the target care information, including the directly expressed needs and indirectly expressed behavioral data of the target caregiver. By analyzing the inherent relationships within the information, a structured needs description can be output. Subsequently, the resource recommendation channel, based on the needs description, initiates a matching calculation process, selecting the most suitable solution from various forms of supporting materials to form the final target recommended resources.

[0028] S3: Update the resource recommender by anonymizing the recommended resources based on the target.

[0029] Specifically, after the resource recommendation process is completed, the resource recommender's update mechanism will be activated. This update mechanism uses the target recommended resource as its core input data. During the update process, the target recommended resource undergoes anonymization, removing all personally identifiable identifiers and sensitive attributes, ensuring that the processed data cannot be linked to any specific individual. This anonymized data is then used to optimize the resource recommender's internal parameters and algorithm model.

[0030] Furthermore, this application also includes: introducing a lung cancer database, collecting disease progression records of lung cancer patients and corresponding care records of lung cancer caregivers; mapping the disease progression records and care records to obtain a disease progression-demand mapping framework, and embedding it into the demand mining channel; matching care demands according to a resource engine, obtaining a demand-resource mapping framework based on the resource matching results, and embedding it into the resource recommendation channel; connecting the demand mining channel and the resource recommendation channel to generate the resource recommender.

[0031] Specifically, a lung cancer database is introduced. This database is a structured collection of data covering medical knowledge, clinical guidelines, and real-world cases related to lung cancer. Two types of data are collected from the database: first, disease progression records of lung cancer patients, which objectively describe the progression of the disease and are divided according to tumor staging standards, such as recording the disease as early, middle, or late stage, and can further record the specific stage such as diagnosis, active treatment, or follow-up rehabilitation; second, care records of corresponding lung cancer caregivers, which include all information generated by caregivers in supporting patients, including questions and feedback they proactively raise, as well as behavioral data generated when using relevant applications or platforms.

[0032] Subsequently, a mapping process is executed, which involves performing a correlation analysis on the collected disease progression records and care records. Mapping refers to a computational method for establishing the correspondence between two different types of data elements. Through mapping operations, a disease progression-demand mapping framework can be constructed, describing the correspondence between different disease progression stages and various needs that caregivers may have. This disease progression-demand mapping framework is embedded into the demand mining channel of the resource recommender, serving as the core knowledge foundation for demand analysis and guiding the prediction and mining of care needs based on the input disease progression information.

[0033] Then, a component of a resource engine is invoked to match the identified care needs. The resource engine is a software module that manages and schedules support resources. Internally, it stores various standardized and tagged resources, such as popular science articles, nursing videos, and access points to psychological counseling services. The matching process refers to calculating the similarity between the characteristics of the need and the attributes of the resources according to a preset algorithm. Based on the results of resource matching, a need-resource mapping framework is constructed. This framework defines the conversion rules between specific types of care needs and the most suitable support resources, and is embedded in the resource recommendation channel of the resource recommender, enabling the channel to quickly transform abstract needs into a concrete list of resources.

[0034] Finally, by connecting the demand mining channel and the resource recommendation channel, a complete resource recommender was generated. Connectivity means establishing a data interface and calling protocol between two previously independent software components, enabling the demand mining channel and the resource recommendation channel to work collaboratively. The demand mining channel is responsible for understanding the context and identifying the problem, while the resource recommendation channel is responsible for providing solutions. Together, they form an automated processing pipeline from raw information input to final resource output—the resource recommender.

[0035] Furthermore, this application also includes: extracting disease stage records and treatment stage records based on the disease stage records; using disease stage and treatment stage as matrix indicators; filling the matrix indicator values ​​through the disease stage records and treatment stage records to obtain a disease stage matrix; wherein the disease stage includes at least early, middle, and late stages, and the treatment stage includes at least diagnosis, treatment, and outcome; extracting active care records and passive care records from the care records; performing care demand analysis on the active care records and passive care records to generate a care demand vector; concatenating the care demand vector according to the disease stage matrix to generate a full-cycle care demand matrix; and combining the disease stage matrix and the full-cycle care demand matrix to obtain the disease-demand mapping framework.

[0036] Specifically, in constructing the disease progression-demand mapping framework, more granular data is extracted based on disease progression stage records. Specifically, this involves extracting disease staging records and treatment stage records. Disease staging records are the result of classifying the progression of lung cancer according to medical standards, including at least three categories: early, intermediate, and advanced. Treatment stage records describe the core aspects of medical intervention, including at least three stages: diagnosis, treatment, and outcome. Subsequently, using disease staging and treatment stage as two-dimensional matrix indicators, a structured disease progression stage matrix is ​​obtained by filling specific disease staging and treatment stage records into the positions defined by these matrix indicators. The disease progression stage matrix can systematically represent the specific disease state of a patient.

[0037] Two different types of data were separated from the care records. The first is active care records, which refer to feedback and statements provided willingly and proactively by caregivers, such as written responses in questionnaires or proactively posted requests for help. The second is passive care records, which refer to behavioral data unconsciously generated by caregivers during the receipt of care services, such as their browsing history and dwell time on information platforms. Next, care needs analysis was performed on both types of records using natural language processing or behavioral pattern recognition technology. This process identified and extracted the core needs of caregivers from the raw data. The goal of the analysis was to transform the unstructured records into a structured care needs vector, a mathematically ordered array where the value in each dimension represents the intensity or probability of a caregiver's need for a specific type of need.

[0038] Furthermore, the disease progression matrix and care demand vectors are integrated. Specifically, based on the specific disease stage identified in the disease progression matrix, one or more corresponding care demand vectors are compared and concatenated. Comparison refers to establishing a logical relationship between the disease progression stage and the demand vectors, while concatenation refers to mathematically combining multiple vectors into a more complex structure. Through these comparison and concatenation operations, a full-cycle care demand matrix is ​​generated, showcasing the overall distribution and changing patterns of caregiver demand vectors across different disease progression stages.

[0039] Finally, by combining the disease progression matrix and the full-cycle care needs matrix—that is, by fusing and associating them—a disease progression-needs mapping framework is obtained, thus forming a complete knowledge model. This framework clearly reveals the mapping relationship between specific disease progression stages and a series of potential care needs, laying a theoretical foundation for subsequent needs prediction.

[0040] Furthermore, this application also includes: using the active care record as the explicit demand script, performing a demand script transformation based on information needs, psychological needs, and support needs to obtain an explicit demand script; using the passive care record as the implicit demand script, performing a demand script transformation based on information needs, psychological needs, and support needs to obtain an implicit demand script; performing structured feature transformation on the explicit demand script and the implicit demand script respectively to obtain explicit demand features and implicit demand features; and concatenating the explicit demand features and the implicit demand features to generate the care demand vector.

[0041] Specifically, proactive care records are defined as explicit needs statements for processing. Explicit needs statements refer to the caregiver's explicit needs expressed directly through language or writing, such as clearly stated questions or difficulties during communication. The process of transforming these explicit needs statements into a needs script is based on a pre-defined needs classification system, which includes at least three core dimensions: informational needs, psychological needs, and supportive needs. Informational needs refer to the need for factual information such as medical knowledge and nursing skills; psychological needs refer to the need for emotional support such as emotional comfort and stress relief; and supportive needs refer to the need for operational support such as practical tools and human assistance. By analyzing the correlation between the explicit needs statements and these three dimensions, they are transformed into a structured needs script, which is a standardized representation of the user's stated needs.

[0042] Next, passive care records are defined as latent need scripts and processed. Latent need scripts refer to unspoken, underlying needs indirectly reflected through caregiver behavioral data, such as behavioral patterns of repeatedly browsing specific web pages or using specific functions. These latent need scripts are then transformed into demand scripts based on the same dimensions. Behavioral pattern analysis is used to infer the potential information, psychological, or support needs behind them, resulting in a latent need script—an inferential representation of the user's underlying intentions.

[0043] Then, structured feature transformation is performed on both the explicit demand script and the implicit demand script. Structured feature transformation refers to converting the semantic or behavioral information contained in the script into feature vectors that can be directly processed by a computer through feature engineering methods. The explicit demand script is transformed into explicit demand features, including quantitative indicators such as word frequency and sentiment polarity; the implicit demand script is transformed into implicit demand features, including statistical quantities such as behavior frequency and sequence patterns.

[0044] Finally, the explicit and implicit needs features are combined into a comprehensive care needs vector through a concatenation operation. Concatenation refers to the mathematical connection of two or more feature vectors in terms of dimensions to form a higher-dimensional feature vector containing more comprehensive information. The resulting care needs vector is the final output, which can simultaneously reflect the quantitative representation of caregivers' explicit expressions and implicit behaviors.

[0045] Furthermore, this application also includes: obtaining care blank records based on the disease stage records; performing care need analysis based on the care blank records to generate need blanks; triggering a resource creation task through the need blanks, reviewing the safety and effectiveness of the created resources to obtain qualified resources; assigning resource profile tags to the qualified resources and adding them to the resource engine.

[0046] Specifically, care gap records are obtained based on disease progression records. Care gap records refer to care problems or support gaps that are not adequately covered by the existing resource system at a specific disease progression stage, as identified through analysis of historical data. Care gap records are identified using statistical methods; for example, a care gap record is formed when caregiver needs frequently arise at a certain disease progression stage but corresponding resource support is lacking.

[0047] Care needs analysis is performed based on care blank records. Data analysis techniques are used to extract specific unmet needs from the blank records to generate clear needs gaps. These gaps are structured and defined care needs that have not yet been met in a specific clinical setting. Their specific forms may include the management of a new type of side effect or the psychological support needs at a specific stage of treatment.

[0048] Subsequently, resource creation tasks are automatically triggered based on the identified demand gaps. A resource creation task is an instruction issued to the resource production end to develop corresponding supporting resources for a specific demand gap. The initial resources created under this task undergo rigorous security and validity reviews. Security reviews ensure that the resource content complies with medical standards and is not misleading, while validity reviews verify that the resources can indeed address the corresponding care needs. Resources that pass the reviews are deemed qualified and qualify for inclusion in the resource system.

[0049] Finally, qualified resources undergo standardization by assigning them resource profile tags to complete their feature descriptions. Resource profile tags are a set of feature identifiers that comprehensively describe resource attributes, including metadata such as resource type, applicable disease course, category of need addressed, and difficulty level. Taggled resources are then formally added to the resource engine, becoming effective components that can be used by the recommendation system.

[0050] Furthermore, this application also includes: parsing care needs based on the care records to generate a care need vector; assigning a need profile label to the care need vector; matching the need profile label with the resource profile label in the resource engine and outputting a resource matching result; and mapping the care need vector and the resource matching result to obtain the need-resource mapping framework.

[0051] Specifically, care needs analysis is performed based on care records. Care needs analysis refers to the process of analyzing and processing raw care records using a computational model to identify the underlying needs information. This process transforms unstructured records into structured care need vectors, which are mathematically represented arrays of features where the values ​​of each dimension quantify the intensity or probability of different types of needs. Based on these care need vectors, corresponding need profile labels are assigned to them, providing semantic identifiers that describe the characteristics of the needs, such as categorizing needs as emergency information queries or long-term psychological support.

[0052] The resource engine matches demand profile tags with predefined resource profile tags. Resource profile tags are a predefined set of feature identifiers used to describe resource attributes. Matching refers to the operation of establishing a correspondence between the demand tag set and the resource tag set using a similarity calculation algorithm. The matching operation outputs a sorted list of candidate resources.

[0053] Furthermore, a mapping operation is performed to handle the relationship between care demand vectors and resource matching results, through a computational method that establishes a correspondence from a mathematical vector space to a resource entity space. By associating the demand patterns expressed by the care demand vectors with the solutions contained in the resource matching results, a demand-resource mapping framework is ultimately obtained. The demand-resource mapping framework is a set of rules or a mathematical model that guides the system on how to transform specific types and intensities of demand into specific resource recommendation strategies.

[0054] Furthermore, this application also includes: matching the demand profile tags with the resource profile tags to obtain a preliminary matching result set; sorting the preliminary matching result set based on preference, quality, and popularity to obtain a preferred resource sequence, a quality resource sequence, and a popular resource sequence; selecting resources from the preferred resource sequence, the quality resource sequence, and the popular resource sequence based on the preferred resource weight, the quality resource weight, and the popular resource weight, and outputting a resource matching result.

[0055] Specifically, similarity is calculated based on the correspondence between resource profile tags and demand profile tags. A preliminary matching result set is obtained through a semantic matching algorithm between tags, containing a list of all candidate resources potentially relevant to the current demand.

[0056] Next, a multi-dimensional resource ranking operation is performed on the preliminary matching result set. Resource ranking refers to the calculation process of prioritizing candidate resources according to specific indicators. Specifically, ranking is performed based on three independent dimensions: first, a preferred resource sequence is generated based on users' historical behavior data, reflecting the degree to which resources match users' personal habits; second, a quality resource sequence is generated based on objective indicators such as the authority and timeliness of resources; and third, a popular resource sequence is generated by referring to statistical indicators such as the recent access volume and user interaction rate of resources. Each sequence is an ordered arrangement of the same batch of resources after evaluation from different perspectives.

[0057] Then, a weighted selection mechanism is used to determine the final output from the preferred resource sequence, the quality resource sequence, and the popularity resource sequence. The weighted selection mechanism assigns specific contribution coefficients to each sequence, namely, the preferred resource weight, the quality resource weight, and the popularity resource weight, which determine the degree of influence of each sequence in the final decision. Based on the preset weight configuration, the resource positions in the three sequences are comprehensively evaluated, and a resource matching result with optimized ranking is finally output.

[0058] Furthermore, this application also includes: retrieving the target matching result set based on the resource recommender; if the active selection of the target caregiver is empty, automatically selecting and generating the target recommended resource; if the active selection of the target caregiver is not empty, receiving the active selection of the target caregiver as the target recommended resource; updating the preferred resource sequence according to the target recommended resource, and updating the popularity resource sequence based on the updated preferred resource sequence.

[0059] Specifically, the target matching result set is retrieved based on the processing results of the resource recommender. The target matching result set is a list of candidate resources formed after multiple rounds of filtering and sorting, which includes several resource options with the highest matching degree to the current care scenario.

[0060] If the target caregiver's active selection is null, meaning the user has not provided explicit feedback on the recommendation, an automatic selection mechanism will be activated. This mechanism will select the highest-ranking resource from the target matching result set according to preset optimization rules, automatically generating the target recommended resource. Conversely, if the target caregiver's active selection is not null, meaning the user has explicitly specified a resource option, this active selection will be directly accepted as the final target recommended resource. This ensures respect for the user's explicit preferences.

[0061] After determining the target recommended resources, a self-updating mechanism is initiated. The preferred resource sequence is updated based on the finally adopted target recommended resources. This preferred resource sequence is an ordered set recording users' historical selection tendencies. The update process is achieved by strengthening the weight of the selected resources in the sequence. Subsequently, based on the updated preferred resource sequence, the popularity resource sequence is updated synchronously. The popularity resource sequence is a dynamic ranking reflecting the overall popularity of resources, and its update logic incorporates the access count and selection frequency of the selected resources into the popularity calculation model.

[0062] Furthermore, this application also includes: receiving the target disease inflection point and triggering a high-risk support mode; simplifying the interactive interface of the resource recommender through the high-risk support mode, increasing the query frequency of the target disease stage information and target care information, and outputting inflection point recommended resources.

[0063] Specifically, when a target disease inflection point is received, it means that a critical turning point in the target patient's condition has been identified, such as transitioning from a stable treatment phase to an acute exacerbation phase or from late-stage treatment to end-of-life care. The identification process triggers a pre-set high-risk support mode, a special operating state designed for periods of significant changes in the patient's condition, aiming to provide more focused and urgent care support.

[0064] Upon entering high-risk support mode, several adaptive adjustments are immediately implemented. First, the interface of the resource recommender is simplified by reducing the display of non-core functional modules and highlighting key information and operation buttons to reduce the cognitive load on target caregivers in emergency situations. Simultaneously, the frequency of inquiries regarding target disease stages and target care information is increased, meaning more frequent and proactive collection of details about disease progression and current care challenges from caregivers. Based on this enhanced information, a specially optimized algorithm outputs inflection point recommended resources. These inflection point recommended resources are emergency response plans screened for disease inflection points, including highly relevant support materials such as symptom crisis management guidelines and emergency contact information sets.

[0065] Furthermore, this application also includes: collecting citation suggestions from the resource engine based on the target recommended resource; if the citation suggestion is answered with agreement to withdraw, deleting the target recommended resource, the target disease stage information, and the target care information; if the citation suggestion is answered with agreement to cite, using the anonymized target recommended resource, target disease stage information, and target care information as the update input for the resource recommender.

[0066] Specifically, a citation suggestion collection process will be initiated with the resource engine based on the identified target recommended resources. Citation suggestions refer to resource usage authorization requests that are automatically generated and submitted to relevant authorized parties, with the purpose of confirming whether the recommended cases can be used for system optimization.

[0067] Based on the response to the citation suggestion received, different data processing paths will be executed. When the response to the citation suggestion received is an agreement to withdraw, the data deletion procedure will be initiated immediately, including the complete removal of the target recommendation resources generated in this interaction, the associated target disease stage information, and the corresponding target care information, ensuring that no records are left in the system.

[0068] Conversely, if the received citation suggestion is answered with agreement to citation, the data anonymization process will be initiated. Anonymization refers to the process of removing all personally identifiable identifiers, making the data unlinkable to a specific individual. The anonymized target recommended resources, target disease stage information, and target care information will be used as training data input into the resource recommender to update and optimize its recommendation algorithm.

[0069] In summary, the lung cancer caregiver needs data mining and resource recommendation method provided in this application has the following technical effects: by achieving the technical goal of constructing an intelligent care support system with multi-dimensional needs perception and dynamic resource mapping, it achieves the technical effects of improved recommendation accuracy, continuously enhanced system adaptability, and stable growth in user satisfaction.

[0070] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0071] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of this application and its equivalents, this application also intends to include such modifications and variations.

Claims

1. A method for data mining and resource recommendation of lung cancer caregivers' needs, characterized in that, include: Receive information on the target disease stage of the target care recipient and the target care information of the target caregiver; A resource recommender is used to perform demand mining and resource recommendation on the target disease stage information and target care information, and outputs target recommended resources; the resource recommender includes a demand mining channel and a resource recommendation channel; The resource recommender is updated with anonymized recommendations based on the target.

2. The method for data mining and resource recommendation of lung cancer caregivers' needs as described in claim 1, characterized in that, Building a resource recommender includes: A lung cancer database was introduced to collect disease progression records of lung cancer patients and corresponding care records of lung cancer caregivers. The disease progression records and care records are mapped to obtain a disease progression-demand mapping framework, which is then embedded into the demand mining channel. Based on the resource engine matching care needs, a demand-resource mapping framework is obtained based on the resource matching results and embedded into the resource recommendation channel; Connect the demand mining channel and the resource recommendation channel to generate the resource recommender.

3. The method for data mining and resource recommendation of lung cancer caregivers' needs as described in claim 2, characterized in that, Mapping the disease progression records and care records yields a disease progression-demand mapping framework, including: Based on the disease stage records, disease progression records and treatment stage records are extracted. Using disease progression and treatment stage as matrix indicators, the matrix indicator values ​​are filled by the disease progression records and treatment stage records to obtain a disease stage matrix. The disease progression stages include at least early, middle and late stages, and the treatment stages include at least diagnosis, treatment and prognosis. Extract active care records and passive care records from the care records, analyze the care needs of the active care records and passive care records, and generate a care need vector; Based on the disease stage matrix, the care demand vector is spliced ​​together to generate a full-cycle care demand matrix; By combining the disease stage matrix and the full-cycle care needs matrix, the disease-needs mapping framework is obtained.

4. The method for data mining and resource recommendation of lung cancer caregivers' needs as described in claim 3, characterized in that, The active care records and passive care records are analyzed for care needs to generate care need vectors, including: Using the aforementioned proactive care records as the script for the table of needs, the script for the table of needs is transformed based on information needs, psychological needs, and assistance needs to obtain the table of needs script. Using the passive care records as the dialogue of latent needs, the script of latent needs is transformed based on the information needs, psychological needs, and assistance needs to obtain the script of latent needs. The table requirement script and the latent requirement script are respectively subjected to structured feature transformation to obtain table requirement features and latent requirement features; The care demand vector is generated by concatenating the table demand features and the latent demand features.

5. The method for data mining and resource recommendation of lung cancer caregivers' needs as described in claim 2, characterized in that, Building a resource engine includes: Obtain blank care records based on the disease progression records; Based on the care blank records, care needs are analyzed to generate a need blank; The resource creation task is triggered by the aforementioned demand blank, and the security and validity of the created resources are reviewed to obtain qualified resources; The qualified resources are assigned resource profile tags and added to the resource engine.

6. The method for data mining and resource recommendation of lung cancer caregivers' needs as described in claim 2, characterized in that, Based on the resource engine matching care needs, a demand-resource mapping framework is derived from the resource matching results, including: Care needs are analyzed based on the care records to generate a care needs vector, and a needs profile label is assigned based on the care needs vector. Based on the resource profile tags in the resource engine, match the demand profile tags and output the resource matching results; The demand-resource mapping framework is obtained by mapping the care demand vector and the resource matching result.

7. The method for data mining and resource recommendation of lung cancer caregivers' needs as described in claim 6, characterized in that, Based on the resource profile tags in the resource engine, the resource matching results are output according to the demand profile tags, including: A preliminary matching result set is obtained by matching the resource profile tags with the demand profile tags; The preliminary matching result set is sorted based on preference, quality, and popularity to obtain a preferred resource sequence, a quality resource sequence, and a popularity resource sequence. Based on the preferred resource weight, quality resource weight, and popularity resource weight, resources are selected from the preferred resource sequence, quality resource sequence, and popularity resource sequence, and the resource matching result is output.

8. The method for data mining and resource recommendation of lung cancer caregivers' needs as described in claim 7, characterized in that, Output target recommended resources, including: The target matching result set is retrieved based on the resource recommender; If the target caregiver's active selection is empty, the target recommended resource will be automatically generated. If the target caregiver's active selection is not empty, the target caregiver's active selection is received and used as the target recommended resource; The preferred resource sequence is updated based on the target recommended resources, and the popularity resource sequence is updated based on the updated preferred resource sequence.

9. The method for data mining and resource recommendation of lung cancer caregivers' needs as described in claim 1, characterized in that, After outputting the target recommended resources, the following is included: Receive the target disease inflection point and trigger the high-risk support mode; The high-risk support mode simplifies the interface of the resource recommender, increases the frequency of queries for target disease stage information and target care information, and outputs inflection point recommended resources.

10. The method for data mining and resource recommendation of lung cancer caregivers' needs as described in claim 1, characterized in that, After outputting the target recommended resources, it also includes: Based on the aforementioned target, the resource recommendation engine will collect reference suggestions; If the response to the suggested advice is "agree to withdraw", delete the target recommended resource, the target disease stage information, and the target care information; If the response to the cited suggestion is "agree to cite", the anonymized target recommended resources, target disease stage information, and target care information will be used as the update inputs for the resource recommender.

Images