Unified workflow-based tumor multidimensional assessment and report generation system and method

Abstract

The application discloses a tumor multidimensional evaluation and report generation system and method based on a unified workflow, relates to the technical field of clinical auxiliary decision and medical workflow optimization, and comprises a unified workflow entrance module, a tumor type identification module, a special evaluation engine scheduling module, a special evaluation engine library module, an evaluation result collection module, an intelligent report generation module, a decision support module and a workflow state visualization module, wherein the modules are connected in series. The unified workflow entrance module provides unified data management; the special evaluation engine library module is internally provided with multiple scoring engines for different tumor types and different diagnosis and treatment scenes; and the intelligent report generation module automatically generates a complete evaluation report. The application can effectively solve the problem of fragmentation of the tumor multidimensional evaluation workflow, effectively realize report integration, and enhance the clinical decision support capability.

Description

Technical Field

[0001] This invention relates to the field of clinical decision support and medical workflow optimization technology, and in particular to a system and method for multidimensional tumor assessment and report generation based on a unified workflow. Background Technology

[0002] In clinical oncology work, tumor assessment and report writing are core components. Currently, different assessment tools are used for different tumor types. At the same time, a comprehensive assessment of the same tumor does not rely on a single scoring or staging mechanism, but rather combines multiple tools to make cross-judgments based on multiple dimensions such as lesion progression, imaging characteristics, biological behavior, and patient tolerance, ultimately leading to the development of a precise treatment plan. For example, TNM staging is used to assess the extent of tumor invasion, lymph node metastasis, and distant metastasis in prostate cancer, while PI-RADS scores combined with multiparametric MRI are used to determine the malignancy probability of lesions on imaging. For breast cancer, TNM staging is used to determine the overall progression of tumor size, lymph node metastasis, and distant metastasis, while BI-RADS scores are used to classify the malignancy risk of lesions on mammography / ultrasound / MRI and screen for suspicious lesions requiring biopsy. For renal cell carcinoma, TNM staging is used to assess the extent of tumor invasion and metastasis, RENAL scores are used to analyze the anatomical location, size, and relationship with the renal hilum vessels of the tumor, assess the difficulty of surgery and the feasibility of nephron preservation, and CCLS scores are used to predict whether the tumor is a clear cell subtype based on imaging characteristics, providing a reference for postoperative adjuvant therapy.

[0003] Manual calculation of complex scores is prone to errors, and differences in the understanding of scoring rules among different doctors further reduce the standardization of the assessment. In Chinese invention application number ZL202510886360.4, entitled "An Automated Method and System for Gastric Cancer TNM Staging Based on a Large Language Model," an automated method and system for gastric cancer TNM staging based on a large language model is provided. This includes acquiring a patient's gastric cancer-related medical dataset; preprocessing the medical dataset; inputting the preprocessed medical dataset into a trained large language model; extracting examination conclusions from the preprocessed medical dataset and concatenating the diagnostic conclusions into a long text, which is then input into the large language model for TNM staging classification; and outputting the patient's TNM staging result. This invention achieves fully automated processing from medical dataset acquisition, preprocessing, feature extraction to final TNM staging classification.

[0004] Although the aforementioned inventions have enabled automated systems for scoring tools, resolved errors that are prone to occur in manual scoring, and improved the standardization of scoring, each scoring tool currently has independent calculation rules and operating interfaces. If a multi-dimensional assessment of tumors is to be conducted, doctors need to frequently switch between multiple systems, repeatedly enter patient clinical data, resulting in a highly fragmented workflow and low efficiency. Furthermore, the scoring results need to be manually transcribed into the final comprehensive report, which is not only time-consuming but also prone to transcription errors, making it impossible to achieve intelligent correlation between assessment results and report content.

[0005] In summary, existing technologies lack a unified working platform to manage different tumor assessment tools, have weak decision support functions, and struggle to provide comprehensive treatment recommendations based on multi-dimensional assessment results, which seriously affects the quality and efficiency of clinical diagnosis and treatment. Therefore, there is an urgent need for a tumor multi-dimensional assessment and report generation system and method based on a unified workflow to solve the above-mentioned technical problems. Summary of the Invention

[0006] To integrate various types of tumor assessment tools, achieve one-time login, multi-dimensional assessment, and automatic generation of complete reports, effectively solving the problem of workflow fragmentation, this invention proposes the following technical solution.

[0007] First aspect: The system is a multi-dimensional tumor assessment and report generation system based on a unified workflow. The system includes a unified workflow entry module, a tumor type identification module, a dedicated assessment engine scheduling module, a dedicated assessment engine library module, an assessment result aggregation module, an intelligent report generation module, a decision support module, and a workflow status visualization module. The unified workflow entry module is used to provide a unified workflow startup interface, a patient clinical data input interface, and to achieve unified data management. The tumor type identification module is used to automatically identify tumor types based on input clinical data or to provide a tumor type selection interface; The dedicated assessment engine scheduling module is used to call one or more assessment engines in the dedicated assessment engine library module according to the tumor type. The dedicated assessment engine library module has multiple scoring engines built-in for different tumor types and different diagnosis and treatment scenarios. Each engine is used to perform professional tumor assessment on clinical data and output assessment results. The evaluation result aggregation module is used to receive the evaluation results output by each dedicated evaluation engine and process them in a unified format. The decision support module is used to generate treatment recommendations and follow-up plans based on the evaluation results; The intelligent report generation module is used to adapt personalized report templates to the evaluation results in a unified format, and add treatment suggestions and follow-up plans to generate a complete clinical evaluation report. The workflow status visualization module is used to display the evaluation progress and result status in real time.

[0008] Furthermore, the tumor type identification module has a built-in large model processing unit, which analyzes and processes the input clinical data based on the 69 types of cancer tumor description templates in the AJCC8 standard.

[0009] Furthermore, the tumor type identification module also has an input interface, which allows doctors to manually correct the tumor type for incomplete information or doubtful identification results, and the corrected data is fed back to this module.

[0010] Furthermore, each scoring engine in the dedicated assessment engine library module includes a natural language understanding unit, a data standardization unit, a missing data handling unit, an assessment result generation unit, and an assessment process traceability unit. The natural language understanding unit automatically extracts scoring parameters from clinical data, the data standardization unit performs standardization processing and boundary checks on the input data, the missing data handling unit provides intelligent prompts for missing fields and waits for supplementary data, the assessment result generation unit calculates scores based on the data of each scoring parameter and generates assessment results, and the assessment process traceability unit is used to record the scoring basis and intermediate judgment process of each engine.

[0011] Furthermore, the dedicated evaluation engine library module includes the RENAL scoring engine, PI-RADS scoring engine, BI-RADS scoring engine, CCLS scoring engine, UCLS scoring engine, and TNM phased engine.

[0012] Furthermore, the dedicated evaluation engine library module also has an extension interface for introducing new scoring engines. Furthermore, the intelligent report generation module includes a dynamic template adaptation unit and a report output unit. The dynamic template adaptation unit is used to select the corresponding report template according to the tumor type and assessment results, and the report output unit is used to output a complete clinical assessment report containing the scoring results, basis, and recommendations.

[0013] The second aspect: A method for multidimensional tumor assessment and report generation based on a unified workflow, using the system described in the first aspect of the present invention, includes the following steps: Step 1: Receive and manage the input clinical data through the unified workflow entry module; Step two: The tumor type identification module analyzes the clinical data and automatically identifies the tumor type or provides a tumor type selection interface for user confirmation. Step 3: The dedicated assessment engine scheduling module calls one or more dedicated assessment engines from the dedicated assessment engine library module based on the identified tumor type. Step four: The dedicated assessment engine being invoked performs a professional assessment of the clinical data and outputs the assessment results and assessment basis. Step 5: The evaluation result collection module receives one or more evaluation results and evaluation basis, processes them in a unified format, and then outputs them to the intelligent report generation module and the decision support module. Step 6: The decision support module generates treatment recommendations and follow-up plans based on the evaluation results in a unified format and outputs them to the intelligent report generation module; Step 7: The intelligent report generation module adapts a personalized report template based on the tumor type and assessment results, and integrates the assessment results, assessment basis, and treatment recommendations to generate a complete clinical assessment report; Step 8: The workflow status visualization module displays the progress of the entire process and the status of the final report in real time.

[0014] Third aspect An electronic device is provided, including at least one processor and a storage medium connected to the processor. The storage medium stores instructions that can be executed by the processor. When the instructions are executed by the processor, they realize the functions of each module of the tumor multidimensional assessment and report generation system based on a unified workflow as described in the first aspect of the present invention.

[0015] The present invention has the following advantages over the prior art: 1. This invention integrates multiple assessment tools for different tumor types and treatment scenarios through a dedicated assessment engine library module, enabling one-time login and multi-dimensional assessment. A unified workflow entry module provides doctors with a single workflow startup interface. All assessment tasks share the same set of basic patient data, avoiding repeated input, improving data management efficiency, and avoiding doctors switching between multiple systems, significantly improving clinical work efficiency. Compared with existing decentralized assessment systems, it shortens the assessment and report generation time and can effectively solve the problem of workflow fragmentation. 2. This invention automatically adapts and integrates evaluation results with report templates through the collaborative work of the evaluation result collection module and the intelligent report generation module. It eliminates the need for manual transcription, avoids transcription errors, and realizes a complete workflow loop from data input, evaluation calculation, result integration to report generation. This solves the problem of the disconnect between evaluation results and report content and can effectively achieve integrated report generation. 3. This invention generates targeted treatment suggestions based on multi-dimensional assessment results, providing doctors with comprehensive treatment direction references, improving the quality of diagnosis and treatment, making up for the shortcomings of weak decision support functions in existing technologies, and effectively enhancing clinical decision support capabilities. Attached Figure Description

[0016] Figure 1 This is a schematic diagram showing the connection relationship between the various modules of the system of the present invention; Figure 2This is a schematic diagram showing the connection relationship between the units of the dedicated evaluation engine library module in the system of the present invention; Figure 3 This is a schematic diagram of the steps of the generation method of the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Reference Figure 1 The tumor multidimensional assessment and report generation system based on a unified workflow includes a unified workflow entry module, a tumor type identification module, a dedicated assessment engine scheduling module, a dedicated assessment engine library module, an assessment result aggregation module, an intelligent report generation module, a decision support module, and a workflow status visualization module.

[0019] The modules are connected sequentially. The output of the unified workflow entry module is connected to the input of the tumor type identification module. The output of the tumor type identification module is connected to the input of the dedicated evaluation engine scheduling module. The output of the dedicated evaluation engine scheduling module is connected to the input of the dedicated evaluation engine library module. The output of the dedicated evaluation engine library module is connected to the input of the evaluation result collection module. The output of the evaluation result collection module is connected to the inputs of the intelligent report generation module and the decision support module, respectively. The output of the intelligent report generation module is connected to the input of the workflow status visualization module. The output of the decision support module is connected to the input of the intelligent report generation module.

[0020] The specific functions and implementation methods of each module are as follows: M1: A unified workflow entry module is built to create a single operation entry point, integrating basic patient data to lay a unified data foundation for subsequent evaluation work and solve the problem of scattered data from multiple systems in existing technologies. In addition to providing user login authentication, the unified workflow entry module also provides patient data management functions, creating a data profile for each patient, and providing clinical data upload functions. It can upload multiple different types of examination reports for a patient at one time, accepting image, PDF, and text files, and extracting text data from the above-mentioned file formats and passing it to the tumor type identification module.

[0021] M2: The tumor type identification module includes a text language processing unit and a large model processing unit. The text language processing unit standardizes the clinical data text, including text cleaning and denoising, medical terminology standardization, key information extraction, text segmentation and rule alignment, and standardized format conversion, outputting standardized JSON. The large model processing unit identifies the tumor type based on the standardized JSON and the built-in AJCC8 69 cancer feature template. The built-in large model processing unit can choose a closed-source medical-specific large model or an open-source general-purpose large model.

[0022] M3: The dedicated assessment engine scheduling module enables on-demand invocation of assessment engines based on the identified tumor type, avoiding resource waste and ensuring accurate matching between assessment tools and tumor types. The core of this module is a maintainable data table of tumor types and corresponding assessment engines. For example, kidney tumors can invoke the RENAL scoring engine and TNM staging engine, while prostate cancer can invoke the PI-RADS scoring engine and TNM staging engine.

[0023] M4: The dedicated evaluation engine library module is the core module of the entire system. Depending on the work requirements, it contains multiple evaluation engines. In this embodiment, the dedicated evaluation engine library includes the RENAL scoring engine, PI-RADS scoring engine, BI-RADS scoring engine, CCLS scoring engine, UCLS scoring engine, and TNM staging engine. RENAL scoring engine: specifically designed for scoring the anatomical complexity of kidney tumors, accurately calculating based on dimensions such as tumor size, location, and depth; PI-RADS scoring engine: An imaging assessment system for prostate cancer that provides standardized scoring based on MRI features; BI-RADS scoring engine: A imaging assessment and reporting system for breast cancer, providing a unified grading standard; CCLS / UCLS scoring engine: A dynamic scoring system based on branch selection that automatically matches the corresponding scoring rules according to the clinical pathway; TNM Installment Engine: As a basic assessment tool; In addition, the dedicated evaluation engine library module also has an extension interface for introducing new scoring engines.

[0024] refer to Figure 2Each scoring engine in the dedicated assessment engine library module includes a natural language understanding unit, a data standardization unit, a missing data handling unit, an assessment result generation unit, and an assessment process traceability unit. The natural language understanding unit automatically extracts scoring parameters from clinical data. The data standardization unit performs standardization processing and boundary checks on the input data. The missing data handling unit provides intelligent prompts for missing fields and waits for supplementary data. The assessment result generation unit calculates scores based on the data of each scoring parameter and generates assessment results. The assessment process traceability unit records the scoring basis and intermediate judgment process of each engine, ensuring the transparency and traceability of the assessment process.

[0025] Taking the RENAL rating engine as an example: (1) The scoring dimensions and score range of the built-in RENAL scoring mechanism in the scoring engine. R (tumor size): 1-3 points; ≤4cm = 1 point; 4cm and <7cm = 2 points; ≥7cm = 3 points E (Protrusion): 1-3 points; ≥50% protrusion = 1 point; <50% protrusion = 2 points; Completely ingrown = 3 points N (distance from the set system): 1-3 points; ≥7mm = 1 point; 4mm and <7mm = 2 points; ≤4mm = 3 points A (Anterior / posterior position): Classification identifier; A = anterior (ventral); P = posterior (dorsal); X = difficult to determine. L (Longitudinal position): 1-3 points; completely located at the pole = 1 point; mostly located at the pole = 2 points; more than 50% crosses the pole = 3 points. Weighting and total score calculation: Each dimension has equal weight. Total score = R + E + N + L (4-12 points). Dimension A serves as a location identifier and does not participate in the scoring but affects surgical planning.

[0026] (2) Natural Language Understanding Unit: Completes tasks such as medical named entity recognition (NER), medical terminology normalization, pronoun resolution and information purification. Extracts from the original text: maximum diameter and unit of tumor; classification of external protrusion; extraction of set system distance; classification of anterior and posterior positions; and classification of longitudinal positions.

[0027] (3) Data standardization unit, which completes the numerical standardization process, including verifying the numerical range, type conversion and boundary check.

[0028] (4) Missing data processing unit: If no relevant features are extracted from the original text, or the data exceeds the reasonable range, a pop-up window will be opened to prompt the user to manually enter the data. This unit provides two interaction modes. For discriminative evaluations such as TNM and RENAL, the system outputs the score, basis and missing item prompts in parallel. The user can supplement any missing information at any time to update the score. For link-based evaluations such as UCLS and CCLS, a serial workflow is adopted. The downstream options are dynamically determined according to the user's selected branch, and the process is gradually advanced until the final result is generated.

[0029] (5) Evaluation result generation unit: The extracted scoring parameters are compared with the scoring criteria to calculate the scoring results.

[0030] (6) Traceability of the evaluation process: Record the basis for judgment in detail, retain and show the original analysis process of the natural language understanding unit, the reasons for decision-making, and provide a transparent scoring logic chain.

[0031] M5: The evaluation results aggregation module summarizes the output results from multiple engines and performs standardization processing to eliminate format differences, providing a unified data format for subsequent report generation and decision support.

[0032] M6: The decision support module associates the assessment results with treatment recommendations based on preset rules, realizing the transformation of assessment results into clinical decisions. The preset rules are data tables that correspond assessment results to treatment recommendations one-to-one according to treatment guidelines, such as a graded recommendation mechanism based on RENAL scores. Low complexity (total score ≤ 6 points): Kidney-preserving surgery is recommended, as the technical risk is relatively low; Moderate complexity (total score 7-9): Feasibility of kidney-preserving surgery needs to be assessed, and it is recommended to perform the procedure by an experienced surgeon. High complexity (total score ≥10 points): Consider radical nephrectomy or multidisciplinary evaluation of treatment options; Authoritative source: AJCC Cancer Staging Manual, 8th Edition, published by the American Joint Committee on Cancer (AJCC).

[0033] M7: The intelligent report generation module automatically integrates multi-dimensional information to generate standardized reports through dynamic template adaptation, without manual intervention, thus avoiding transcription errors.

[0034] M8: The workflow status visualization module enables visualized management of the entire assessment process by tracking the assessment progress in real time, making it easier for doctors to control the pace of their work.

[0035] The present invention also discloses a method for multidimensional tumor assessment and report generation based on a unified workflow, which is implemented using the system disclosed in the above embodiments. refer to Figure 3 A method for multidimensional tumor assessment and report generation based on a unified workflow includes the following steps: Step 1: Doctors input patients' clinical data, including pathological diagnoses and imaging features, through a unified workflow entry module. This module manages the input data in a unified manner, ensuring that all assessment tasks share the same set of basic data and avoiding duplicate input. Step 2: Using the tumor type identification module, clinical data is analyzed and processed based on the cancer tumor description template in the AJCC8 standard to automatically identify tumor types; Step 3: The dedicated assessment engine scheduling module calls the corresponding dedicated assessment engine from the dedicated assessment engine library module based on the confirmed tumor type. Step 4: The dedicated assessment engine is invoked to start the assessment process. The built-in natural language understanding unit automatically extracts scoring parameters from the clinical description. The data standardization unit performs standardization processing and boundary checks. If there are missing fields, the missing data processing unit will provide intelligent prompts. The assessment is completed after the doctor supplements the data, and the assessment results and assessment basis are output. Step 5: The evaluation result collection module receives the evaluation results and evaluation basis, performs unified format processing, converts the non-standard format results output by different engines into a unified format, and then transmits them to the intelligent report generation module and the decision support module. Step 6: Based on the evaluation results in a unified format and combined with the scoring-decision mapping rules, the decision support module generates targeted diagnosis and treatment suggestions and follow-up plans, and transmits them to the intelligent report generation module. Step 7: The intelligent report generation module selects the corresponding personalized report template based on the tumor type and assessment results, and integrates the assessment results, assessment basis and treatment suggestions to generate a complete clinical assessment report containing the scoring results, basis and suggestions. Step 8: The workflow status visualization module displays the progress information and final report status of the entire assessment process in real time. Doctors can use this module to understand the progress of the assessment work in real time and obtain a complete clinical assessment report in a timely manner for clinical diagnosis and treatment.

[0036] The present invention also provides an electronic device, including at least one processor and a storage medium connected to the processor. The storage medium stores instructions that can be executed by the processor. When the instructions are executed by the processor, they realize the functions of each module of the tumor multidimensional assessment and report generation system based on a unified workflow as described in the above embodiments.

[0037] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A multi-dimensional tumor assessment and report generation system based on a unified workflow, characterized in that, The system includes a unified workflow entry module, a tumor type identification module, a dedicated assessment engine scheduling module, a dedicated assessment engine library module, an assessment result aggregation module, an intelligent report generation module, a decision support module, and a workflow status visualization module. The unified workflow entry module is used to provide a unified workflow startup interface, a patient clinical data input interface, and to achieve unified data management. The tumor type identification module is used to automatically identify tumor types based on input clinical data or to provide a tumor type selection interface; The dedicated assessment engine scheduling module is used to call one or more assessment engines in the dedicated assessment engine library module according to the tumor type. The dedicated assessment engine library module has multiple scoring engines built-in for different tumor types and different diagnosis and treatment scenarios. Each engine is used to perform professional tumor assessment on clinical data and output assessment results. The evaluation result aggregation module is used to receive the evaluation results output by each dedicated evaluation engine and process them in a unified format. The decision support module is used to generate treatment recommendations and follow-up plans based on the evaluation results; The intelligent report generation module is used to adapt personalized report templates to the evaluation results in a unified format, and add treatment suggestions and follow-up plans to generate a complete clinical evaluation report. The workflow status visualization module is used to display the evaluation progress and result status in real time.

2. The tumor multidimensional assessment and report generation system based on a unified workflow according to claim 1, characterized in that, The tumor type identification module has a built-in large model processing unit, which analyzes and processes the input clinical data based on the 69 types of cancer tumor description templates in the AJCC8 standard.

3. The tumor multidimensional assessment and report generation system based on a unified workflow according to claim 2, characterized in that, The tumor type identification module also has an input interface, which allows doctors to manually correct the tumor type for incomplete information or doubtful identification results. The corrected data is then fed back into this module.

4. The tumor multidimensional assessment and report generation system based on a unified workflow according to claim 1, characterized in that, Each scoring engine in the dedicated assessment engine library module includes a natural language understanding unit, a data standardization unit, a missing data handling unit, an assessment result generation unit, and an assessment process traceability unit. The natural language understanding unit automatically extracts scoring parameters from clinical data, the data standardization unit performs standardization processing and boundary checks on the input data, the missing data handling unit provides intelligent prompts for missing fields and waits for supplementary data, the assessment result generation unit calculates scores based on the data of each scoring parameter and generates assessment results, and the assessment process traceability unit is used to record the scoring basis and intermediate judgment process of each engine.

5. The tumor multidimensional assessment and report generation system based on a unified workflow according to claim 4, characterized in that, The dedicated evaluation engine library module includes the RENAL scoring engine, PI-RADS scoring engine, BI-RADS scoring engine, CCLS scoring engine, UCLS scoring engine, and TNM phased engine.

6. The tumor multidimensional assessment and report generation system based on a unified workflow according to claim 4, characterized in that, The dedicated evaluation engine library module also has an extension interface for introducing new scoring engines.

7. The tumor multidimensional assessment and report generation system based on a unified workflow according to claim 1, characterized in that, The intelligent report generation module includes a dynamic template adaptation unit and a report output unit. The dynamic template adaptation unit is used to select the corresponding report template according to the tumor type and assessment results. The report output unit is used to output a complete clinical assessment report containing the scoring results, basis, and recommendations.

8. A method for multi-dimensional tumor assessment and report generation based on a unified workflow, characterized in that, Using the system as described in any one of claims 1-7 includes the following steps: Step 1: Receive and manage the input clinical data through the unified workflow entry module; Step two: The tumor type identification module analyzes the clinical data and automatically identifies the tumor type or provides a tumor type selection interface for user confirmation. Step 3: The dedicated assessment engine scheduling module calls one or more dedicated assessment engines from the dedicated assessment engine library module based on the identified tumor type. Step four: The dedicated assessment engine being invoked performs a professional assessment of the clinical data and outputs the assessment results and assessment basis. Step 5: The evaluation result collection module receives one or more evaluation results and evaluation basis, processes them in a unified format, and then outputs them to the intelligent report generation module and the decision support module. Step 6: The decision support module generates treatment recommendations and follow-up plans based on the evaluation results in a unified format and outputs them to the intelligent report generation module; Step 7: The intelligent report generation module adapts a personalized report template based on the tumor type and assessment results, and integrates the assessment results, assessment basis, and treatment recommendations to generate a complete clinical assessment report; Step 8: The workflow status visualization module displays the progress of the entire process and the status of the final report in real time.

9. An electronic device comprising at least one processor and a storage medium connected to the processor, the storage medium storing instructions executable by the processor, characterized in that... When the instructions are executed by the processor, they implement the functions of each module of the tumor multidimensional assessment and report generation system based on a unified workflow as described in claim 1.

Images