Prediction method, system and equipment for liver transplantation prognosis of liver cancer

Abstract

The invention discloses a method, a system and equipment for predicting liver transplantation prognosis of liver cancer. The liver cancer liver transplantation prognosis prediction model construction method comprises the following steps: based on prior literature knowledge of liver cancer liver transplantation influence factors, carrying out unsupervised learning on a basic large language model to obtain an initial large language model; a data set is obtained, natural language transformation is conducted on the data set according to the cue word template, the initial large language model is trained, a pre-trained large language model is obtained, and the data set comprises clinical data of a plurality of liver transplantation patients; multi-dimensional information of typical cases in the liver transplantation patient is obtained to serve as tagged fine adjustment data, fine adjustment is conducted on the pre-trained large language model, a prediction model is obtained, and the prediction model is used for prognosis prediction of liver transplantation of liver cancer. By adopting the embodiment of the invention, the liver cancer liver transplantation prognosis prediction model with expert-level evaluation ability can be constructed, so that the accurate prediction of the prognosis of the liver cancer liver transplantation patient is realized.

Description

Technical Field

[0001] This application relates to the interdisciplinary fields of smart healthcare and artificial intelligence, and in particular to a method, system, and device for predicting the prognosis of liver transplantation in patients with liver cancer. Background Technology

[0002] The postoperative prognosis of liver transplant patients is crucial for the rational allocation of scarce liver resources, which necessitates accurate prediction of the prognosis of liver transplant patients.

[0003] Currently, prognostic prediction models for liver transplantation in liver cancer mainly fall into two categories: one is based on statistical methods, which construct predictive formulas through variable selection, function fitting, and correlation analysis; however, these methods make strict assumptions about data distribution. The other is based on machine learning methods, which directly extract features from labeled data and make predictions; however, these methods heavily rely on large-scale, multi-center structured clinical and omics data and often face problems of data heterogeneity and missing values, requiring extensive preprocessing. Furthermore, patients' cancer treatment histories are complex and diverse, and existing models often struggle to fully characterize the interactions between treatments, while also generally suffering from insufficient interpretability, thus hindering their widespread clinical application. Summary of the Invention

[0004] Therefore, it is necessary to provide a method, system, and device for predicting the prognosis of liver cancer liver transplantation, which can construct a predictive model for liver cancer liver transplantation with expert-level assessment capabilities, thereby achieving accurate prediction of the prognosis of liver cancer liver transplantation patients.

[0005] In a first aspect, embodiments of this application provide a method for constructing a predictive model for the prognosis of liver transplantation in liver cancer, including:

[0006] Based on prior literature knowledge of factors influencing liver transplantation in liver cancer, an unsupervised learning process is performed on the basic large language model to obtain an initial large language model.

[0007] A dataset is obtained, and after performing natural language conversion on the dataset according to the prompt word template, the initial large language model is trained to obtain a pre-trained large language model. The dataset includes clinical data from multiple liver transplant patients.

[0008] Multidimensional information of typical cases among liver transplant patients is obtained as labeled fine-tuning data. The pre-trained large language model is then fine-tuned to obtain a prediction model, which is used to predict the prognosis of liver transplantation for liver cancer.

[0009] Furthermore, based on prior literature knowledge of factors influencing liver transplantation in liver cancer, an unsupervised learning process is performed on the basic large language model to obtain an initial large language model, including:

[0010] Obtain the basic large language model;

[0011] Literature and guidelines on influencing factors of liver transplantation for liver cancer were collected. These literature and guidelines were used as prior knowledge to perform unsupervised learning on the basic large language model to obtain an initial large language model.

[0012] Further, after obtaining the dataset and performing natural language conversion on the dataset according to the prompt word template, the initial large language model is trained to obtain a pre-trained large language model. The dataset includes clinical data from multiple liver transplant patients, including:

[0013] Clinical data from multiple liver transplant patients were obtained, and a dataset was generated based on the clinical data from the multiple liver transplant patients.

[0014] Obtain the training set from the dataset;

[0015] Perform natural language conversion on the data in the training set based on the prompt word template;

[0016] The initial large language model is trained using the training set after natural language conversion to obtain the pre-trained large language model.

[0017] Furthermore, the prompt template is a five-segment template, which includes general information about the liver transplant patient, tumor characteristics, downstaging treatment, immune status, and nutritional assessment.

[0018] Furthermore, it also includes:

[0019] A validation set is obtained based on the dataset.

[0020] The prediction model is hyperparameter tuned and model calibrated based on the validation set.

[0021] Furthermore, it also includes:

[0022] The test set is obtained based on the dataset.

[0023] The prediction model is validated and its performance is evaluated based on the test set.

[0024] Furthermore, the multidimensional information obtained from typical cases among liver transplant patients, as labeled fine-tuned data, includes:

[0025] Typical cases from liver transplant patients were selected from the dataset;

[0026] The clinical data of the typical cases were processed using natural language processing to obtain multidimensional information on the typical cases among the liver transplant patients.

[0027] Secondly, a predictive method for the prognosis of liver transplantation in patients with liver cancer is provided, including:

[0028] Obtain clinical data from liver transplant patients;

[0029] The clinical data of the liver transplant patients are input into the prediction model to obtain the prognostic prediction results of liver cancer transplantation in the liver transplant patients. The prediction model is a prediction model obtained according to the prediction model construction method for the prognosis of liver cancer transplantation based on the first aspect and any possible implementation of the first aspect.

[0030] Thirdly, a predictive system for the prognosis of liver transplantation in liver cancer is provided, including:

[0031] The acquisition module is used to obtain clinical data from liver transplant patients.

[0032] The prediction module is used to input the clinical data of the liver transplant patient into the prediction model to obtain the prognostic prediction result of the liver transplant patient with liver cancer. The prediction model is a prediction model obtained according to the prediction model construction method for the prognosis of liver transplantation with liver cancer based on the first aspect and any possible implementation of the first aspect.

[0033] Fourthly, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the method for constructing a predictive model for the prognosis of liver transplantation for liver cancer as described in the first aspect and any possible implementation of the first aspect, or the method for predicting the prognosis of liver transplantation for liver cancer as described in the second aspect.

[0034] The embodiments of this application firstly involve unsupervised learning of a basic large language model based on prior literature knowledge of influencing factors of liver transplantation for liver cancer, resulting in an initial large language model. Secondly, a dataset is obtained, and natural language conversion is performed on the dataset according to prompt word templates to train the initial large language model, resulting in a pre-trained large language model. The dataset includes clinical data from multiple liver transplant patients. Finally, multidimensional information from typical cases among liver transplant patients is obtained as labeled fine-tuning data to fine-tune the pre-trained large language model, resulting in a prediction model. This prediction model is used to predict the prognosis of liver transplantation for liver cancer. Therefore, a prognostic prediction model for liver transplantation for liver cancer with expert-level evaluation capabilities can be constructed, thereby achieving accurate prediction of the prognosis of liver transplant patients with liver cancer. Attached Figure Description

[0035] Other features, objects, and advantages of this application will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:

[0036] Figure 1A flowchart of the method for constructing a predictive model for liver transplantation prognosis of liver cancer provided in the embodiments of this application;

[0037] Figure 2 This is a schematic diagram of the large model fine-tuning method provided in the embodiments of this application;

[0038] Figure 3 This is a schematic diagram of natural language conversion of clinical information provided in the embodiments of this application;

[0039] Figure 4 This is a model performance evaluation diagram provided in the embodiments of this application;

[0040] Figure 5 This is a structural block diagram of a liver cancer liver transplant prognosis prediction system provided in an embodiment of this application;

[0041] Figure 6 This is a structural block diagram of the computer device provided in the embodiments of this application. Detailed Implementation

[0042] The present application will now be described in further detail with reference to the embodiments and accompanying drawings. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the application. Furthermore, it should be noted that, for ease of description, only the parts relevant to the application are shown in the accompanying drawings.

[0043] It should be noted that, unless otherwise specified, the embodiments and features of the embodiments in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0044] The following describes in detail, with reference to the accompanying drawings, a method, system, and device for predicting the prognosis of liver transplantation for liver cancer according to embodiments of this application.

[0045] Figure 1 This is a flowchart illustrating a method for constructing a predictive model for liver transplantation prognosis in liver cancer according to an embodiment of this application. Figure 1 As shown, the method for constructing a predictive model for liver transplantation prognosis in liver cancer according to an embodiment of this application includes the following steps:

[0046] S101: Based on prior literature knowledge of factors influencing liver transplantation in liver cancer, an unsupervised learning process is performed on the basic large language model to obtain an initial large language model.

[0047] In one embodiment of this application, the step of performing unsupervised learning on a basic large language model based on prior literature knowledge of factors influencing liver transplantation for liver cancer to obtain an initial large language model includes: obtaining a basic large language model; collecting literature and guidelines on factors influencing liver transplantation for liver cancer, using the literature and guidelines on factors influencing liver transplantation for liver cancer as the prior literature knowledge, and performing unsupervised learning on the basic large language model to obtain an initial large language model.

[0048] Large Language Models (LLMs) are Transformer-based AI models capable of understanding and generating human language. Through deep semantic understanding and supervised fine-tuning, they can become expert-level models in the field. In a specific example, the GLM-4-Plus architecture is used as the foundation for the LLM model. In other examples, other architectures can be applied based on experience or practical needs.

[0049] In one embodiment of this application, prior literature knowledge includes, but is not limited to, systematic reviews and clinical summaries on risk factors for recurrence of liver cancer after transplantation, both domestically and internationally. By inputting this prior literature knowledge into the model, the model can learn and absorb core knowledge in the field of liver cancer and liver transplantation influencing factors through reading and understanding these texts, thereby establishing a basic understanding of this field and laying the foundation for subsequent training and prediction.

[0050] S102: Obtain the dataset, and after performing natural language conversion on the dataset according to the prompt word template, train the initial large language model to obtain a pre-trained large language model, wherein the dataset includes clinical data of multiple liver transplant patients.

[0051] In one embodiment of this application, obtaining a dataset, performing natural language conversion on the dataset according to a prompt word template, and training an initial large language model to obtain a pre-trained large language model, wherein the dataset includes clinical data of multiple liver transplant patients, comprising: obtaining clinical data of multiple liver transplant patients and obtaining a dataset based on the clinical data of multiple liver transplant patients; obtaining a training set from the dataset; performing natural language conversion on the data in the training set according to the prompt word template; and training the initial large language model based on the natural language converted training set to obtain the pre-trained large language model.

[0052] In a specific example, the dataset consists of the UNOS database cohort and the Tsinghua Changgung Hospital cohort. The UNOS database cohort includes 15,942 pathologically diagnosed liver cancer patients who underwent liver transplantation between 1994 and 2024. Patients who died perioperatively (within one month post-surgery), had insufficient follow-up (no death or recurrence within 3 years), or had incomplete medical records were excluded, resulting in 11,349 eligible patients. Stratified random sampling was used to divide these patients into training, validation, and test sets at a ratio of 70%:15%:15%. The Tsinghua Changgung Hospital cohort includes 233 liver cancer patients who underwent liver transplantation between 2018 and 2022. Patients who died perioperatively, had insufficient follow-up (<2 years), or had interrupted follow-up were excluded using the same criteria, resulting in 222 cases that met the analysis requirements. These cases were randomly divided into equal training and test sets.

[0053] Among them, such as Figure 3 As shown, the clinical information obtained from liver transplant patients is raw, unprocessed data. By logically transforming the clinical values ​​and examination results into natural language descriptions, the model's ability to understand and reason about complex contexts can be improved. For example, simple data indicators such as "Age: XX years old; Gender: XX..." can be transformed into: "Patient, XX years old, male, infected with hepatitis B virus for XX years, has been receiving XX treatment..." This approach aligns better with the thinking logic of medical experts and makes the model more interpretable.

[0054] In one embodiment of this application, the prompt template is a five-segment template, which includes general information about the liver transplant patient, tumor characteristics, downstaging treatment, immune status, and nutritional assessment.

[0055] The general criteria include, but are not limited to, age, gender, and comorbidities; tumor characteristics include, but are not limited to, the number and size of lesions; downstaging treatments include, but are not limited to, transcatheter arterial chemoembolization (TACE), radiotherapy, targeted therapy, and immunotherapy; immune status includes, but is not limited to, lymphocyte subsets and inflammatory factor levels; and nutritional assessments include, but are not limited to, albumin (ALB) and body mass index (BMI).

[0056] By logically transforming clinical information into the aforementioned natural language description, the model's learning ability can be enhanced, allowing it to more efficiently utilize previously learned prior literature knowledge. The training set data segmented in the above steps is then input into the model in this way to train a pre-trained model.

[0057] S103: Obtain multidimensional information of typical cases among liver transplant patients as labeled fine-tuning data, fine-tune the pre-trained large language model to obtain a prediction model, wherein the prediction model is used to predict the prognosis of liver transplantation for liver cancer.

[0058] In one embodiment of this application, obtaining multidimensional information of typical cases among liver transplant patients as labeled fine-tuned data includes: screening typical cases among liver transplant patients from the dataset; and performing natural language processing on the clinical data of the typical cases to obtain multidimensional information of the typical cases among liver transplant patients.

[0059] The purpose of fine-tuning labeled samples is to specifically learn the key patterns in liver cancer liver transplantation prediction, enabling the model to quickly match the detailed features of the scenario and optimize the model's performance.

[0060] In a specific example, such as Figure 2 As shown, 100 typical cases were selected from the UNOS database cohort and the Tsinghua Chang Gung Hospital cohort, and their clinical information and follow-up outcomes were input into a pre-trained large language model to construct a predictive model for the prognosis of liver cancer liver transplantation, namely the HCC-LT-PROBOT predictive model.

[0061] In one embodiment of this application, the method further includes: obtaining a validation set based on the dataset; and performing hyperparameter tuning and model calibration on the prediction model based on the validation set.

[0062] Hyperparameters are parameters that need to be manually set before deep learning training begins; these parameters cannot be automatically learned from training data. They directly affect model performance and the training process and require optimization. Hyperparameters include, but are not limited to, learning rate, number of model layers, and batch size. Model calibration refers to the process of adjusting the match between the output probability of the machine learning model and the actual prediction accuracy, ensuring that the model's prediction probability reflects true accuracy. For example, if the model's confidence level for predicting "patient survival" is generally too high on the validation set, calibration can reduce the output confidence level to make the prediction more reliable. By performing hyperparameter tuning and model calibration on the validation set, the accuracy and stability of the model's prognostic prediction for liver transplantation in liver cancer can be effectively improved.

[0063] In one embodiment of this application, the method further includes: obtaining a test set based on the dataset; and performing model validation and performance evaluation on the prediction model based on the test set.

[0064] Specifically, the prediction model proposed in this application was used on the independent test set divided in the above steps to calculate the prediction accuracy, F1 score and AUC value. At the same time, survival analysis (Kaplan-Meier curve) was combined to evaluate the survival prediction ability of the model in different risk stratifications, and a comparative analysis was conducted with the classic Milan criteria to verify its clinical application value.

[0065] The F1 score is the harmonic mean of the model's accuracy and recall, used to evaluate the overall performance of the model; the AUC value is the area under the ROC curve of the model, used to evaluate the model's ability to distinguish between positive and negative samples.

[0066] Model performance evaluation results are as follows Figure 4 As shown in Table 1, the HCC-LT-PROBOT prediction model proposed in this application has high performance metrics for 1-year and 2-year overall survival (OS) and recurrence-free survival (FS). It can be seen that the prediction model proposed in this application has high accuracy, model performance and ability to distinguish between positive and negative samples when predicting 1-year overall survival and recurrence-free survival. At the same time, although the prediction accuracy is slightly reduced in the prediction of 2-year overall survival and recurrence-free survival, it still remains at a high level, and the model performance and ability to distinguish between positive and negative samples are still good.

[0067] like Figure 4 As shown, the left and right figures respectively display the Kaplan-Meier curves under the Milan standard and the prediction standard proposed in this application. Figure 4 As can be seen, the overall survival curves of patients in both figures show a downward trend over time, indicating that the overall survival rate of patients gradually decreases over time. At the same time, it can be seen that compared with the survival analysis prediction based on the Milan standard, the overall survival rate of samples with an OS prediction of 1, based on the model proposed in this application, shows a more obvious downward trend. This indicates that the model proposed in this application has more accurate predictions, stronger ability to distinguish between positive and negative samples, and better model performance.

[0068] According to the method for constructing a predictive model for the prognosis of liver transplantation in liver cancer according to embodiments of this application, firstly, based on prior literature knowledge of influencing factors of liver transplantation in liver cancer, an unsupervised learning process is performed on a basic large language model to obtain an initial large language model; secondly, a dataset is obtained, and natural language conversion is performed on the dataset according to prompt word templates to train the initial large language model, resulting in a pre-trained large language model. The dataset includes clinical data from multiple liver transplant patients; finally, multidimensional information from typical cases among liver transplant patients is obtained as labeled fine-tuning data to fine-tune the pre-trained large language model, resulting in a predictive model. This predictive model is used to predict the prognosis of liver transplantation in liver cancer. Therefore, a predictive model for the prognosis of liver transplantation in liver cancer with expert-level evaluation capabilities can be constructed, thereby achieving accurate prediction of the prognosis of liver transplant patients with liver cancer.

[0069] Furthermore, embodiments of this application provide a method for predicting the prognosis of liver transplantation for liver cancer, comprising: obtaining clinical data of liver transplant patients; inputting the clinical data of liver transplant patients into a prediction model to obtain the prognostic prediction result of liver transplantation for liver cancer in the liver transplant patients, wherein the prediction model is based on the flowchart mentioned above. Figure 1 The predictive model for liver transplantation in liver cancer is described using a predictive model construction method.

[0070] The method for predicting the prognosis of liver transplantation in liver cancer according to embodiments of this application firstly involves unsupervised learning of a basic large language model based on prior literature knowledge of influencing factors of liver transplantation in liver cancer, resulting in an initial large language model. Secondly, a dataset is obtained, and natural language conversion is performed on the dataset according to prompt word templates to train the initial large language model, resulting in a pre-trained large language model. The dataset includes clinical data from multiple liver transplant patients. Finally, multidimensional information from typical cases among liver transplant patients is obtained as labeled fine-tuning data to fine-tune the pre-trained large language model, resulting in a prediction model. This prediction model is used to predict the prognosis of liver transplantation in liver cancer. Therefore, a prognosis prediction model for liver transplantation in liver cancer with expert-level evaluation capabilities can be constructed, thereby achieving accurate prediction of the prognosis of liver transplant patients with liver cancer.

[0071] Specific limitations regarding the predictive methods for liver transplantation prognosis in liver cancer can be found in the above-mentioned limitations on the predictive model construction method for liver transplantation prognosis, and will not be repeated here. Each module of the aforementioned predictive system for liver transplantation prognosis can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the computer device's memory as software, so that the processor can call and execute the corresponding operations of each module.

[0072] Figure 5 This is a structural block diagram of a liver cancer liver transplant prognosis prediction system according to an embodiment of this application. Figure 5 As shown, the liver cancer liver transplant prognosis prediction system according to an embodiment of this application includes: an acquisition module 510 and a prediction module 520, wherein:

[0073] Module 510 is used to obtain clinical data from liver transplant patients;

[0074] Prediction module 520 is used to input the clinical data of the liver transplant patient into the prediction model to obtain the prognostic prediction result of the liver transplant patient with liver cancer, wherein the prediction model is based on the flowchart mentioned above. Figure 1 The predictive model for liver transplantation in liver cancer is described using a predictive model construction method.

[0075] According to the liver transplant prognosis prediction system for liver cancer according to embodiments of this application, firstly, based on prior literature knowledge of influencing factors of liver cancer and liver transplantation, an unsupervised learning process is performed on a basic large language model to obtain an initial large language model; secondly, a dataset is obtained, and natural language conversion is performed on the dataset according to prompt word templates to train the initial large language model, resulting in a pre-trained large language model. The dataset includes clinical data from multiple liver transplant patients; finally, multidimensional information from typical cases among liver transplant patients is obtained as labeled fine-tuning data to fine-tune the pre-trained large language model, resulting in a prediction model. This prediction model is used to predict the prognosis of liver cancer and liver transplantation. Therefore, a liver cancer and liver transplantation prognosis prediction model with expert-level evaluation capabilities can be constructed, thereby achieving accurate prediction of the prognosis of liver cancer and liver transplant patients.

[0076] Specific limitations regarding the predictive system for liver transplantation prognosis in liver cancer can be found in the above-described limitations on the predictive model construction method for liver transplantation prognosis, and will not be repeated here. Each module of the aforementioned predictive system for liver transplantation prognosis can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the computer device's memory as software, so that the processor can call and execute the corresponding operations of each module.

[0077] The following is for reference. Figure 6 , Figure 6 A schematic diagram of a computer device structure suitable for implementing embodiments of this application is shown.

[0078] like Figure 6As shown, the computer system 1000 includes a central processing unit (CPU) 1001, which can perform various appropriate actions and processes based on programs stored in read-only memory (ROM) 1002 or programs loaded from storage section 1008 into random access memory (RAM) 1003. The RAM 1003 also stores various programs and data required for the system's operating instructions. The CPU 1001, ROM 1002, and RAM 1003 are interconnected via a bus 1004. An input / output (I / O) interface 1005 is also connected to the bus 1004.

[0079] The following components are connected to I / O interface 1005: an input section 1006 including a keyboard, mouse, etc.; an output section 1007 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and speakers, etc.; a storage section 1008 including a hard disk, etc.; and a communication section 1009 including a network interface card such as a LAN card, modem, etc. The communication section 1009 performs communication processing via a network such as the Internet. A drive 1010 is also connected to I / O interface 1005 as needed. A removable medium 1011, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on drive 1010 as needed so that computer programs read from it can be installed into storage section 1008 as needed.

[0080] Specifically, according to embodiments of this application, the flowchart above refers to... Figure 1 The described process can be implemented as a computer-readable storage medium. For example, embodiments of this application include a computer-readable storage medium comprising a computer program containing program code for performing the methods shown in the flowchart, such as performing: unsupervised learning of a basic large language model based on prior literature knowledge of influencing factors of liver transplantation for liver cancer to obtain an initial large language model; obtaining a dataset, performing natural language conversion on the dataset according to prompt word templates, training the initial large language model to obtain a pre-trained large language model, wherein the dataset includes clinical data from multiple liver transplant patients; obtaining multidimensional information of typical cases among liver transplant patients as labeled fine-tuning data, fine-tuning the pre-trained large language model to obtain a prediction model, wherein the prediction model is used for prognostic prediction of liver transplantation for liver cancer.

[0081] Specifically, according to embodiments of this application, the flowchart above refers to... Figure 1The described process can be implemented as a computer software program. For example, embodiments of this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowchart, such as performing: unsupervised learning of a basic large language model based on prior literature knowledge of influencing factors of liver cancer liver transplantation to obtain an initial large language model; obtaining a dataset, performing natural language conversion on the dataset according to prompt word templates, training the initial large language model to obtain a pre-trained large language model, wherein the dataset includes clinical data of multiple liver transplant patients; obtaining multidimensional information of typical cases among liver transplant patients as labeled fine-tuning data, fine-tuning the pre-trained large language model to obtain a prediction model, wherein the prediction model is used for prognostic prediction of liver cancer liver transplantation.

[0082] In such an embodiment, the computer program includes program code for performing the methods shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network via communication section 1009, and / or installed from removable media 1011. When the computer program is executed by central processing unit (CPU) 1001, it performs the functions defined in the system of this application.

[0083] It should be noted that the computer-readable medium shown in this application can be a computer-readable signal medium or a computer-readable storage medium, or any combination of the two. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this application, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this application, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media can also be any computer-readable medium other than computer-readable storage media, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wireless, wire, optical fiber, RF, etc., or any suitable combination thereof.

[0084] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operational instructions of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two connected blocks may actually be executed substantially in parallel, or they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified functions or operational instructions, or using a combination of dedicated hardware and computer instructions.

[0085] The units or modules described in the embodiments of this application can be implemented in software or hardware. The described units or modules can also be located in a processor. The names of these units or modules do not, in certain circumstances, constitute a limitation on the unit or module itself.

[0086] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0087] The above embodiments merely illustrate several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. A method for constructing a predictive model for the prognosis of liver transplantation in liver cancer, characterized in that, include: Based on prior literature knowledge of factors influencing liver transplantation in liver cancer, an unsupervised learning process is performed on the basic large language model to obtain an initial large language model. A dataset is obtained, and after performing natural language conversion on the dataset according to the prompt word template, the initial large language model is trained to obtain a pre-trained large language model. The dataset includes clinical data from multiple liver transplant patients. Multidimensional information of typical cases among liver transplant patients is obtained as labeled fine-tuning data. The pre-trained large language model is then fine-tuned to obtain a prediction model, which is used to predict the prognosis of liver transplantation for liver cancer.

2. The method for constructing a predictive model for liver transplantation prognosis in liver cancer according to claim 1, characterized in that, The prior literature knowledge based on the influencing factors of liver transplantation for liver cancer is used to perform unsupervised learning on the basic large language model to obtain an initial large language model, including: Obtain the basic large language model; Literature and guidelines on influencing factors of liver transplantation for liver cancer were collected. These literature and guidelines were used as prior knowledge to perform unsupervised learning on the basic large language model to obtain an initial large language model.

3. The method for constructing a predictive model for liver transplantation prognosis in liver cancer according to claim 1, characterized in that, The dataset is obtained, and after performing natural language conversion on the dataset according to the prompt word template, the initial large language model is trained to obtain a pre-trained large language model. The dataset includes clinical data from multiple liver transplant patients, including: Clinical data from multiple liver transplant patients were obtained, and a dataset was generated based on the clinical data from the multiple liver transplant patients. Obtain the training set from the dataset; Perform natural language conversion on the data in the training set based on the prompt word template; The initial large language model is trained using the training set after natural language conversion to obtain the pre-trained large language model.

4. The method for constructing a predictive model for liver transplantation prognosis in liver cancer according to claim 3, characterized in that, The prompt template is a five-segment template, which includes general information about the liver transplant patient, tumor characteristics, downstaging treatment, immune status, and nutritional assessment.

5. The method for constructing a predictive model for liver transplantation prognosis in liver cancer according to claim 1, characterized in that, Also includes: A validation set is obtained based on the dataset. The prediction model is hyperparameter tuned and model calibrated based on the validation set.

6. The method for constructing a predictive model for liver transplantation prognosis in liver cancer according to claim 1, characterized in that, Also includes: The test set is obtained based on the dataset. The prediction model is validated and its performance is evaluated based on the test set.

7. The method for constructing a predictive model for liver transplantation prognosis in liver cancer according to claim 1, characterized in that, The multidimensional information obtained from typical cases of liver transplant patients, as labeled fine-tuned data, includes: Typical cases from liver transplant patients were selected from the dataset; The clinical data of the typical cases were processed using natural language processing to obtain multidimensional information on the typical cases among the liver transplant patients.

8. A method for predicting the prognosis of liver transplantation in patients with liver cancer, characterized in that, include: Obtain clinical data from liver transplant patients; The clinical data of the liver transplant patients are input into the prediction model to obtain the prognostic prediction results of liver cancer transplantation in the liver transplant patients, wherein the prediction model is a prediction model obtained by the method for constructing a prediction model for liver cancer transplantation according to any one of claims 1-7.

9. A predictive system for the prognosis of liver transplantation in liver cancer, characterized in that, include: The acquisition module is used to obtain clinical data from liver transplant patients. The prediction module is used to input the clinical data of the liver transplant patient into the prediction model to obtain the prognostic prediction result of the liver transplant patient with liver cancer, wherein the prediction model is a prediction model obtained by the method for constructing a prediction model for the prognosis of liver transplantation with liver cancer according to any one of claims 1-7.

10. A computing device, characterized in that, The computing device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. The processor executes the program to implement the method for constructing a predictive model for the prognosis of liver cancer transplantation according to claims 1-7, or the method for predicting the prognosis of liver cancer transplantation according to claim 8.

Images