Primary liver cancer discharge preparation evaluation system and method based on patient information

Abstract

The invention is applicable to the technical field of medical control systems, and provides a primary liver cancer discharge preparation evaluation system and method based on patient information, and the system comprises an information extraction module which is used for extracting key entities from patient case information; the health propaganda and education recommendation module is used for recommending personalized health propaganda and education content to the patient according to the information extraction result; the patient port module is used for filling a hospital discharge preparation measurement table by a patient; the medical care module is used for completing patient admission assessment; the hospital discharge evaluator module is used for calculating a comprehensive score according to the scores of the patient port module and the medical care module and judging whether the patient meets the hospital discharge standard, the workload of medical care personnel can be relieved, meanwhile, a more scientific and thought hospital discharge preparation scheme can be provided for the patient, and the hospital discharge accuracy of the patient is improved. The traditional Chinese medicine composition has important practical significance for improving the overall treatment effect of the primary liver cancer.

Description

Technical Field

[0001] The present invention relates to the technical field of medical control systems, and in particular to a primary liver cancer discharge preparation evaluation system and method based on patient information. Background Art

[0002] In healthcare services, primary liver cancer is a serious condition whose treatment and discharge management require precise and detailed assessments. Currently, most hospitals use evaluation systems that rely primarily on the clinical experience of medical staff and the results of regular examinations, such as liver function tests, tumor marker levels, and imaging studies. While these traditional methods can assess a patient's condition and treatment effectiveness to a certain extent, they lack a systematic, standardized solution for determining whether a patient meets discharge criteria and for effectively and comprehensively considering the patient's physical condition, treatment response, and changes in their condition after continued treatment.

[0003] Existing evaluation methods can lead to the following problems: Evaluation results may be biased due to individual differences; they fail to fully consider multiple factors, including patient physical data, medication dosage, treatment frequency, and treatment progress; and they struggle to provide patients with comprehensive discharge guidance and follow-up management recommendations. Therefore, in response to these current circumstances, there is an urgent need to develop a patient-informed primary liver cancer discharge readiness evaluation system and method to overcome the shortcomings of current practical applications. Summary of the Invention

[0004] The purpose of the present invention is to provide a primary liver cancer discharge preparation evaluation system and method based on patient information, aiming to solve the problems in the above-mentioned background technology.

[0005] The present invention is implemented as follows: a primary liver cancer discharge preparation evaluation system based on patient information, comprising:

[0006] An information extraction module, configured to extract key entities from patient case information;

[0007] A health education recommendation module is used to recommend personalized health education content to patients based on the information extraction results;

[0008] A patient port module, wherein the patient port module is used for patients to fill in a discharge readiness measurement form;

[0009] A medical care module, which is used to complete the patient admission assessment;

[0010] and a discharge evaluator module, which is used to calculate a comprehensive score based on the scores of the patient port module and the medical care module to determine whether the patient meets the discharge criteria.

[0011] As a further solution of the present invention: the information extraction module includes a named entity recognition model, and the named entity recognition model includes a BERT pre-trained model, a BiLSTM layer and a CRF layer;

[0012] The BERT pre-trained model is used to extract character features from case text and generate a 768-dimensional feature vector;

[0013] The BiLSTM layer is used to extract bidirectional semantic features of text;

[0014] The CRF layer is used to optimize the label sequence through the Viterbi algorithm and output the optimal labeling result.

[0015] As a further solution of the present invention: the input of the BERT pre-training model includes character index, segment vector and position vector, and feature extraction is achieved through a multi-head self-attention mechanism and a multi-layer Transformer encoder. The bottom layer extracts basic semantic information, and the top layer extracts deep semantic information, which are fused to form a comprehensive semantic feature vector.

[0016] As a further solution of the present invention: the health education recommendation module adopts a user-based collaborative filtering algorithm to construct a user-content interaction matrix, calculates user similarity, and recommends health education content to target users based on the historical behavior of similar users.

[0017] As a further solution of the present invention: the discharge readiness scale of the patient port module includes scoring items such as physical condition preparation, physical strength, energy, self-care ability, understanding of self-care, ability to deal with life needs, ability to complete medical treatment, emotional support, personal care assistance, housework assistance and medical care needs assistance, and the patient scores each item on a scale of 0-10.

[0018] As a further solution of the present invention: the admission assessment form of the medical care module includes assessment indicators such as age, admission status, daily living activities ability, carrying tubes and admission status, past hospitalization history, self-care ability, pain assessment ability, family or caregiver ability, cooperation and cognitive status. The system automatically fills in the score based on the information extraction result, and sends it to the discharge evaluator module after confirmation by the treating physician.

[0019] As a further solution of the present invention: the scoring calculation formula of the discharge assessor module is:

[0020]

[0021] Among them, P u For patients to self-assess, P s The total score of the patient's self-assessment, D d For admission assessment form evaluation, D sThe total score of the admission evaluation form is calculated according to the scoring formula. If the patient's comprehensive score is greater than 0.6, the discharge evaluator module determines that the patient meets the discharge criteria; otherwise, the patient needs to continue to receive medical and nursing services and will not be discharged for the time being.

[0022] A method for evaluating discharge readiness for primary liver cancer based on patient information is applied to the above-mentioned system for evaluating discharge readiness for primary liver cancer based on patient information. The method comprises the following steps:

[0023] Step 1: Use the BIO annotation method to construct a training dataset for the named entity recognition model and divide the data into a training set and a test set;

[0024] Step 2: Use the training set to train the named entity recognition model, which is processed through the BERT pre-training model, BiLSTM layer, and CRF layer in sequence to output the sequence labeling results;

[0025] Step 3: Based on the collaborative filtering algorithm, recommend health education content to patients according to the sequence annotation results;

[0026] Step 4: The patient fills out the discharge readiness scale through the patient portal module, and the score is sent to the discharge assessor module;

[0027] Step 5: The medical care module automatically generates the admission assessment form score based on the sequence annotation results, and sends it to the discharge assessment module after confirmation by the physician;

[0028] Step 6: The discharge assessor module calculates a comprehensive score based on the scoring formula, determines discharge eligibility, and provides personalized guidance.

[0029] As a further solution of the present invention: in the BIO marking method, BX represents the beginning of the tag type X, IX represents the middle position of the tag type X, and O represents not belonging to any tag type.

[0030] As a further solution of the present invention: the Bi LSTM layer includes forward and reverse LSTM units, and information screening is achieved through a forget gate, a memory gate, and an output gate. The output feature vector is passed through a softmax layer to generate a label probability set.

[0031] Compared with the prior art, the present invention has the following beneficial effects:

[0032] This invention focuses on the discharge readiness assessment for patients with primary liver cancer. Utilizing natural language processing (NLP) and recommendation algorithms, it fundamentally addresses the current technological gap in assessing whether patients with primary liver cancer are suitable for discharge. This not only conserves hospital medical resources but also alleviates the additional financial pressure on patients who meet discharge criteria but fail to be discharged in the later stages of their treatment. The specific benefits are as follows:

[0033] Accurately assessing patient discharge readiness: By building and training a named entity recognition (NER) model, the system can accurately extract and identify key content (such as diseases, symptoms, medications, and treatments) from patient case information. Utilizing a BERT pre-trained model, a bidirectional long short-term memory (Bi-LSTM) layer, and a conditional random field (CRF) layer, the model ensures high recognition rate and accuracy, significantly improving the system's accuracy in identifying key entities in primary liver cancer cases.

[0034] Personalized health education content recommendation: Based on the user's collaborative filtering algorithm, the system associates the patient's sequence annotation results with health education content and recommends relevant content to the target user by calculating the learning behavior of similar users. This personalized recommendation method makes health education content more tailored to the patient's actual needs, improving the utilization rate of health education content and learning effect of patients;

[0035] Efficient discharge readiness assessment process: The system integrates the patient portal and the medical and nursing modules. After the patient completes the discharge readiness scale, the information is automatically sent to the discharge assessor module. The system automatically fills in the admission assessment form based on the sequence annotation results. After confirmation by the treating physician in the medical and nursing module, it is sent to the discharge assessor module. This automated assessment process reduces manual operation and subjective bias, improving assessment efficiency.

[0036] Reliable discharge assessment mechanism: The discharge assessment module uses a comprehensive scoring system, with the patient's discharge scale accounting for 40% and the medical and nursing module's assessment scale accounting for 60% to determine the final score. A score greater than 0.6 indicates that the patient can be discharged; otherwise, continued medical and nursing services are required. This comprehensive assessment mechanism ensures the comprehensiveness and reliability of the discharge assessment, ensuring a safe and smooth transition for patients after discharge. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the specific embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0038] Figure 1 It is a system flow chart of the present invention.

[0039] Figure 2 Schematic diagram of the information extraction module in the present invention.

[0040] Figure 3 This is the BERT model diagram in the present invention.

[0041] Figure 4Schematic diagram of the BiLSTM layer structure in the present invention.

[0042] Figure 5 Schematic diagram of the collaborative filtering algorithm in the present invention.

[0043] Figure 6 This is a functional diagram of the modules in the present invention. DETAILED DESCRIPTION

[0044] The technical solution of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the embodiments described are only some embodiments of the present invention, not all embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0045] In the description of the present invention, it should be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," etc., indicating orientations or positional relationships, are based on the orientations or positional relationships shown in the accompanying drawings and are intended solely to facilitate and simplify the description of the present invention. They are not intended to indicate or imply that the devices or components referred to must have, be constructed, or operate in a specific orientation, and therefore should not be construed as limitations on the present invention. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0046] In the description of the present invention, it should be noted that, unless otherwise expressly specified or limited, the terms "mounted," "connected," and "connected" should be understood in a broad sense. For example, they may refer to fixed, detachable, or integral connections; mechanical or electrical connections; direct or indirect connections through an intermediary; and internal communication between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on specific circumstances.

[0047] The present invention will be further explained below with reference to specific embodiments.

[0048] See also Figures 1-6 The embodiment of the present invention provides a primary liver cancer discharge preparation evaluation system based on patient information, comprising:

[0049] An information extraction module, configured to extract key entities from patient case information; the information extraction module comprising a named entity recognition model, comprising a BERT pre-trained model, a bidirectional long short-term memory network (BiLSTM) layer, and a conditional random field (CRF) layer;

[0050] The BERT pre-trained model is used to extract character features from case text and generate a 768-dimensional feature vector;

[0051] The BiLSTM layer is used to extract bidirectional semantic features of text;

[0052] The CRF layer is used to optimize the label sequence through the Viterbi algorithm and output the optimal annotation result. The input of the BERT pre-trained model includes character index, segment vector and position vector. Feature extraction is achieved through a multi-head self-attention mechanism and a multi-layer Transformer encoder. The bottom layer extracts basic semantic information and the top layer extracts deep semantic information, which are then fused to form a comprehensive semantic feature vector.

[0053] A health education recommendation module is used to recommend personalized health education content to patients based on the information extraction results. The module uses a user-based collaborative filtering algorithm to construct a user-content interaction matrix, calculates user similarity, and recommends health education content to target users based on the historical behavior of similar users.

[0054] A patient portal module is used for patients to complete a discharge readiness scale; the discharge readiness scale includes scoring items such as physical readiness, physical strength, energy, self-care ability, understanding of self-care, ability to handle life needs, ability to complete medical treatment, emotional support, personal care assistance, housework assistance, and medical care needs assistance. Patients score each item on a scale of 0-10.

[0055] The medical care module is used to complete the patient admission assessment. The admission assessment form of the medical care module includes assessment indicators such as age, admission status, daily living activities ability, tube carrying and admission status, previous hospitalization history, self-care ability, pain assessment ability, family or caregiver ability, cooperation and cognitive status. The system automatically fills in the score based on the information extraction results, and after confirmation by the treating physician, it is sent to the discharge assessment module;

[0056] And a discharge evaluator module, which is used to calculate a comprehensive score based on the scores of the patient port module and the medical care module to determine whether the patient meets the discharge criteria; the score calculation formula of the discharge evaluator module is:

[0057]

[0058] Among them, P u For patients to self-assess, P s The total score of the patient's self-assessment, D d For admission assessment form evaluation, D sThe total score of the admission evaluation form is calculated according to the scoring formula. If the patient's comprehensive score is greater than 0.6, the discharge evaluator module determines that the patient meets the discharge criteria; otherwise, the patient needs to continue to receive medical and nursing services and will not be discharged for the time being.

[0059] In an embodiment of the present invention, by constructing and training a named entity recognition (NER) model, the system can accurately extract and identify key content (such as diseases, symptoms, drugs, and treatment methods) from patient case information. Utilizing the BERT pre-training model, bidirectional long short-term memory network (BiLSTM), and conditional random field (CRF) layer, the model's high recognition rate and accuracy are ensured, greatly improving the system's accuracy in identifying key entities in primary liver cancer cases. Based on a user-based collaborative filtering algorithm, the system associates the patient's sequence annotation results with health education content and recommends relevant content to target users by calculating the learning behavior of similar users. This personalized recommendation method makes health education content more tailored to the patient's actual needs, improving the utilization rate and learning effect of the patient's health education content. The system integrates the patient port and medical care module. After the patient fills out the discharge readiness scale, the information is automatically sent to the discharge evaluator module; the system automatically fills in the admission assessment form based on the sequence annotation results, which is confirmed by the treating physician in the medical care module and sent to the discharge evaluator module. This automated evaluation process reduces manual operation and subjective bias, improving evaluation efficiency. The Discharge Evaluator module uses a comprehensive scoring mechanism, with the patient's discharge scale accounting for 40% and the medical and nursing module's assessment scale accounting for 60% to determine the final score. A score greater than 0.6 indicates that the patient can be discharged; otherwise, continued medical and nursing care is required. This comprehensive assessment mechanism ensures the comprehensiveness and reliability of the discharge assessment, ensuring a safe and smooth transition for patients after discharge.

[0060] See also Figures 1-6 The embodiment of the present invention provides a method for evaluating discharge preparation for primary liver cancer based on patient information, which is applied to the above-mentioned system for evaluating discharge preparation for primary liver cancer based on patient information. The method includes the following steps:

[0061] Step 1: Construct a dataset for training the named entity recognition model. Use the BIO annotation method to annotate each character in the patient case information. After the annotation is completed, divide the annotated information into a training set and a test set.

[0062] The BIO annotation marks each element in the text sequence as "BX", "IX", or "O", where "BX" indicates that the tag containing this element belongs to type X and is the beginning of the current tag. "IX" indicates that the tag containing this element belongs to class X and is located in the middle of the current tag. "O" indicates that the current element does not belong to any type.

[0063] Step 2: Build a named entity recognition model and use the training set to train the model. The named entity recognition model mainly includes the BERT pre-training model, the bidirectional long short-term memory network BiLSTM layer and the conditional random field CRF layer. The information extraction module is as follows: Figure 2 As shown, the module operation process is as follows:

[0064] Step 21: Use the BERT pre-trained model to extract character features from all samples in the training set (such as Figure 3 ). First, before feeding the sample into the BERT model, the vocabulary is queried to determine the index of each character in the sample. Next, these word indices are fed into the BERT model along with the segment vector and position vector. The BERT model uses its built-in encoding mechanism and multi-head self-attention mechanism to convert each character into a 768-dimensional feature vector. This feature vector can represent the multiple meanings of the same character in different contexts. The self-attention calculation formula is as follows:

[0065]

[0066] Symbols meaning:

[0067] -Q: query matrix, dimension is (n×d k ), where n is the sequence length, d k is the dimension of the query vector.

[0068] -K: key matrix, dimension is (m×d k ), where m is the length of the key-value pairs (usually the same as the sequence length).

[0069] -V: value matrix, dimension is (m×d v ), d v is the dimension of the value vector.

[0070] -d k : The dimension of the key vector, used to scale the dot product to avoid gradient vanishing due to excessive values.

[0071] -T means matrix transpose.

[0072]

[0073] Symbols meaning:

[0074] - : The query projection matrix of the i-th head, dimension is (d model ×d k )

[0075] - : The key projection matrix of the i-th head, dimension is (d model ×d k )

[0076] - : The value projection matrix of the i-th head, dimension is (d model ×d v )

[0077] -d model : The dimension of the input vector (usually 512 in Transformer)

[0078] -d k ,d v : The dimension of the key / value vector of each header, usually d k =d v =d model / h, where h is the number of heads.

[0079] MultiHead(Q,K,V)=Concat(head1,…,head h )W O

[0080] Symbols meaning:

[0081] -h: The number of attention heads (typically 8 in Transformer).

[0082] -Concat: concatenate the outputs of all heads along the column, the concatenated dimensions are (n×h·d v )

[0083] -W O (Output Projection Matrix): Multi-head attention output projection matrix (O = Output), dimension is (h·d v ×d model )

[0084] Where Q is the query matrix, K is the key matrix, and V is the value matrix. The multi-head attention mechanism uses a mapping matrix, which multiplies the Q, K, and V matrices, maps them to a smaller dimension, and then concatenates them using Concat. The result is then multiplied by the mapping matrix WO to produce the final result of the multi-head attention mechanism. Each character's features are then independently processed through a feedforward neural network layer to capture local features. The multi-layer Transformer encoder extracts features from each sample layer by layer, extracting basic semantic information from the bottom layer and deeper semantic information from the top layer. The features from the bottom and top layers are then fused, and the mean operation is used to combine the semantic information from different layers to form a comprehensive semantic feature vector. Finally, the feature vector is mapped to 768 dimensions.

[0085] Step 22: Input the semantic feature vector obtained in step 21 into the Bidirectional Long Short-Term Memory Network (BiLSTM) layer for processing (e.g. Figure 4 ), the input text is first converted into various feature vectors (such as word embeddings, position embeddings, and grammatical features), which are then concatenated along the feature dimensions to form a unified input vector. This input vector is fed into the BiLSTM layer, which contains forward and reverse LSTM units, each with 150 dimensions. Leveraging the bidirectional memory network characteristics of BiLSTM, global information from front to back and back to front can be effectively extracted. The BiLSTM layer primarily consists of a forget gate, a memory gate, and an output gate. The forget gate determines what information to discard, the memory gate determines what information to remember, and the output gate, based on the forget and memory gates, determines the final output information. Ultimately, the feature vector output by the BiLSTM layer can be used through a softmax layer to obtain a set of probabilities for the labels corresponding to each character.

[0086] Step 23: The label probability set obtained in Step 22 is input to the Conditional Random Field (CRF) layer. The CRF layer autonomously learns the features between label sequences and applies corresponding rules and constraints to the label sequences. When calculating the optimal label sequence, the CRF layer uses the Viterbi algorithm for optimization. The Viterbi algorithm effectively reduces the label sequence optimization time through dynamic programming, improving the overall performance and accuracy of the model. Ultimately, the CRF layer outputs the optimal label sequence, providing high-quality sequence labeling results.

[0087] Step 3: Use user-based collaborative filtering algorithms (such as Figure 5 The process of recommending health education content based on sequence annotation results (such as diseases, symptoms, and therapeutic drugs) is as follows: First, the user's sequence annotation results are associated with the corresponding health education content, and a user-content interaction matrix is ​​constructed. The values ​​in the matrix represent the user's interest in or learning behavior of the content. Then, based on this matrix, a user-based collaborative filtering algorithm is used to calculate the similarity between users to find other users with similar sequence annotation results as the target user. Based on the historical behavior of these similar users, health education content that they have learned or are interested in is recommended to the target user, thereby providing personalized health learning suggestions.

[0088] Step 4: On the Discharge Readiness Scale prepared in the patient portal, patients are asked to fill out a score from 0 to 10 based on their actual feelings regarding physical readiness, strength, energy, self-care ability, understanding of self-care, ability to cope with daily needs, ability to complete medical treatments, and access to emotional support, personal care assistance, household chores, and medical care needs. After completing the score in the patient portal, the patient's score is sent to the Discharge Evaluator module.

[0089] Step 5: The patient admission assessment form in the medical care module includes multiple assessment indicators and scoring items, such as age, admission status, ability to carry a tube and admission, previous hospitalization history, self-care ability, pain assessment ability, family or caregiver ability, cooperation, and cognitive status. The medical care module first automatically fills in the scores for each assessment item based on the sequence annotation results and calculates the total score, which requires confirmation by the treating physician. The scoring data is then sent to the discharge evaluator module through the medical care module.

[0090] Step 6: The discharge assessor module calculates the scores based on the patient's discharge scale and admission assessment form. The scoring formula is as follows:

[0091]

[0092] Among them, P u For patients to self-assess, P s The total score of the patient's self-assessment, D d For admission assessment form evaluation, D s This module calculates the score based on the admission assessment form. If the patient's overall score is greater than 0.6, the module determines that the patient meets discharge criteria. Otherwise, the patient will continue to receive medical and nursing services and will not be discharged temporarily. This module assesses the patient's discharge readiness and the subsequent support measures needed. The medical team then provides the patient with personalized discharge guidance and support recommendations to ensure a stable transition to home care after discharge.

[0093] In this embodiment, the method can not only reduce the workload of medical staff, but also provide patients with a more scientific and thoughtful discharge preparation plan, which has important practical significance for improving the overall treatment effect of primary liver cancer.

[0094] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A primary liver cancer discharge preparation evaluation system based on patient information, characterized in that: include: An information extraction module, configured to extract key entities from patient case information; A health education recommendation module is used to recommend personalized health education content to patients based on the information extraction results; A patient port module, wherein the patient port module is used for patients to fill in a discharge readiness measurement form; A medical care module, which is used to complete the patient admission assessment; and a discharge evaluator module, which is used to calculate a comprehensive score based on the scores of the patient port module and the medical care module to determine whether the patient meets the discharge criteria.

2. The primary liver cancer discharge preparation evaluation system based on patient information according to claim 1, characterized in that: The information extraction module includes a named entity recognition model, and the named entity recognition model includes a BERT pre-training model, a BiLSTM layer and a CRF layer; The BERT pre-trained model is used to extract character features from case text and generate a 768-dimensional feature vector; The BiLSTM layer is used to extract bidirectional semantic features of text; The CRF layer is used to optimize the label sequence through the Viterbi algorithm and output the optimal labeling result.

3. The primary liver cancer discharge preparation evaluation system based on patient information according to claim 2, characterized in that: The input of the BERT pre-trained model includes character index, segment vector and position vector. Feature extraction is achieved through a multi-head self-attention mechanism and a multi-layer Transformer encoder. The bottom layer extracts basic semantic information, and the top layer extracts deep semantic information, which are then fused to form a comprehensive semantic feature vector.

4. The primary liver cancer discharge preparation evaluation system based on patient information according to claim 1, characterized in that: The health education recommendation module adopts a user-based collaborative filtering algorithm to construct a user-content interaction matrix, calculates user similarity, and recommends health education content to target users based on the historical behaviors of similar users.

5. The primary liver cancer discharge preparation evaluation system based on patient information according to claim 1, characterized in that: The discharge readiness scale of the patient portal module includes scoring items such as physical readiness, physical strength, energy, self-care ability, understanding of self-care, ability to handle life needs, ability to complete medical treatment, emotional support, personal care assistance, housework assistance and medical care needs assistance. Patients score each item on a scale of 0-10.

6. The primary liver cancer discharge preparation evaluation system based on patient information according to claim 1, characterized in that: The admission assessment form of the medical care module includes assessment indicators such as age, admission status, ability to carry daily living activities, tube carrying and admission status, past hospitalization history, self-care ability, pain assessment ability, family or caregiver ability, cooperation and cognitive status. The system automatically fills in the score based on the information extraction results, and sends it to the discharge evaluator module after confirmation by the treating physician.

7. The primary liver cancer discharge preparation evaluation system based on patient information according to claim 1, characterized in that: The scoring calculation formula of the discharge assessor module is: Among them, P u For patients to self-assess, P s The total score of the patient's self-assessment, D d For admission assessment form evaluation, D s The total score of the admission evaluation form is calculated according to the scoring formula. If the patient's comprehensive score is greater than 0.6, the discharge evaluator module determines that the patient meets the discharge criteria; otherwise, the patient needs to continue to receive medical and nursing services and will not be discharged for the time being.

8. A method for evaluating discharge readiness for primary liver cancer based on patient information, applied to a system for evaluating discharge readiness for primary liver cancer based on patient information as claimed in any one of claims 1 to 7, characterized in that: The method comprises the following steps: Step 1: Use the BIO annotation method to construct a training dataset for the named entity recognition model and divide the data into a training set and a test set; Step 2: Use the training set to train the named entity recognition model, which is processed through the BERT pre-training model, BiLSTM layer, and CRF layer in sequence to output the sequence labeling results; Step 3: Based on the collaborative filtering algorithm, recommend health education content to patients according to the sequence annotation results; Step 4: The patient fills out the discharge readiness scale through the patient portal module, and the score is sent to the discharge assessor module; Step 5: The medical care module automatically generates the admission assessment form score based on the sequence annotation results, and sends it to the discharge assessment module after confirmation by the physician; Step 6: The discharge assessor module calculates a comprehensive score based on the scoring formula, determines discharge eligibility, and provides personalized guidance.

9. The method for evaluating discharge readiness for primary liver cancer based on patient information according to claim 8, characterized in that: In the BIO marking method, BX indicates the beginning of tag type X, IX indicates the middle position of tag type X, and O indicates that it does not belong to any tag type.

10. The method for evaluating discharge readiness for primary liver cancer based on patient information according to claim 8, characterized in that: The BiLSTM layer contains forward and reverse LSTM units, and implements information screening through forget gate, memory gate and output gate. The output feature vector is passed through the softmax layer to generate a label probability set.

Images