Diagnostic prediction device and system based on knowledge caching and personalized federated learning

Abstract

The present invention provides a diagnostic prediction device and system based on knowledge caching and personalized federated learning, which relates to the field of model learning and medical imaging diagnosis. The present invention constructs a dynamic knowledge cache by receiving medical sample index information uploaded by multiple clients; creates an index relationship mapping table from label to index and index to hash value based on the medical sample index information; initializes each medical sample and establishes an index relationship mapping between each medical sample and the R most similar samples; receives the category probability vector and its index uploaded by each client, updates the knowledge cache; then updates the index relationship mapping table and retrieves the knowledge vectors of the R most similar samples; uses a weighted average method to generate a comprehensive knowledge representation; distributes the comprehensive knowledge representation to each client for personalized federated learning; and uses the trained client models for medical diagnosis prediction. The present invention solves the problems of high communication costs, poor model structure compatibility, and difficulty in obtaining public data sets.

Description

Technical Field

[0001] The present invention relates to the technical field of model learning and medical imaging diagnosis, and in particular to a diagnosis prediction device and system based on knowledge caching and personalized federated learning. Background Art

[0002] In recent years, with the widespread application of medical imaging technologies (such as X-rays, CT scans, and MRIs), the potential of artificial intelligence in medical imaging diagnosis has become increasingly apparent. However, the sensitive nature of medical data and the stringent requirements of relevant privacy regulations have limited the use of traditional centralized machine learning methods. Federated learning, by collaboratively training models across distributed devices and avoiding the sharing of raw data, has become an effective means of protecting data privacy.

[0003] Existing technologies for the application of personalized federated learning in medical imaging diagnosis primarily employ architectures based on parameter interaction. The approach is as follows: each client (e.g., a hospital imaging workstation) trains local medical imaging data using a unified model structure, generating large-scale model parameters that are then uploaded to a central server. The server aggregates these parameters through weighted averaging, updates the global model, and distributes the updated parameters. To mitigate the lack of personalization, some methods incorporate additional public datasets (e.g., public imaging libraries) to assist in training, or fine-tune the global model on the client side to adapt it to local needs. However, architectures based on parameter interaction require frequent transmission of large-scale model parameters, resulting in high communication costs between distributed medical devices. Furthermore, the model structure suffers from poor compatibility, typically requiring a unified model structure that is difficult to adapt to the diverse hardware performance and diagnostic needs of different medical institutions. Furthermore, some methods utilize additional public datasets, but obtaining such data in the medical field is difficult due to privacy restrictions.

[0004] In view of this, the applicant filed this application after studying the existing technology. Summary of the Invention

[0005] This invention aims to provide a diagnostic prediction device and system based on knowledge caching and personalized federated learning to address the high communication costs, poor model structure compatibility, and difficulty accessing public datasets that plague traditional federated learning architectures based on parameter interaction. By introducing dynamic knowledge caching and a multimodal feature fusion mechanism, combined with indexed relationship mapping and a weighted knowledge fusion strategy, this invention aims to achieve more efficient distributed medical imaging diagnostic model training while meeting the strict privacy protection requirements of the medical field.

[0006] In order to solve the above technical problems, the present invention is implemented through the following technical solutions:

[0007] A diagnostic prediction device based on knowledge caching and personalized federated learning, applied to the edge server of a distributed medical network, includes a dynamic knowledge cache construction unit, an index relationship mapping unit, a most similar sample mapping unit, a knowledge cache update unit, a knowledge vector retrieval unit, a comprehensive knowledge representation generation unit, and a personalized learning unit;

[0008] The dynamic knowledge cache construction unit is used to receive medical sample index information uploaded by multiple clients and construct a dynamic knowledge cache; wherein the medical sample index information includes a hash value, index and label of the medical sample;

[0009] The index relationship mapping unit is used to create an index relationship mapping table of label to index and index to hash value according to the medical sample index information, so as to associate each label with a corresponding index and each index with a corresponding hash value;

[0010] The most similar sample mapping unit is used to initialize each medical sample and establish an index relationship mapping between each medical sample and R most similar samples;

[0011] The knowledge cache updating unit is used to receive the category probability vector and its index uploaded by each client, so as to update the latest inference prediction result of the client to the knowledge cache;

[0012] The knowledge vector retrieval unit is used to update the index relationship mapping table according to the updated knowledge cache, and retrieve the knowledge vectors of the R most similar samples for each medical sample;

[0013] The comprehensive knowledge representation generating unit is used to fuse the knowledge vectors to generate comprehensive knowledge representation by adopting a weighted average method;

[0014] The personalized learning unit is used to distribute the comprehensive knowledge representation to each client; wherein, the comprehensive knowledge representation is used to trigger the client model of each client to perform personalized federated learning training of the distributed medical network based on the comprehensive knowledge representation, and perform medical diagnosis prediction based on the trained client model to output a prediction label.

[0015] Preferably, the medical sample index information is obtained by each client performing feature fusion and privacy encoding processing on the locally stored medical image samples, specifically:

[0016] Each client reads a medical sample dataset in a local database; wherein the medical sample dataset includes medical image samples, patient metadata, and diagnostic labels, which are expressed as:

[0017] ;

[0018] in, represents a medical sample dataset; k∈{1,2,…,K} represents the kth client, and K represents the total number of clients in the distributed medical network; is the i-th image sample; for corresponding patient metadata; for corresponding diagnostic labels; is the sample size;

[0019] Each client uses its own pre-trained client model to process medical image samples and patient metadata to extract corresponding image features and text features. The client model includes a convolutional neural network and a Deepseek model. The convolutional neural network processes the input medical image samples and extracts the spatial features of the image through convolution and pooling operations to capture the texture and structural information in the image. The image features are expressed as follows:

[0020] ;

[0021] in, represents the image features of the i-th image sample of the k-th client; 、 are the pre-trained weight and bias parameters respectively;

[0022] The Deepseek model processes the input patient metadata and extracts text semantic features through matrix operations to reflect the individual characteristics of the patient. The text features are expressed as follows:

[0023] ;

[0024] in, Represents patient metadata Text features of 、 are the pre-trained weight and bias parameters respectively;

[0025] The image features and the text features are combined to obtain a comprehensive feature vector, which is expressed as:

[0026] ;

[0027] in, is the comprehensive feature vector corresponding to the i-th image sample; Represents a splicing operation;

[0028] The differential privacy hash algorithm is used to encode and protect the comprehensive feature vector. That is, the comprehensive feature vector is mapped to a low-dimensional space through a random projection matrix, and then Gaussian noise perturbation is added to generate a hash value of fixed length. The expression is:

[0029] ;

[0030] ;

[0031] in, is the hash value corresponding to the i-th image sample of the k-th client; DPH represents the differential privacy hash algorithm; H is the random projection matrix; Indicates that the mean square is 0 and the variance is Gaussian noise; and is the differential privacy parameter; ΔF is the sensitivity of the feature, indicating the maximum range of variation of the feature; sign represents the encoding operation;

[0032] Each client uploads the corresponding hash value, diagnostic label, and index obtained from the comprehensive feature vector to the edge server through an encrypted channel.

[0033] Preferably, when initializing each medical sample, each medical sample is initialized to a knowledge vector of an all-zero vector, representing its initial state before training; the dimension of the knowledge vector is consistent with the number of categories of the diagnostic label.

[0034] Preferably, when establishing an index relationship mapping between each medical sample and the R most similar samples,

[0035] The cosine similarity between the hash values of all medical samples is calculated, and a time decay factor is introduced to reduce the weight of the medical sample data uploaded earlier. The R most similar samples are screened out, and the improved HNSW algorithm is used to accelerate the retrieval process. The expression is:

[0036] ;

[0037] in, represents the R most similar samples of the i-th image sample of the k-th client; Represents two hash values 、 The cosine similarity of k, Represents the index of the client; i, j represent the index of the image sample; Indicates the time decay rate, which is used to control the influence of historical data; t indicates the current timestamp; Indicates the The upload timestamp of the jth image sample of a client; Top-R means selecting the top R samples with the highest similarity.

[0038] Preferably, the category probability vector is generated by each client using a client model to infer and predict the input medical sample and the predicted category probability corresponding to the medical sample generated by forward propagation; the number of elements in the category probability vector is consistent with the number of categories.

[0039] Preferably, the expression of the comprehensive knowledge representation is:

[0040] ;

[0041] ;

[0042] in, representing the comprehensive knowledge representation; The knowledge vector representing the i-th image sample of the k-th client, that is, the knowledge vector corresponding to the index (k, i); Represents the knowledge vector of the s-th similar sample; Represents the weight of the corresponding knowledge vector; R represents the number of similar samples.

[0043] Preferably, the weight of the comprehensive knowledge representation The cosine similarity of the hash values corresponding to the image samples and the similarity of the patient metadata are normalized and expressed as:

[0044] ;

[0045] ;

[0046] in, represents the hash value of the i-th image sample of the k-th client; 、 Respectively represent the sth, The hash value of similar samples; cos represents cosine similarity; represents the patient metadata of the i-th image sample of the k-th client; 、 Respectively represent the sth, Patient metadata for similar samples; Indicates similarity; represents the L2 norm;

[0047] 、 Indicates the patient age of the corresponding sample, through patient metadata get; 、 The vector representing the severity of the patient's condition for the corresponding sample is obtained by combining the patient metadata with the corresponding diagnostic label and scoring by experts.

[0048] Preferably, when each client performs personalized federated learning training for a distributed medical network, a comprehensive objective function is used to optimize the model; wherein the comprehensive objective function includes a supervision loss function and a distillation loss function, and is calculated by dynamic weighting;

[0049] The supervised loss function is calculated using cross entropy loss, which adjusts the model parameters by comparing the difference between the model predicted category probability and the true label to ensure that the model prediction is consistent with the local diagnostic label;

[0050] The distillation loss function uses KL divergence to measure the distribution difference between the local category probability vector and the comprehensive knowledge representation, so that the local reasoning result is consistent with the knowledge cache; the expressions are:

[0051] ;

[0052] ;

[0053] ;

[0054] in, represents a medical sample dataset; Represents the client model comprehensive objective function The trainable parameters of is the i-th image sample; is the true value of the label; Represents the true value of the c-th class label; represents the supervised loss function; express The predicted probability of represents the softmax normalization function; represents the dynamic weight of the comprehensive objective function; KL represents the distillation loss function; represents the predicted probability on the knowledge cache; C represents the total number of label categories; represents the predicted probability of the c-th class label; represents the predicted probability of the c-th class label on the knowledge cache; represents the probability prediction function.

[0055] Preferably, the dynamic weight of the comprehensive objective function Adjusted according to the heterogeneity of local medical sample data and the severity of the patient's condition, the formula is:

[0056] ;

[0057] in, Represents a local medical sample dataset Global data distribution on knowledge cache KL divergence of; Represents patient metadata The severity score of the disease is obtained by combining the patient metadata with the corresponding labels and then scoring by experts; 、 are the weights corresponding to the KL divergence and the disease severity score;

[0058] Each client calculates the gradient of the comprehensive objective function and updates the trainable parameters of the model according to the learning rate , the expression is:

[0059] ;

[0060] in, is the learning rate; ) is the comprehensive objective function right gradient.

[0061] The present invention also provides a diagnosis and prediction system based on knowledge caching and personalized federated learning, comprising an edge server and a plurality of clients; each client is connected to the edge server via communication;

[0062] The client is used to read the medical sample dataset in the local database and process the medical sample dataset using a pre-trained client model to extract corresponding features for splicing and fusion to obtain a comprehensive feature vector; the comprehensive feature vector is then encoded and protected using a differential privacy hashing algorithm to obtain medical sample index information, which is uploaded to the edge server; the medical sample index information includes the hash value, index and label of the medical sample;

[0063] The edge server is configured to construct a dynamic knowledge cache based on the medical sample index information received from multiple clients; and to create an index relationship mapping table from label to index and from index to hash value based on the medical sample index information, so as to associate each label with a corresponding index and each index with a corresponding hash value; and then, initialize each medical sample and establish an index relationship mapping between each medical sample and the R most similar samples;

[0064] The client is used to use the client model to infer the input medical sample data and generate a category probability vector through forward propagation, and upload it to the edge server together with the corresponding index;

[0065] The edge server is used to receive category probability vectors and their indexes uploaded by multiple clients to update the client's latest inference prediction results to the knowledge cache; update the index relationship mapping table based on the updated knowledge cache, and retrieve the knowledge vectors of the R most similar samples for each medical sample; then use a weighted average method to fuse the knowledge vectors to generate a comprehensive knowledge representation, and distribute the comprehensive knowledge representation to each client;

[0066] The client is used to receive the comprehensive knowledge representation distributed by the edge server and perform personalized federated learning and optimization training of the distributed medical network; the trained model is used for remote medical imaging diagnosis and prediction.

[0067] In summary, compared with the prior art, the present invention has the following beneficial effects:

[0068] The edge server of the present invention fuses the knowledge of the R most similar samples based on the similarity of hash values and metadata (such as age and disease severity) to form personalized guidance for each client. It is particularly suitable for disease screening and remote diagnosis in distributed medical networks, providing an efficient and secure solution for smart medical care.

[0069] This invention reduces the communication overhead between distributed medical devices, supports heterogeneous model architecture, adapts to the computing capabilities of different devices, realizes personalized medical imaging diagnosis without the need for public data sets, and improves the adaptability and prediction accuracy of client models for different patient groups.

[0070] This invention uses the differential privacy hashing algorithm (DPH) to encode the comprehensive features of medical images and patient metadata into fixed-length hash values and adds Gaussian noise perturbation to effectively prevent the leakage of original data. Compared with the traditional method of directly transmitting complete model parameters, it greatly improves privacy security.

[0071] The client of the present invention only uploads category probability vectors of the same size as the number of categories (e.g., only two values for a two-category problem), and performs global fusion based on the knowledge cache and time decay factor of the edge server. The communication cost is reduced to less than one thousandth of that of traditional federated learning. It also supports heterogeneous model deployment and adapts to different hardware requirements. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments. It should be understood that the following drawings only illustrate certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained based on these drawings without paying any creative work.

[0073] Figure 1A schematic diagram of the structure of a diagnostic prediction device based on knowledge caching and personalized federated learning provided in Example 1.

[0074] Figure 2 This is a flowchart of a diagnostic prediction device based on knowledge caching and personalized federated learning provided in Example 1.

[0075] Figure 3 A schematic diagram of a diagnostic prediction system based on knowledge caching and personalized federated learning provided in Example 2.

[0076] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments. DETAILED DESCRIPTION

[0077] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the invention for which protection is sought, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0078] Example 1

[0079] Embodiment 1 of the present invention provides a diagnostic prediction device based on knowledge caching and personalized federated learning, which can be implemented by a diagnostic prediction device based on knowledge caching and personalized federated learning (hereinafter referred to as a diagnostic prediction device), and in particular, executed by one or more processors in the diagnostic prediction device.

[0080] In this embodiment, the diagnostic prediction device may be an electronic device equipped with a processor, which has a computer program of the diagnostic prediction device based on knowledge caching and personalized federated learning and can be executed, such as a computer, a smart phone, a smart tablet, a workstation, etc., which is not limited here.

[0081] In this embodiment, knowledge caching is a key feature of the distributed medical network system. It allows edge servers to store and update medical image features and knowledge learned from clients (such as portable ultrasound machines and hospital imaging workstations). This mechanism enables the system to quickly access and infer new medical image data, improving processing speed and accuracy. Furthermore, knowledge caching helps reduce computing pressure on edge servers, improving overall system performance and efficiency.

[0082] In this embodiment, personalized federated learning is another core feature of the distributed medical network system. It allows clients to perform a certain degree of model training and learning locally. This allows the system to fully leverage each client's specific data and characteristics, improving model accuracy and adaptability. Federated learning also ensures data privacy and security, as raw data never leaves the client; only processed features or model parameters are transmitted to edge servers for further analysis and training.

[0083] like Figure 1-Figure 2 As shown, a diagnostic prediction device based on knowledge caching and personalized federated learning is applied to the edge server of a distributed medical network, including a dynamic knowledge cache construction unit, an index relationship mapping unit, a most similar sample mapping unit, a knowledge cache update unit, a knowledge vector retrieval unit, a comprehensive knowledge representation generation unit and a personalized learning unit.

[0084] Dynamic knowledge cache construction unit: used to receive medical sample index information uploaded by multiple clients and construct a dynamic knowledge cache; wherein the medical sample index information includes the hash value, index and label of the medical sample.

[0085] In this embodiment, the overall architecture of the distributed medical network system includes multiple clients and an edge server, each connected to the edge server via communication. These components utilize a distributed computing framework and privacy-preserving mechanisms to enable efficient and personalized model training, addressing the high communication costs, insufficient model heterogeneity, and limited personalization capabilities of traditional approaches.

[0086] When the system starts, each client extracts features from the locally stored medical image data and performs privacy protection processing, laying the foundation for subsequent knowledge sharing. In this embodiment, the medical sample index information is obtained by each client performing feature fusion and privacy encoding processing on the locally stored medical image samples, specifically:

[0087] Each client reads a medical sample dataset from a local database. The medical sample dataset includes medical image samples (e.g., CT scan images), patient metadata (e.g., age, gender, medical history), and diagnostic labels (e.g., "pulmonary nodule positive" or "normal"), expressed as:

[0088] ;

[0089] in, represents a medical sample dataset; k∈{1,2,…,K} represents the kth client, and K represents the total number of clients in the distributed medical network; is the i-th image sample; for corresponding patient metadata; for corresponding diagnostic labels; is the sample size;

[0090] Each client uses its own pre-trained client model to process medical image samples and patient metadata to extract corresponding image features and text features. The client model includes a convolutional neural network and a Deepseek model. For image samples, the convolutional neural network processes the input medical image samples, extracting the spatial features of the image through convolution and pooling operations to capture the texture and structure information in the image, and obtaining image features, which are expressed as:

[0091] ;

[0092] in, represents the image features of the i-th image sample of the k-th client; 、 are the pre-trained weight and bias parameters respectively;

[0093] For metadata, the patient metadata is processed by the Deepseek model (such as Deepseek R1), and the text semantic features are extracted through matrix operations to reflect the individual characteristics of the patient. The text features are obtained, which are expressed as follows:

[0094] ;

[0095] in, Represents patient metadata Text features of 、 are the pre-trained weight and bias parameters respectively;

[0096] After the extraction is completed, the image features and the text features are spliced and fused to obtain a comprehensive feature vector so as to uniformly process the two types of information. The expression is:

[0097] ;

[0098] in, is the comprehensive feature vector corresponding to the i-th image sample; Represents a splicing operation.

[0099] Due to the sensitivity of medical data, directly uploading the original features may lead to privacy leakage. Therefore, the client uses a differential privacy hash algorithm to encode and protect the comprehensive feature vector. That is, the comprehensive feature vector is mapped to a low-dimensional space through a random projection matrix, and then Gaussian noise perturbation is added to generate a fixed-length hash value. The expression is:

[0100] ;

[0101] ;

[0102] in, is the hash value corresponding to the i-th image sample of the k-th client; DPH represents the differential privacy hashing algorithm; H is a random projection matrix (e.g., it can be generated by the Johnson-Lindenstrauss theorem); Indicates that the mean square is 0 and the variance is Gaussian noise; and is the differential privacy parameter; ΔF is the sensitivity of the feature, which indicates the maximum range of variation of the feature (for example, it can be taken as the maximum L2 norm difference of the feature vectors in the training set); sign represents the encoding operation.

[0103] In this embodiment, the Differential Privacy Hashing (DPH) algorithm is an innovative approach that combines differential privacy protection technology with a hash algorithm, aiming to achieve data de-identification while strictly protecting privacy. DPH works by adding differentially private random noise (such as Laplace or Gaussian noise) to the traditional hashing process. This makes the output impossible to reverse engineer back to the original data, ensuring that the hash output distributions of adjacent data sets (which differ by only one record) are nearly identical.

[0104] This method ensures data privacy while preserving the similarity relationship between samples. After encoding, the client will hash the value , sample index (k,i) and diagnostic label Upload to the edge server through an encrypted channel (such as TLS protocol) to provide secure data for subsequent cache construction.

[0105] Index relationship mapping unit: used to create an index relationship mapping table from label to index and from index to hash value according to the medical sample index information, so as to associate each label with the corresponding index and each index with the corresponding hash value.

[0106] In this embodiment, after receiving the hash value, index, and tag uploaded by the client, the edge server builds a dynamic knowledge cache (KC) to efficiently organize distributed data and support knowledge sharing in subsequent training.

[0107] First, the server creates a label-to-index mapping table, associating each diagnostic label with the corresponding sample index. For example, the "positive lung nodule" label is mapped to the index of all related samples to facilitate fast retrieval of samples by category. The update method is as follows:

[0108] ;

[0109] in, Diagnostic label The corresponding index set; It is a set merge operation; is the sample index, indicating the i-th image sample of the k-th client.

[0110] Next, the hash value of each sample is stored in the index map, and a lookup table from the sample index to the hash value is established to provide a data basis for subsequent similarity calculations. The storage method is as follows:

[0111] ;

[0112] in, Sample index The corresponding hash value is , A set of hash values.

[0113] The most similar sample mapping unit is used to initialize each medical sample and establish an index relationship mapping between each medical sample and the R most similar samples.

[0114] In this unit, the server initializes each medical sample as a knowledge vector of all zero vectors, representing its initial state before training. The dimension of the knowledge vector is consistent with the number of categories of the diagnostic label. The initialization process is as follows:

[0115] ;

[0116] in, Index of samples The corresponding knowledge vector, C is the total number of diagnostic categories, such as "positive" and "negative" corresponding to C = 2, Represents a real number.

[0117] Next, an index relationship mapping between each medical sample and the R most similar samples is established to support knowledge fusion.

[0118] Calculate the cosine similarity between the hash values of all medical samples, and introduce a time decay factor to reduce the weight of medical sample data uploaded earlier (for example, the weight of samples uploaded more than 1 month is halved). Then select the R most similar samples and use the improved HNSW algorithm to accelerate the retrieval process. The expression is:

[0119] ;

[0120] in, represents the R most similar samples of the i-th image sample of the k-th client; Represents two hash values 、 The cosine similarity of k, Represents the index of the client; i, j represent the index of the image sample; Indicates the time decay rate, which is used to control the influence of historical data; t indicates the current timestamp; Indicates the The upload timestamp of the jth image sample of the client; Indicates selecting the first R samples with the highest similarity.

[0121] In this embodiment, the improved HNSW (Hierarchical Navigable Small World) algorithm may adopt a graph traversal strategy improvement method to perform directional heuristic pruning; or may adopt an index structure enhancement method of dynamic hierarchical weight adjustment and incremental quantization coding.

[0122] Through the above index mapping, knowledge cache provides efficient data support for distributed training.

[0123] Knowledge cache update unit: used to receive the category probability vector and its index uploaded by each client to update the client's latest inference prediction results to the knowledge cache.

[0124] This step is the model training phase for each client, which extracts local knowledge. At the beginning of each round of model training, each client uses its own model to infer image samples and extract knowledge representations to participate in global knowledge fusion. Clients select the appropriate model architecture based on their own hardware performance, and these models do not need to be consistent with other clients.

[0125] The client model uses inference prediction to predict the input medical sample and generates the predicted class probability corresponding to the medical sample through forward propagation, namely the class probability vector logits. The number of elements in the class probability vector is consistent with the number of classes. For example, for a two-class problem, logits has two values. The data volume of the class probability vector is much smaller than the complete model parameters. The client uploads it to the server along with the index (k, i), significantly reducing communication overhead. The class probability vector is represented as:

[0126] ;

[0127] in, is the logits vector of the i-th sample of the k-th client; Trainable parameters representing the comprehensive objective function of the client model; is the i-th image sample; Represents the probability prediction function of the client model.

[0128] After receiving the logits uploaded by each client, the server first updates the knowledge cache and records the latest inference results in the cache. The update method is as follows:

[0129] .

[0130] Knowledge vector retrieval unit: used to update the index relationship mapping table according to the updated knowledge cache, and retrieve the knowledge vectors of the R most similar samples for each medical sample.

[0131] In this unit, the knowledge set of the R most similar samples for each sample is retrieved according to the index relationship mapping. For example, the logits of the R=10 samples most similar to the current sample are extracted from the cache. The results are as follows:

[0132] ;

[0133] in, Represents the knowledge vector of the Rth similar sample.

[0134] Comprehensive knowledge representation generation unit: used to generate comprehensive knowledge representation by fusing the knowledge vectors using a weighted average method.

[0135] Furthermore, the expression of the comprehensive knowledge representation is:

[0136] ;

[0137] , that is, the weights must meet the normalization conditions.

[0138] in, representing the comprehensive knowledge representation; The knowledge vector representing the i-th image sample of the k-th client, that is, the knowledge vector corresponding to the index (k, i); Represents the knowledge vector of the s-th similar sample; Represents the weight of the corresponding knowledge vector; R represents the number of similar samples.

[0139] The weight of the comprehensive knowledge representation The cosine similarity of the hash values corresponding to the image samples and the patient metadata similarity are normalized to reflect the semantic relevance between samples. Patient metadata similarity is calculated based on patient age differences and disease severity. For example, samples with an age difference of less than 5 years are more similar, and samples with similar disease scores are given greater weight.

[0140] The expression is:

[0141] ;

[0142] ;

[0143] in, represents the hash value of the i-th image sample of the k-th client; 、 Respectively represent the sth, The hash value of similar samples; cos represents cosine similarity; represents the patient metadata of the i-th image sample of the k-th client; 、 Respectively represent the sth, Patient metadata for similar samples; Indicates similarity; represents the L2 norm;

[0144] 、 Indicates the patient age of the corresponding sample, through patient metadata get; 、 The vector representing the severity of the patient's condition for the corresponding sample is obtained by combining the patient metadata with the corresponding diagnostic label and scoring by experts.

[0145] When experts grade, different diseases have corresponding diagnostic labels, and the expert team will assign severity scores to these labels. For example:

[0146] Chronic diseases (such as hypertension): may be scored on a graded basis (grade 1 / grade 2 / grade 3).

[0147] Acute diseases (such as myocardial infarction): combined with Killip classification or TIMI scoring system.

[0148] Complex diseases: Quantify the risk of multiple organ failure using scoring tools such as APACHE II and SOFA.

[0149] Clinical experts (such as attending physicians and above) can also be invited to perform double-blind scoring of the cases. Each expert independently quantifies the [symptom duration] and [number of organs affected] indicators in the metadata using a 0-10 score system based on the "ICD-11 Clinical Severity Assessment Guidelines", and finally takes a weighted average value (weight: for example, 0.6 for chief physicians / 0.4 for deputy chief physicians).

[0150] Personalized learning unit: used to distribute the comprehensive knowledge representation to each client; wherein, the comprehensive knowledge representation is used to trigger the client model of each client to perform personalized federated learning training of the distributed medical network based on the comprehensive knowledge representation, and perform medical diagnosis prediction based on the trained client model and output prediction labels.

[0151] Comprehensive knowledge representation after fusion It is distributed to the corresponding client as guidance information for personalized training.

[0152] The client receives the comprehensive knowledge representation After that, the model is optimized based on local data. When training the client model, a comprehensive objective function is used for optimization. The comprehensive objective function includes the supervision loss function and the distillation loss function, which are weighted by dynamic weights.

[0153] The supervised loss function is calculated using cross entropy loss, which adjusts the model parameters by comparing the difference between the model predicted category probability and the true label to ensure that the model prediction is consistent with the local diagnostic label;

[0154] The distillation loss function uses KL divergence to measure the distribution difference between the local category probability vector and the comprehensive knowledge representation, so that the local reasoning result is consistent with the knowledge cache; the expressions are:

[0155] ;

[0156] ;

[0157] ;

[0158] in, represents a medical sample dataset; Represents the client model comprehensive objective function The trainable parameters of is the i-th image sample; is the true value of the label; Represents the true value of the c-th class label; represents the supervised loss function; express The predicted probability of represents the softmax normalization function; represents the dynamic weight of the comprehensive objective function; KL represents the distillation loss function; represents the predicted probability on the knowledge cache; C represents the total number of label categories; represents the predicted probability of the c-th class label; represents the predicted probability of the c-th class label on the knowledge cache; represents the probability prediction function.

[0159] To achieve personalization, dynamic weights are adjusted according to the heterogeneity of local data and the severity of the patient's condition. For example, samples with large data distribution differences or severe conditions are given higher weights. That is, the dynamic weights of the comprehensive objective function are Adjusted according to the heterogeneity of local medical sample data and the severity of the patient's condition, the formula is:

[0160] ;

[0161] in, Represents a local medical sample dataset Global data distribution on knowledge cache KL divergence of; Represents patient metadata The severity score of the disease is obtained by combining the patient metadata with the corresponding labels and then scoring by experts; 、 are the weights corresponding to the KL divergence and the disease severity score;

[0162] Each client updates the model parameters using the gradient descent method by calculating the gradient of the comprehensive objective function and updating the trainable parameters of the model according to the learning rate. , the expression is:

[0163] ;

[0164] in, is the learning rate; ) is the comprehensive objective function right gradient.

[0165] Through these steps, the trained client models are used for remote medical diagnosis, prediction, and disease screening, improving the model's adaptability and prediction accuracy for different patient populations. For example, on client 1, the patient's medical imaging data (such as CT images) is input into a model trained using knowledge caching and personalized federated learning for diagnosis and prediction, resulting in a prediction result, such as "positive for lung nodules" or "normal." This prediction incorporates updated knowledge uploaded by each client, improving its accuracy.

[0166] This technical solution achieves efficient distributed training while protecting privacy, and improves the personalized performance of the model through knowledge fusion.

[0167] In summary, compared with the prior art, the present invention has the following beneficial effects:

[0168] In practical applications, the device can be deployed in a distributed hospital network and is suitable for disease screening and remote diagnosis. For example, in a pulmonary nodule screening task, portable ultrasound machines or imaging workstations in different hospitals serve as client devices, entering patient metadata and diagnostic labels via text boxes. Following the aforementioned steps, personalized federated learning is completed. The edge server achieves efficient knowledge sharing and integration through a dynamic knowledge cache construction unit and knowledge cache update unit, significantly reducing communication costs and improving model personalization performance.

[0169] In the dynamic knowledge cache construction unit, the combined features of medical images and patient metadata are encoded into fixed-length hash values using the differentially private hashing algorithm (DPH). Gaussian noise perturbation is added to prevent leakage of the original data. The client only uploads logits vectors that are the same as the number of categories. Combined with the edge server's knowledge cache and time decay factor, global fusion is performed, reducing communication costs to less than one thousandth of traditional federated learning. It also supports heterogeneous model architectures and adapts to different hardware requirements. The server combines the knowledge of the R most similar samples based on the similarity of hash values and metadata to form personalized guidance, providing an efficient and secure solution for smart healthcare.

[0170] The present invention adopts an incremental knowledge update mechanism. When each medical sample first enters the knowledge cache, it initializes its knowledge vector as an all-zero vector. Subsequently, it receives the latest reasoning results uploaded by the client, updates the knowledge vector of the corresponding medical sample, and timestamps the time point of each update to ensure that the data in the knowledge cache always reflects the latest reasoning results.

[0171] The present invention adopts an adaptive similarity adjustment system. It generates an initial similarity matrix by calculating the cosine similarity between the hash values of medical samples, generates an adjusted similarity matrix by introducing a time decay factor, and selects the top R samples with the highest similarity from the adjusted similarity matrix. The decay rate of the time decay factor is dynamically adjusted according to the difference between the upload timestamp and the current timestamp to ensure that the influence of historical data gradually weakens over time.

[0172] This method uses a multidimensional weight distribution mechanism to generate similarity weights by calculating the cosine similarity between image sample hash values, and metadata weights by calculating the similarity between patient age and disease severity vectors. The final weight is obtained by normalizing the similarity weights and metadata weights, combining image features and individual patient characteristics to improve the accuracy of knowledge vector fusion. The comprehensive knowledge representation is distributed to all clients for global model updates. Based on the client's hardware performance and diagnostic requirements, the comprehensive knowledge representation is tailored or enhanced and distributed to specific clients for personalized model optimization. A differentiated distribution strategy is used to meet the hardware performance and diagnostic requirements of different medical institutions.

[0173] The present invention adopts a dynamic gradient adjustment mechanism to generate a gradient update direction by calculating the gradient of the model's trainable parameters with a comprehensive objective function. The learning rate is dynamically adjusted to generate a gradient update step size based on the heterogeneity of local data and the patient's disease severity score. The efficiency and effectiveness of model optimization are improved by combining local data characteristics with individual patient characteristics.

[0174] Example 2

[0175] like Figure 3 As shown, the second embodiment of the present invention further provides a diagnosis and prediction system based on knowledge caching and personalized federated learning, including an edge server and multiple clients; each client is connected to the edge server through communication; wherein, the client is used to read the medical sample data set in the local database, and use the pre-trained client model to process the medical sample data set to extract corresponding features for splicing and fusion to obtain a comprehensive feature vector; then, a differential privacy hash algorithm is used to encode and protect the comprehensive feature vector to obtain medical sample index information, which is uploaded to the edge server; wherein, the medical sample index information includes the hash value, index and label of the medical sample;

[0176] The edge server is configured to construct a dynamic knowledge cache based on the medical sample index information received from multiple clients; and to create an index relationship mapping table from label to index and from index to hash value based on the medical sample index information, so as to associate each label with a corresponding index and each index with a corresponding hash value; and then, initialize each medical sample and establish an index relationship mapping between each medical sample and the R most similar samples;

[0177] The client is used to use the client model to infer the input medical sample data and generate a category probability vector through forward propagation, and upload it to the edge server together with the corresponding index;

[0178] The edge server is used to receive category probability vectors and their indexes uploaded by multiple clients to update the client's latest inference prediction results to the knowledge cache; update the index relationship mapping table based on the updated knowledge cache, and retrieve the knowledge vectors of the R most similar samples for each medical sample; then use a weighted average method to fuse the knowledge vectors to generate a comprehensive knowledge representation, and distribute the comprehensive knowledge representation to each client;

[0179] The client is used to receive the comprehensive knowledge representation distributed by the edge server and perform personalized federated learning and optimization training of the distributed medical network; the trained model is used for remote medical imaging diagnosis and prediction.

[0180] In the several embodiments provided in the embodiments of the present invention, it should be understood that the disclosed devices and methods can also be implemented in other ways. The device and method embodiments described above are merely illustrative. For example, the flowcharts in the accompanying drawings show the possible architectures, functions, and operations of the devices, methods, and computer program products according to multiple embodiments of the present invention. In this regard, each box in the flowchart or block diagram can represent a module, program segment, or part of a code, which contains one or more executable instructions for implementing the specified logical functions. It should also be noted that in some alternative implementations, the functions marked in the boxes can also occur in an order different from that marked in the drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, as well as the combination of boxes in the block diagram and / or flowchart, can be implemented using a dedicated hardware-based system that performs the specified functions or actions, or can be implemented using a combination of dedicated hardware and computer instructions.

[0181] In addition, the functional modules in the various embodiments of the present invention may be integrated together to form an independent part, or each module may exist independently, or two or more modules may be integrated to form an independent part.

[0182] If the functions are implemented in the form of software modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, electronic device, or network device, etc.) to perform all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage media include various media that can store program code, such as USB flash drives, mobile hard drives, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical disks. It should be noted that, in this document, the terms "comprise," "include," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, article, or device that includes a series of elements includes not only those elements but also other elements not explicitly listed, or also includes elements inherent to such process, method, article, or device. Without further constraints, an element defined by the phrase "comprises a..." does not preclude the existence of additional identical elements in the process, method, article or apparatus that includes the element.

[0183] The terms used in the embodiments of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The singular forms "a", "an", "the" and "the" used in the embodiments of the present invention and the appended claims are also intended to include plural forms unless the context clearly indicates otherwise.

[0184] It should be understood that the term "and / or" as used herein is merely a description of the relationship between associated objects, indicating that three possible relationships exist. For example, "A and / or B" can represent: A exists alone, A and B exist simultaneously, or B exists alone. Furthermore, the character " / " in this document generally indicates that the associated objects are in an "or" relationship.

[0185] The word "if," as used herein, may be interpreted as "at the time of" or "when" or "in response to determining" or "in response to detecting," depending on the context. Similarly, the phrases "if it is determined" or "if (stated condition or event) is detected" may be interpreted as "when it is determined" or "in response to the determination" or "when detecting (stated condition or event)" or "in response to detecting (stated condition or event)," depending on the context.

[0186] The "first" and "second" mentioned in the embodiments are merely used to distinguish similar objects and do not represent a specific ordering of the objects. It is understood that the specific order or precedence of "first" and "second" can be interchanged where appropriate. It should be understood that the objects distinguished by "first" and "second" can be interchanged where appropriate, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein.

[0187] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art will readily appreciate that various modifications and variations of the present invention are possible. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention are intended to be within the scope of protection of the present invention.

Claims

1. A diagnostic prediction device based on knowledge caching and personalized federated learning, applied to the edge server of a distributed medical network, characterized in that: It includes a dynamic knowledge cache construction unit, an index relationship mapping unit, a most similar sample mapping unit, a knowledge cache update unit, a knowledge vector retrieval unit, a comprehensive knowledge representation generation unit, and a personalized learning unit; The dynamic knowledge cache construction unit is used to receive medical sample index information uploaded by multiple clients and construct a dynamic knowledge cache; wherein the medical sample index information includes a hash value, index and label of the medical sample; The index relationship mapping unit is used to create an index relationship mapping table of label to index and index to hash value according to the medical sample index information, so as to associate each label with a corresponding index and each index with a corresponding hash value; The most similar sample mapping unit is used to initialize each medical sample and establish an index relationship mapping between each medical sample and R most similar samples; The knowledge cache updating unit is used to receive the category probability vector and its index uploaded by each client, so as to update the latest inference prediction result of the client to the knowledge cache; The knowledge vector retrieval unit is used to update the index relationship mapping table according to the updated knowledge cache, and retrieve the knowledge vectors of the R most similar samples for each medical sample; The comprehensive knowledge representation generating unit is used to fuse the knowledge vectors to generate comprehensive knowledge representation by adopting a weighted average method; The personalized learning unit is used to distribute the comprehensive knowledge representation to each client; wherein the comprehensive knowledge representation is used to trigger the client model of each client to perform personalized federated learning training of the distributed medical network based on the comprehensive knowledge representation, and perform medical diagnosis prediction based on the trained client model and output a prediction label; When establishing the index relationship mapping between each medical sample and the R most similar samples, The cosine similarity between the hash values of all medical samples is calculated, and a time decay factor is introduced to reduce the weight of the medical sample data uploaded earlier. The R most similar samples are screened out, and the improved HNSW algorithm is used to accelerate the retrieval process. The expression is: ； in, represents the R most similar samples of the i-th image sample of the k-th client; Represents two hash values 、 The cosine similarity of k, Represents the index of the client; i, j represent the index of the image sample; Indicates the time decay rate, which is used to control the influence of historical data; t indicates the current timestamp; Indicates the The upload timestamp of the jth image sample of the client; Indicates selecting the top R samples with the highest similarity; The category probability vector is generated by each client using the client model to infer and predict the input medical sample and generate the predicted category probability corresponding to the medical sample through forward propagation; the number of elements in the category probability vector is consistent with the number of categories.

2. A diagnostic prediction device based on knowledge caching and personalized federated learning according to claim 1, characterized in that ,The medical sample index information is obtained by each client performing feature fusion and privacy encoding on the ,locally stored medical image samples, specifically: Each client reads a medical sample dataset in a local database; wherein the medical sample dataset includes medical image samples, patient metadata, and diagnostic labels, which are expressed as: ； in, represents a medical sample dataset; k∈{1,2,…,K} represents the kth client, and K represents the total number of clients in the distributed medical network; is the i-th image sample; for corresponding patient metadata; for corresponding diagnostic labels; is the sample size; Each client uses its own pre-trained client model to process medical image samples and patient metadata to extract corresponding image features and text features. The client model includes a convolutional neural network and a Deepseek model. The convolutional neural network processes the input medical image samples and extracts the spatial features of the image through convolution and pooling operations to capture the texture and structural information in the image. The image features are expressed as follows: ； in, represents the image features of the i-th image sample of the k-th client; 、 are the pre-trained weight and bias parameters respectively; The Deepseek model processes the input patient metadata and extracts text semantic features through matrix operations to reflect the individual characteristics of the patient. The text features are expressed as follows: ； in, Represents patient metadata Text features of 、 are the pre-trained weight and bias parameters respectively; The image features and the text features are combined to obtain a comprehensive feature vector, which is expressed as: ； in, is the comprehensive feature vector corresponding to the i-th image sample; Represents a splicing operation; The differential privacy hash algorithm is used to encode and protect the comprehensive feature vector. That is, the comprehensive feature vector is mapped to a low-dimensional space through a random projection matrix, and then Gaussian noise perturbation is added to generate a hash value of fixed length. The expression is: ； ； in, is the hash value corresponding to the i-th image sample of the k-th client; DPH represents the differential privacy hash algorithm; H is the random projection matrix; Indicates that the mean square is 0 and the variance is Gaussian noise; and is the differential privacy parameter; ΔF is the sensitivity of the feature, indicating the maximum range of variation of the feature; sign represents the encoding operation; Each client uploads the corresponding hash value, diagnostic label, and index obtained from the comprehensive feature vector to the edge server through an encrypted channel.

3. A diagnostic prediction device based on knowledge caching and personalized federated learning according to claim 1, characterized in that When initializing each medical sample, each medical sample is initialized as a knowledge vector of an all-zero vector, representing its initial state before training; the dimension of the knowledge vector is consistent with the number of categories of the diagnostic label.

4. A diagnostic prediction device based on knowledge caching and personalized federated learning according to claim 1, characterized in that ,The expression of the comprehensive knowledge representation is: ； ； in, representing the comprehensive knowledge representation; The knowledge vector representing the i-th image sample of the k-th client, that is, the knowledge vector corresponding to the index (k, i); Represents the knowledge vector of the s-th similar sample; Represents the weight of the corresponding knowledge vector; R represents the number of similar samples.

5. A diagnostic prediction device based on knowledge caching and personalized federated learning according to claim 4, characterized in that , the weight of the comprehensive knowledge representation The cosine similarity of the hash values corresponding to the image samples and the similarity of the patient metadata are normalized and expressed as: ； ； in, represents the hash value of the i-th image sample of the k-th client; 、 Respectively represent the sth, The hash value of similar samples; cos represents cosine similarity; represents the patient metadata of the i-th image sample of the k-th client; 、 Respectively represent the sth, Patient metadata for similar samples; Indicates similarity; represents the L2 norm; 、 Indicates the patient age of the corresponding sample, through patient metadata get; 、 The vector representing the severity of the patient's condition for the corresponding sample is obtained by combining the patient metadata with the corresponding diagnostic label and scoring by experts.

6. A diagnostic prediction device based on knowledge caching and personalized federated learning according to claim 1, characterized in that When each client performs personalized federated learning training on a distributed medical network, a comprehensive objective function is used to optimize the model; wherein, the comprehensive objective function includes a supervision loss function and a distillation loss function, and is calculated by dynamic weighting; The supervised loss function is calculated using cross entropy loss, which adjusts the model parameters by comparing the difference between the model predicted category probability and the true label to ensure that the model prediction is consistent with the local diagnostic label; The distillation loss function uses KL divergence to measure the distribution difference between the local category probability vector and the comprehensive knowledge representation, so that the local reasoning result is consistent with the knowledge cache; the expressions are: ； ； ； in, represents a medical sample dataset; Represents the client model comprehensive objective function The trainable parameters of is the i-th image sample; is the true value of the label; Represents the true value of the c-th class label; represents the supervised loss function; express The predicted probability of represents the softmax normalization function; represents the dynamic weight of the comprehensive objective function; KL represents the distillation loss function; represents the predicted probability on the knowledge cache; C represents the total number of label categories; represents the predicted probability of the c-th class label; represents the predicted probability of the c-th class label on the knowledge cache; represents the probability prediction function.

7. A diagnostic prediction device based on knowledge caching and personalized federated learning according to claim 6, characterized in that , the dynamic weight of the comprehensive objective function Adjusted according to the heterogeneity of local medical sample data and the severity of the patient's condition, the formula is: ； in, Represents a local medical sample dataset Global data distribution on knowledge cache KL divergence of Represents patient metadata The severity score of the disease is obtained by combining the patient metadata with the corresponding labels and then scoring by experts; 、 are the weights corresponding to the KL divergence and the disease severity score; Each client calculates the gradient of the comprehensive objective function and updates the trainable parameters of the model according to the learning rate , the expression is: ； in, is the learning rate; ) is the comprehensive objective function right gradient.

8. A diagnosis and prediction system based on knowledge caching and personalized federated learning, characterized by: It includes an edge server and multiple clients; each client is connected to the edge server through communication; The client is used to read the medical sample dataset in the local database and process the medical sample dataset using a pre-trained client model to extract corresponding features for splicing and fusion to obtain a comprehensive feature vector; the comprehensive feature vector is then encoded and protected using a differential privacy hashing algorithm to obtain medical sample index information, which is uploaded to the edge server; the medical sample index information includes the hash value, index and label of the medical sample; The edge server is configured to construct a dynamic knowledge cache based on the medical sample index information received from multiple clients; and to create an index relationship mapping table from label to index and from index to hash value based on the medical sample index information, so as to associate each label with a corresponding index and each index with a corresponding hash value; and then, initialize each medical sample and establish an index relationship mapping between each medical sample and the R most similar samples; The client is used to use the client model to infer the input medical sample data and generate a category probability vector through forward propagation, and upload it to the edge server together with the corresponding index; The edge server is used to receive category probability vectors and their indexes uploaded by multiple clients to update the client's latest inference prediction results to the knowledge cache; update the index relationship mapping table based on the updated knowledge cache, and retrieve the knowledge vectors of the R most similar samples for each medical sample; then use a weighted average method to fuse the knowledge vectors to generate a comprehensive knowledge representation, and distribute the comprehensive knowledge representation to each client; The client is used to receive the comprehensive knowledge representation distributed by the edge server and perform personalized federated learning and optimization training of the distributed medical network; the trained model is used for remote medical imaging diagnosis and prediction; When establishing the index relationship mapping between each medical sample and the R most similar samples, The cosine similarity between the hash values of all medical samples is calculated, and a time decay factor is introduced to reduce the weight of the medical sample data uploaded earlier. The R most similar samples are screened out, and the improved HNSW algorithm is used to accelerate the retrieval process. The expression is: ； in, represents the R most similar samples of the i-th image sample of the k-th client; Represents two hash values 、 The cosine similarity of k, Represents the index of the client; i, j represent the index of the image sample; Indicates the time decay rate, which is used to control the influence of historical data; t indicates the current timestamp; Indicates the The upload timestamp of the jth image sample of the client; Indicates selecting the top R samples with the highest similarity; The category probability vector is generated by each client using the client model to infer and predict the input medical sample and generate the predicted category probability corresponding to the medical sample through forward propagation; the number of elements in the category probability vector is consistent with the number of categories.

Images