Patient physical condition assessment method, device and equipment and storage medium

Abstract

The embodiment of the invention relates to a patient physical condition assessment method, device and equipment and a storage medium. The method comprises the following steps: according to patient case data, carrying out experience dependence evaluation on the physical condition of a patient to obtain an experience evaluation result; performing rule dependence evaluation on the physical condition of the patient according to the patient medical record data to obtain a rule evaluation result; and under the condition that at least one of the experience assessment result and the rule assessment result comprises a severe assessment result, carrying out risk assessment feedback on the physical condition of the patient. According to the technical scheme, high-efficiency and high-reliability automatic patient physical power condition evaluation independent of manual work can be achieved.

Description

Technical Field

[0001] This invention relates to the field of data analysis technology, and in particular to a method, apparatus, device, and storage medium for assessing a patient's physical condition. Background Technology

[0002] In clinical practice, particularly in oncology, accurate assessment of a patient's performance status is crucial for developing treatment plans, predicting prognosis, and selecting candidates for clinical trials. Currently, the ECOG-PS score (Eastern Cooperative Oncology Group Performance Status) is a widely adopted standardized assessment tool internationally. However, large-scale, automated extraction of PS scores from electronic health records remains a significant challenge, as the vast majority of relevant information is embedded in unstructured clinical texts, such as physician progress notes and nursing records.

[0003] Traditional methods involve manual summarization, where clinicians, nurses, or research coordinators manually assign PS scores to patients by reading large amounts of scattered clinical documents. While reliable, this approach is inefficient and ill-suited to the demands of the big data era. The rise of deep learning has brought a breakthrough; recurrent neural networks and convolutional neural networks can automatically learn feature representations from raw text, reducing reliance on manual feature engineering. However, it still requires large-scale, high-quality manual data annotation, and manual adjudication and verification are still necessary when scoring criteria from different sources contradict each other. Summary of the Invention

[0004] This invention provides a method, apparatus, device, and storage medium for assessing a patient's physical condition, with the aim of achieving highly efficient, highly reliable, and automated assessment of a patient's physical condition that does not rely on manual intervention.

[0005] In a first aspect, embodiments of the present invention provide a method for assessing a patient's physical condition, comprising:

[0006] Based on patient case data, an experience-dependent assessment of the patient's physical condition was conducted to obtain the experience assessment results.

[0007] Based on the patient's medical record data, a rule-dependent assessment of the patient's physical condition was performed to obtain the rule assessment results.

[0008] If at least one of the experience assessment results and the rule assessment results includes a critical illness assessment result, a risk assessment feedback is provided on the patient's physical condition.

[0009] Optionally, the step of performing an experience-dependent assessment of the patient's physical condition based on patient case data to obtain an experience assessment result includes:

[0010] Unstructured text data is obtained from the patient's medical record data;

[0011] The experience dependency evaluation is performed on the unstructured text data to obtain an experience-quantified score;

[0012] If the experience-based quantitative score reaches the critical illness scoring threshold, the critical illness assessment result is generated based on the experience-based quantitative score.

[0013] Optionally, the step of performing the experience dependency evaluation on the unstructured text data to obtain an experience quantification score includes:

[0014] The unstructured text data is input into the evaluation agent;

[0015] Obtain the experience quantification score output by the evaluation agent;

[0016] The evaluation agent is used to perform the experience dependency evaluation on the input unstructured text data based on preset evaluation knowledge, and output the experience quantification score.

[0017] Optionally, the step of performing a rule-dependent assessment of the patient's physical condition based on the patient's medical record data to obtain a rule assessment result includes:

[0018] According to preset judgment rules, judgment event data and cumulative event data are identified in the patient's medical record data;

[0019] If the existence of the determination event data is detected, the critical illness assessment result is generated based on the determination event data.

[0020] Optionally, after identifying judgment event data and cumulative event data in the patient's medical record data according to preset judgment rules, the method further includes:

[0021] The accumulated event data is subjected to quantitative scoring to obtain a rule-based quantitative score;

[0022] If the rule-quantified score reaches the cumulative score threshold, the critical illness assessment result is generated based on the cumulative event data.

[0023] Optionally, the risk assessment feedback on the patient's physical condition includes:

[0024] From the patient case data, obtain empirical evidence data associated with the empirical assessment results and rule evidence data associated with the rule assessment results;

[0025] If a conflict is found between the empirical evidence data and the rule-based evidence data, a severe illness risk assessment is performed on the empirical evidence data and the rule-based evidence data to obtain a severe illness assessment result, and the risk assessment feedback is performed based on the severe illness assessment result.

[0026] Optionally, the step of determining a severe illness risk assessment between the empirical evidence data and the rule-based evidence data when a conflict is found between them, and obtaining a severe illness assessment result, includes:

[0027] The experience evaluation results, the rule evaluation results, the experience evidence data, and the rule evidence data are input into the adjudication agent;

[0028] Obtain the critical illness decision result output by the decision-making agent;

[0029] The adjudication agent is used to make a severe illness risk adjudication on the input empirical evidence data and rule evidence data according to the preset adjudication logic, and output the severe illness adjudication result.

[0030] Secondly, embodiments of the present invention provide a patient physical condition assessment device, comprising:

[0031] The experience assessment module is used to conduct an experience-dependent assessment of the patient's physical condition based on the patient's case data, and obtain the experience assessment results.

[0032] The rule evaluation module is used to perform rule-dependent evaluation of the patient's physical condition based on the patient's medical record data, and obtain the rule evaluation result.

[0033] The assessment report module is used to provide risk assessment feedback on the patient's physical condition when at least one of the experience assessment results and the rule assessment results includes a critical illness assessment result.

[0034] Thirdly, embodiments of the present invention provide a patient physical condition assessment device, comprising:

[0035] One or more processors;

[0036] Memory, used to store one or more programs;

[0037] When the one or more programs are executed by the one or more processors, the one or more processors implement the patient physical condition assessment method provided in any embodiment of the present invention.

[0038] Fourthly, embodiments of the present invention provide a storage medium containing computer-executable instructions, which, when executed by a computer processor, are used to perform a patient physical condition assessment method as provided in any embodiment of the present invention.

[0039] This invention provides a method, apparatus, device, and storage medium for assessing patient physical condition. It performs an experience-dependent assessment of patient physical condition based on patient medical record data to obtain an experience assessment result, and performs a rule-dependent assessment of patient physical condition based on patient medical record data to obtain a rule assessment result. Thus, when at least one of the experience assessment result and the rule assessment result includes a critical illness assessment result, it provides risk assessment feedback on the patient's physical condition. This solves the problem of insufficient automatic analysis capability for different types of medical data in existing patient physical condition assessments, achieving a highly efficient, highly reliable, and automated patient physical condition assessment that does not rely on manual intervention. Attached Figure Description

[0040] Figure 1 A flowchart of a method for assessing a patient's physical condition provided in Embodiment 1 of the present invention;

[0041] Figure 2 This is a schematic diagram of the implementation architecture of a patient physical condition assessment method provided in Embodiment 1 of the present invention;

[0042] Figure 3 This is a flowchart illustrating a method for assessing a patient's physical condition according to Embodiment 1 of the present invention.

[0043] Figure 4 This is a schematic diagram of a rule dependency evaluation process provided in Embodiment 1 of the present invention;

[0044] Figure 5 This is a schematic diagram of a risk assessment feedback process provided in Embodiment 1 of the present invention;

[0045] Figure 6 This is a schematic diagram of a patient physical condition assessment device provided in Embodiment 2 of the present invention;

[0046] Figure 7 This is a schematic diagram of a patient physical condition assessment device provided in Embodiment 3 of the present invention. Detailed Implementation

[0047] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and not intended to limit it. Furthermore, it should be noted that, for ease of description, the accompanying drawings show only the parts relevant to the present invention, and not all of the structures.

[0048] Example 1

[0049] Figure 1 This is a flowchart of a method for assessing a patient's physical condition according to Embodiment 1 of the present invention. This embodiment is applicable to the automated assessment of a patient's physical condition. The method can be executed by a patient physical condition assessment device, which can be implemented by hardware and / or software and is generally integrated into an electronic device, such as a computer device. The method specifically includes:

[0050] Step 110: Based on the patient's case data, conduct an experience-dependent assessment of the patient's physical condition to obtain the experience assessment results.

[0051] Among these, patient medical record data can be data recording the patient's medical treatment. Experience-dependent assessment can be an operation that judges the patient's physical condition based on pre-learned clinical experience and experience in assessing the patient's physical condition. Experience assessment results can be the results obtained by judging the patient's physical condition based on pre-learned clinical experience and experience in assessing the patient's physical condition, and are used to describe the judged patient's physical condition.

[0052] In this embodiment, the core principle of experience-dependent assessment is to simulate the process of human analysis of patient medical record data and assessment of patient physical condition. This can be achieved based on artificial intelligence technologies such as machine learning, large models, or AI agents. By applying the experience data from human analysis of patient medical record data, along with the corresponding conclusions and principles of patient physical condition assessment, to the learning phase of artificial intelligence, the corresponding model or agent can simulate the human data analysis process, thereby acquiring the ability to perform experience-dependent assessment and obtaining the output experience assessment results based on the input patient medical record data.

[0053] Furthermore, the experience assessment results can include conclusions from experience-dependent assessments. These conclusions can include qualitative conclusions, such as a graded description of the patient's performance status, and quantitative conclusions, such as the patient's PS score. Optionally, the experience assessment results can also include the assessment basis for the experience assessment results, such as relevant clinical knowledge pre-learned by artificial intelligence; optionally, the experience assessment results can also include patient medical record data associated with the experience assessment results, such as original content in the patient medical record data that indicates the patient's performance status is severe. The above implementation allows the experience assessment results to be provided for human review or further research, overcoming the shortcomings of existing machine learning methods in terms of the lack of interpretability and unreliability in the model decision-making process.

[0054] In one optional implementation, the patient's physical condition is assessed based on experience-dependent evaluation of the patient's medical records to obtain an experience-dependent evaluation result. This may include: obtaining unstructured text data from the patient's medical records; performing an experience-dependent evaluation on the unstructured text data to obtain an experience-based quantitative score; and generating a critical illness assessment result based on the experience-based quantitative score if the experience-based quantitative score reaches a critical illness score threshold.

[0055] Unstructured text data can be natural language content without a specific format from patient medical records. Experience-based quantitative scores are numerical parameters used to evaluate a patient's physical condition through experience-dependent assessment; higher scores indicate poorer physical condition and more severe symptoms. The severity score threshold is the lowest experience-based quantitative score corresponding to a patient's poor physical condition that would warrant a diagnosis of severe illness. The severity assessment result describes the outcome, based on experience, where a patient's poor physical condition would warrant a diagnosis of severe illness.

[0056] Specifically, patient medical record data typically includes both structured and unstructured text data. Structured data usually refers to data with a specific format, such as disease diagnoses, physical examination indicators, and care levels in an EHR (Electronic Health Record). Unstructured text data typically refers to natural language text written or recorded by healthcare professionals without a specific format, and may include patient complaints, present medical history, and progress notes.

[0057] Because structured data has a specific format and its text and / or numerical values ​​are within a known range, the patient's physical condition described by the text and / or numerical values ​​can be determined through simple comparison. Therefore, to reduce computational burden, experience-dependent assessment can be performed only on unstructured text data. Through experience-dependent assessment, an experience-quantified score can be obtained from the unstructured text data. Based on the pre-determined mapping relationship in the experience-dependent assessment, a higher experience-quantified score is obtained when the patient's physical condition reflected by the unstructured text data is worse. Therefore, when the experience-quantified score reaches a preset critical illness score threshold, it indicates that the patient's physical condition is poor and can be judged as having entered a critical illness state. The critical illness assessment result can then be generated from the experience-quantified score in the experience assessment results.

[0058] In one optional implementation, the experience-based quantitative score can be the PS score; correspondingly, the severity score threshold can be set to 2. Therefore, when the PS score is greater than or equal to 2, a corresponding severity assessment result can be generated.

[0059] In one optional implementation, performing an experience-dependent evaluation on unstructured text data to obtain an experience-quantified score may include: inputting the unstructured text data into an evaluation agent; and obtaining the experience-quantified score output by the evaluation agent.

[0060] The evaluation agent is used to perform experience-dependent evaluation on the input unstructured text data based on pre-set evaluation knowledge, and output an experience-quantified score.

[0061] Specifically, an AI agent is an intelligent system capable of perceiving its environment and taking actions to achieve specific goals. It is typically based on an LLM (Large Language Model) and possesses autonomy, adaptability, and interactivity, enabling it to operate independently in complex and ever-changing environments. The assessment agent can be a pre-optimized agent that reads unstructured text data such as the patient's chief complaint, medical history, and progress notes. Based on pre-set assessment knowledge, including the agent's built-in clinical knowledge and / or patient performance evaluation experience such as the ECOG-PS scoring criteria, it analyzes the unstructured text data and outputs an experience-quantified score. Preferably, the output of the assessment agent can be a structured JSON object including the experience-quantified score, reasoning basis, and key citations.

[0062] Step 120: Based on the patient's medical record data, conduct a rule-dependent assessment of the patient's physical condition to obtain the rule assessment results.

[0063] Rule-dependent assessment can be an operation that judges a patient's physical condition based on pre-defined rules. The result of the rule assessment can be the result obtained by judging the patient's physical condition according to the pre-defined rules.

[0064] In this embodiment, the core principle of rule-dependent assessment is to perform feature matching on the content of patient medical record data according to explicit rules, and to conduct rule-dependent assessment based on patient medical record data that matches preset rules, thereby obtaining rule assessment results with higher reliability than experience-based assessment results. Optionally, rule-dependent assessment can be implemented through keyword matching. For example, the presence of "Level 1 Nursing Care" in the nursing record can indicate that the patient's physical condition is severe or corresponds to a preset higher PS score. Optionally, rule-dependent assessment can also be implemented by comparing the values ​​of key indicators with preset value ranges. For example, an albumin test result less than 25 g / L in laboratory test results can indicate that the patient's physical condition is severe or corresponds to a preset higher PS score.

[0065] Furthermore, the rule assessment results may include conclusions from rule-dependent assessments, which may include qualitative conclusions, such as a graded description of the patient's physical condition; or quantitative conclusions, such as the patient's PS score. Optionally, the rule assessment results may also include the assessment basis for the rule assessment results, such as pre-defined relevant rules; optionally, the rule assessment results may also include patient medical record data associated with the rule assessment results, such as text content containing specific keywords and indicators and values ​​that meet specific numerical conditions in the patient medical record data.

[0066] In one optional implementation, the patient's physical condition is assessed based on rule-dependent evaluation of the patient's medical record data to obtain the rule evaluation result. This may include: identifying judgment event data and cumulative event data in the patient's medical record data according to preset judgment rules; and generating a critical illness assessment result based on the judgment event data when the existence of judgment event data is identified.

[0067] The preset judgment rules can be pre-determined rules that compare patient medical record data with preset data content and identify judgment event data and cumulative event data based on the comparison results. Judgment event data can be data describing a patient's physical condition that is significantly poor, and which can be used to determine whether the patient's condition has entered the critical stage through rule-dependent assessment. Cumulative event data can be data describing a patient's physical condition that cannot be determined to have entered the critical stage through rule-dependent assessment.

[0068] Specifically, the preset judgment rules may include pre-defined keywords and / or key indicator value ranges. When keywords and / or key indicators within the specified value ranges appear in patient medical record data, the preset judgment rules can be used to identify whether the patient medical record data belongs to judgment event data or cumulative event data. It should be noted that judgment event data and cumulative event data may include structured data from patient medical records and unstructured text data that can be analyzed using rule-dependent evaluation.

[0069] Decision event data is typically pre-defined as patient medical record data corresponding to various high-risk events that can be analyzed using rule-dependent assessment, such as keywords like "transfer to ICU," "invasive ventilation," and "critical condition notification." When decision event data is identified in a patient's medical record, it indicates that the patient's physical condition is poor and the illness has entered a severe stage, thus allowing the generation of a critical condition assessment result.

[0070] In an optional implementation, after identifying judgment event data and cumulative event data in the patient's medical record data according to preset judgment rules, the process may further include: performing quantitative scoring on the cumulative event data to obtain a rule-based quantitative score; and generating a critical illness assessment result based on the cumulative event data when the rule-based quantitative score reaches the cumulative score threshold.

[0071] The quantitative scoring process involves determining a rule-based quantitative score for the patient's physical condition based on the data from each cumulative event. The rule-based quantitative score is a numerical parameter used to evaluate the patient's physical condition through rule-dependent assessment; a higher score indicates a worse physical condition and more severe symptoms. The cumulative score threshold is the lowest rule-based quantitative score that corresponds to a patient's poor physical condition being deemed to have progressed to a severe stage of illness.

[0072] Specifically, cumulative event data includes patient medical record data corresponding to various non-high-risk events that can be analyzed using rule-dependent assessment. When cumulative event data is identified in patient medical record data, it can be quantitatively scored to obtain corresponding rule-based quantitative scores.

[0073] It should be noted that the cumulative event data may contain multiple parts describing the patient's physical condition. Each part of the cumulative event data can be quantified and scored separately, and the final result is a rule-based quantitative score by combining the quantitative scores of each part. In one optional implementation, the rule-based quantitative score can be the sum of the PS scores obtained from each part of the cumulative event data; correspondingly, the cumulative score threshold can be 5. Therefore, when the rule-based quantitative score reaches the cumulative score threshold, it indicates that the patient's physical condition is poor and the disease has entered a severe stage, thus a severe illness assessment result can be generated accordingly.

[0074] Step 130: Provide a risk assessment feedback on the patient's physical condition, provided that at least one of the experience assessment results and rule assessment results includes a critical illness assessment result.

[0075] Among these, the critical condition assessment result can indicate that the patient's physical condition is poor and can be classified as a critically ill patient. Risk assessment feedback can be an operation that combines experience-based assessment results and rule-based assessment results to judge and provide feedback on the patient's physical condition.

[0076] Specifically, if the experience-based or rule-based assessment results include a critical condition assessment, it indicates that, based on experience and / or rules, the patient's physical condition is deemed poor, posing a certain risk and requiring the attention of medical staff. Therefore, a risk assessment of the patient's physical condition is necessary in this situation.

[0077] In this embodiment, the risk assessment feedback includes a comprehensive judgment of the patient's physical condition based on both experience-based and rule-based assessment results. If there are conflicting results between the experience-based and rule-based assessments—for example, the experience-based assessment may not include a severe illness assessment result, indicating a good physical condition, while the rule-based assessment does include a severe illness assessment result—the conflict can be resolved using artificial intelligence methods. This can be achieved by re-querying relevant patient case data and reviewing both the experience-based and rule-based assessment results to arrive at a final conclusion. If the experience-based and rule-based assessment results are consistent, both indicating a severe illness assessment result, the final conclusion can be obtained by combining both results. Finally, the risk assessment feedback can be used to inform the user of the patient's physical condition assessment results.

[0078] In one optional implementation, risk assessment feedback on the patient's physical condition may include: obtaining empirical evidence data associated with the empirical assessment results and rule evidence data associated with the rule assessment results from the patient's case data; if it is determined that there is a conflict between the empirical evidence data and the rule evidence data, making a severe illness risk determination on the empirical evidence data and the rule evidence data to obtain a severe illness determination result, and making risk assessment feedback based on the severe illness determination result.

[0079] Empirical evidence data can be patient case data upon which empirical assessment results are based, while rule-based evidence data can be patient case data upon which rule-based assessment results are based. Conflict in evidence data occurs when there is a significant difference between the patient's physical condition described by the empirical evidence data and the rule-based evidence data. Severity risk assessment can be an operation that re-evaluates whether a patient's physical condition falls within the severe stage of their illness based on both empirical and rule-based evidence data. The severity assessment result can be the patient's physical condition determined through the severity risk assessment.

[0080] Specifically, when it is determined that a risk assessment of the patient's physical condition is necessary, the relevant raw data can be obtained from the patient's medical records, including empirical evidence and rule-based evidence. If there is a conflict between the empirical and rule-based evidence—for example, the empirical evidence describes the patient's physical condition as acceptable, while the rule-based evidence contains judgment event data—a severe illness risk assessment can be made by combining the empirical and rule-based evidence. This involves comprehensively considering the specific content of all empirical and rule-based evidence to reassess the patient's physical condition, obtaining a severe illness assessment result, and then conducting the corresponding risk assessment feedback.

[0081] In an optional implementation, if it is determined that there is a conflict between empirical evidence data and rule-based evidence data, a severe illness risk assessment is performed on the empirical evidence data and rule-based evidence data to obtain a severe illness assessment result. This may include: inputting the empirical assessment result, rule assessment result, empirical evidence data and rule-based evidence data into the assessment agent; and obtaining the severe illness assessment result output by the assessment agent.

[0082] The adjudication agent is used to make a severe illness risk assessment based on the input empirical evidence data and rule evidence data according to the preset adjudication logic, and output the severe illness assessment result.

[0083] Specifically, the adjudicating agent can be a specially optimized agent that can simulate a senior clinical expert, identify and adjudicate the conflict between empirical evidence data and rule-based evidence data based on pre-set adjudication logic, and output accurate adjudication results for severe cases.

[0084] In an optional implementation, the pre-defined decision logic may prioritize rule evaluation results and rule evidence data over experience evaluation results and experience evidence data.

[0085] In an optional implementation, after obtaining empirical evidence data associated with the empirical assessment results and rule evidence data associated with the rule assessment results from the patient case data, the method may further include: generating a severe case acceptance result based on the empirical assessment results, rule assessment results, empirical evidence data, and rule evidence data, in the case that it is determined that there is no evidence data conflict between the empirical evidence data and the rule evidence data, so as to conduct risk assessment feedback based on the severe case acceptance result.

[0086] In this context, the acceptance of a severe case assessment result can be based on the patient's physical condition determined by the combined description of empirical assessment results, rule-based assessment results, empirical evidence data, and rule-based evidence data. Specifically, if there is no conflict between the empirical evidence data and the rule-based evidence data, indicating that the patient's physical condition described by the empirical assessment results, rule-based assessment results, empirical evidence data, and rule-based evidence data is consistent or similar, then a severe case assessment result can be directly generated for corresponding risk assessment feedback.

[0087] In one optional implementation, the severe illness ruling and the severe illness acceptance result may include the final PS score, detailed reasoning for the judgment, and the confidence level of the final PS score. Further optionally, the severe illness ruling and the severe illness acceptance result may be structured JSON objects. Correspondingly optionally, the risk assessment feedback based on the severe illness ruling or the severe illness acceptance result may be to directly output the severe illness ruling or the severe illness acceptance result for the user to use for further analysis.

[0088] For example, Figure 2This is a schematic diagram illustrating the implementation architecture of a patient physical condition assessment method provided in Embodiment 1 of the present invention. Figure 3 This is a flowchart illustrating a method for assessing a patient's physical condition according to Embodiment 1 of the present invention. (Combined with...) Figure 2 and 3 As shown, the patient performance status assessment method can be analyzed based on the patient's EHR data as patient case data. On one hand, Phase One is achieved through the preliminary assessment AI-Agent module, which performs an experience-dependent assessment of the patient's performance status to obtain the experience assessment result. This phase aims to extract patient functional status information from unstructured clinical text. By calling a specially optimized preliminary assessment AI-Agent, the system reads the patient's chief complaint, present medical history, and progress notes, analyzes them based on built-in clinical knowledge such as the ECOG-PS scoring standard, and outputs a structured JSON object containing the preliminary score, reasoning basis, and key citations as the experience assessment result. If the preliminary score reaches the preset severity criteria, such as a PS score ≥ 2, this result is taken as the severity assessment result and can serve as one of the clues to initiate Phase Three.

[0089] On the other hand, Phase Two is implemented through a rules engine module, which performs rule-dependent assessments of the patient's physical condition to obtain the rule assessment results. This phase runs in parallel with Phase One, focusing on identifying objective clues of severe illness from the patient's full EHR data. For example, Figure 4 This is a schematic diagram of a rule dependency evaluation process provided in Embodiment 1 of the present invention. Figure 4 As shown, the rule engine performs high-speed scanning of structured and unstructured text, and executes a two-layer judgment logic based on configurable preset judgment rules, the "List of Critical Case Clues." This includes a direct judgment layer, used to detect the existence of high-risk events that could lead to veto, identifying and judging event data such as ICU transfer, invasive ventilation, and critical condition notification; and a risk accumulation layer, used to quantify and score accumulated event data that did not match the direct rules, generating a rule-based quantitative score, the "Objective Risk Total Score." If a case matches any direct judgment rule, or if the Objective Risk Total Score exceeds a set threshold (≥5 points), an objective critical condition is determined, serving as a critical condition assessment result and triggering the Phase Three processing flow.

[0090] Furthermore, through the intelligent triage and trigger module, based on the aforementioned triggering conditions, Phase Three is initiated. If the experience-based or rule-based assessment results include a severe illness assessment, a risk assessment of the patient's physical condition is provided. This phase, as the core, is initiated after severe illness clues are identified in Phase One or Phase Two. All suspected severe cases undergo expert review to ensure the accuracy and reliability of the judgment results. For example, Figure 5 This is a schematic diagram of a risk assessment feedback process provided in Embodiment 1 of the present invention. Figure 5As shown, the process can specifically include a judgment material packaging step, which automatically integrates the subjective assessment report (Phase 1) based on experience evaluation results, the objective judgment results (Phase 2) based on rule evaluation results, and relevant original medical records included in the experience evidence data and rule evidence data to generate a "final judgment material package." Further, a deep reasoning and adjudication step is executed, passing the material package to an adjudication AI-Agent with complex clinical reasoning capabilities. This adjudication AI-Agent can simulate senior clinical experts based on pre-set adjudication logic, identifying and adjudicating potential conflicts between subjective and objective information based on pre-set logic such as "objective evidence takes precedence over subjective description," and outputting a precise final PS score. A further interpretable conclusion output step is executed, generating a structured JSON object through the adjudication AI-Agent, containing the final PS score, detailed judgment reasons, and confidence score, for risk assessment feedback. This result has high interpretability and reliability, can be directly adopted in most scenarios, and significantly reduces reliance on backend manual review.

[0091] In a specific application scenario, the goal is to determine a patient's critical condition and PS score. The patient's medical record data includes unstructured text from the doctor's progress notes ("patient is weak, limited in activity, requires assistance to walk, bedridden most of the time, and only partially able to care for themselves") and structured data from nursing records ("Level 1 nursing care"). Through the initial subjective assessment in Phase 1, an AI-Agent analyzes the progress notes to identify key descriptions such as "bedridden most of the time" and "only partially able to care for themselves," outputting an initial PS score of 3 based on the ECOG-PS standard. Since PS ≥ 2, Phase 3 processing can be triggered. Through objective clue judgment in Phase 2, a rule engine scans the entire data to identify the "Level 1 nursing care" clue, awarding 1 point in the risk accumulation layer. Assuming no other high-risk clues, although the total objective score is not high, Phase 1 has provided a critical condition trigger signal. Therefore, in the comprehensive review and final judgment in Phase 3, the assessment results of Phase 1 trigger this phase. The AI-Agent's "PS 3" assessment report and the rule engine's objective clue of "Level 1 Care" are packaged and passed to the adjudication AI-Agent. The adjudication AI-Agent performs a comprehensive analysis and finds that the subjective description of "primarily bedridden" is highly consistent with the objective clue of "Level 1 Care," with no conflict. Therefore, the final output is a PS score of 3 for the patient, classifying them as a critically ill patient. The adjudication AI-Agent's rationale is: "The preliminary subjective assessment of a PS score of 3 is consistent with the objective clue of 'Level 1 Care,' with no conflict; the preliminary assessment conclusion is accepted." The entire process is fully automated, accurately handling routine cases with consistent evidence.

[0092] In another specific application scenario, the goal was also to determine a patient's critical condition and PS score. The patient's medical record data included unstructured text written by the outpatient physician in the record: "Patient's general condition is good, ECOG 1 point"; structured data from the same day's laboratory test results: "Albumin <25g / L"; and structured data from the doctor's orders: "High-flow oxygen therapy". Through the initial subjective assessment in Phase One, an AI-Agent was used to analyze the outpatient records, outputting a preliminary assessment conclusion of PS 1. Since PS < 2, this result itself did not trigger a critical condition clue. In the objective clue determination in Phase Two, the rule engine scanned all data in parallel and successfully matched two high-risk clues: "Albumin < 25g / L" (1 point for risk accumulation layer) and "High-flow oxygen therapy" (2 points for risk accumulation layer); thus, the total objective risk score accumulated to 3 points. More importantly, since "High-flow oxygen therapy" was configured as one of the rules for directly determining a critical condition, the engine determined that the case had an objective critical condition. This determination became a strong clue triggering Phase Three. Therefore, in Phase Three, the objective severity assessment result from Phase Two triggered this phase. The preliminary assessment AI-Agent's "PS 1 point" report and the rule engine's "Objective Severe Illness (High-Flow Oxygen, Low Protein)" assessment report are packaged and passed to the adjudication AI-Agent. During analysis, the adjudication AI-Agent immediately identifies a serious logical conflict between the subjective assessment "PS 1 point, condition acceptable" and the objective indicator "requires high-flow oxygen." Based on its built-in advanced principle of "objective evidence takes precedence over subjective description," the adjudication AI-Agent performs deep reasoning that "a patient requiring high-flow oxygen has severely impaired respiratory function and could not possibly be in a state with a PS score of 1." Therefore, the preliminary assessment needs to be revised upwards, outputting the patient's PS score as 3. The AI-Agent's assessment was based on the following reasoning: "Comprehensive analysis initially yielded a PS score of 1, but this is significantly inconsistent with the objective fact that the patient requires high-flow oxygen therapy. Based on the principle of prioritizing objective evidence and referring to the ECOG scoring criteria, the patient has at least lost the ability to work and is severely limited in activity; therefore, the final assessment is a PS score of 3." This application scenario demonstrates the intelligence and robustness of the patient physical condition assessment method provided in Embodiment 1 of this invention in handling complex conflicting problems, avoiding the misjudgments of existing technologies.

[0093] The technical solution of this embodiment performs an experience-dependent assessment of the patient's physical condition based on patient case data to obtain an experience assessment result, and performs a rule-dependent assessment of the patient's physical condition based on patient medical record data to obtain a rule assessment result. Thus, when at least one of the experience assessment result and the rule assessment result includes a critical illness assessment result, a risk assessment feedback is provided for the patient's physical condition. This solves the problem of insufficient automatic analysis capability for different types of medical data in existing patient physical condition assessments, and achieves a highly efficient, highly reliable, and automated patient physical condition assessment that does not rely on manual intervention.

[0094] Example 2

[0095] Figure 6 This is a schematic diagram of the structure of a patient physical condition assessment device provided in Embodiment 2 of the present invention, as shown below. Figure 6 As shown, the patient physical condition assessment device includes: an experience assessment module 610, a rule assessment module 620, and an assessment report module 630, wherein...

[0096] The experience assessment module 610 is used to conduct an experience-dependent assessment of the patient's physical condition based on the patient's case data, and obtain the experience assessment results.

[0097] The rule evaluation module 620 is used to perform rule-dependent evaluation on the patient's physical condition based on the patient's medical record data, and obtain the rule evaluation result.

[0098] The assessment report module 630 is used to provide risk assessment feedback on the patient's physical condition when at least one of the experience assessment results and the rule assessment results includes a critical illness assessment result.

[0099] The technical solution of this embodiment performs an experience-dependent assessment of the patient's physical condition based on patient case data to obtain an experience assessment result, and performs a rule-dependent assessment of the patient's physical condition based on patient medical record data to obtain a rule assessment result. Thus, when at least one of the experience assessment result and the rule assessment result includes a critical illness assessment result, a risk assessment feedback is provided for the patient's physical condition. This solves the problem of insufficient automatic analysis capability for different types of medical data in existing patient physical condition assessments, and achieves a highly efficient, highly reliable, and automated patient physical condition assessment that does not rely on manual intervention.

[0100] Optionally, the experience assessment module 610 may include: a text acquisition unit for acquiring unstructured text data from the patient's medical record data; an experience quantification unit for performing the experience-dependent assessment on the unstructured text data to obtain an experience quantification score; and an experience threshold unit for generating the critical illness assessment result based on the experience quantification score when it is determined that the experience quantification score reaches a critical illness score threshold.

[0101] Optionally, the experience quantification unit can be specifically used to: input the unstructured text data into the evaluation agent; obtain the experience quantification score output by the evaluation agent; wherein the evaluation agent is used to perform experience dependency evaluation on the input unstructured text data according to preset evaluation knowledge, and output the experience quantification score.

[0102] Optionally, the rule evaluation module 620 may include: a rule recognition unit, used to identify judgment event data and cumulative event data in the patient medical record data according to preset judgment rules; and a judgment event unit, used to generate the critical illness assessment result based on the judgment event data when the existence of the judgment event data is identified.

[0103] Optionally, the rule evaluation module 620 may further include: a rule quantification unit, used to perform quantification scoring on the accumulated event data to obtain a rule quantification score; and a rule threshold unit, used to generate the critical illness assessment result based on the accumulated event data when it is determined that the rule quantification score has reached a cumulative score threshold.

[0104] Optionally, the assessment report module 630 may include: an evidence acquisition unit, used to acquire empirical evidence data associated with the empirical assessment result and rule evidence data associated with the rule assessment result from the patient case data; and a conflict resolution unit, used to make a severe illness risk resolution between the empirical evidence data and the rule evidence data when it is determined that there is an evidence data conflict between the empirical evidence data and the rule evidence data, to obtain a severe illness resolution result, so as to provide the risk assessment feedback based on the severe illness resolution result.

[0105] Optionally, the conflict resolution unit can be specifically used to: input the experience assessment result, the rule assessment result, the experience evidence data, and the rule evidence data into the resolution agent; obtain the severe illness resolution result output by the resolution agent; wherein, the resolution agent is used to make a severe illness risk resolution on the input experience evidence data and the rule evidence data according to a preset resolution logic, and output the severe illness resolution result.

[0106] The patient physical condition assessment device provided in the embodiments of the present invention can execute the patient physical condition assessment method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the method.

[0107] Example 3

[0108] Figure 7 This is a schematic diagram of the structure of a patient physical condition assessment device provided in Embodiment 3 of the present invention, as shown below. Figure 7As shown, the patient physical condition assessment device includes a processor 510, a memory 520, an input device 530, and an output device 540; the number of processors 510 in the patient physical condition assessment device can be one or more. Figure 7 Taking a processor 510 as an example; the processor 510, memory 520, input device 530, and output device 540 in the patient physical condition assessment device can be connected via a bus or other means. Figure 7 Taking the example of a connection between China and Israel via a bus.

[0109] The memory 520, as a computer-readable storage medium, can be used to store software programs, computer-executable programs, and modules, such as the program instructions / modules corresponding to the patient physical condition assessment method in this embodiment of the invention (e.g., the experience assessment module 610, rule assessment module 620, and assessment report module 630 in the patient physical condition assessment device). The processor 510 executes various functional applications and data processing of the patient physical condition assessment device by running the software programs, instructions, and modules stored in the memory 520, thereby realizing the aforementioned patient physical condition assessment method.

[0110] The memory 520 may primarily include a program storage area and a data storage area. The program storage area may store the operating system and at least one application program required for a given function; the data storage area may store data created based on terminal usage. Furthermore, the memory 520 may include high-speed random access memory and non-volatile memory, such as at least one disk storage device, flash memory, or other non-volatile solid-state storage device. In some instances, the memory 520 may further include memory remotely configured relative to the processor 510, which can be connected to the patient's physical condition assessment device via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0111] Input device 530 can be used to receive input digital or character information, and to generate key signal inputs related to user settings and function control of the patient's physical condition assessment device. Output device 540 may include a display device such as a screen.

[0112] Example 4

[0113] Embodiment 4 of the present invention also provides a storage medium containing computer-executable instructions, which, when executed by a computer processor, are used to perform a method for assessing a patient's physical condition, including:

[0114] Based on patient case data, an experience-dependent assessment of the patient's physical condition was conducted to obtain the experience assessment results.

[0115] Based on the patient's medical record data, a rule-dependent assessment of the patient's physical condition was performed to obtain the rule assessment results.

[0116] If at least one of the experience assessment results and the rule assessment results includes a critical illness assessment result, a risk assessment feedback is provided on the patient's physical condition.

[0117] Of course, the computer-executable instructions provided in the embodiments of the present invention are not limited to the method operations described above, but can also perform related operations in the patient physical condition assessment method provided in any embodiment of the present invention.

[0118] Based on the above description of the implementation methods, those skilled in the art can clearly understand that the present invention can be implemented using software and necessary general-purpose hardware, and of course, it can also be implemented using hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk, or optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0119] It is worth noting that in the embodiments of the above-mentioned patient physical condition assessment device, the various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be realized; in addition, the specific names of each functional unit are only for easy differentiation and are not used to limit the scope of protection of the present invention.

[0120] Although the present invention has been described in detail above with general descriptions, specific embodiments, and experiments, modifications or improvements can be made to it, which will be obvious to those skilled in the art. Therefore, all such modifications or improvements made without departing from the spirit of the present invention fall within the scope of protection claimed by the present invention.

Claims

1. A method for assessing a patient's physical condition, characterized in that, include: Based on patient case data, an experience-dependent assessment of the patient's physical condition was conducted to obtain the experience assessment results. Based on the patient's medical record data, a rule-dependent assessment of the patient's physical condition was performed to obtain the rule assessment results. If at least one of the experience assessment results and the rule assessment results includes a critical illness assessment result, a risk assessment feedback is provided on the patient's physical condition.

2. The method according to claim 1, characterized in that, The process of conducting an experience-dependent assessment of a patient's physical condition based on patient case data to obtain the experience assessment results includes: Unstructured text data is obtained from the patient's medical record data; The experience dependency evaluation is performed on the unstructured text data to obtain an experience-quantified score; If the experience-based quantitative score reaches the critical illness scoring threshold, the critical illness assessment result is generated based on the experience-based quantitative score.

3. The method according to claim 2, characterized in that, The empirical dependency evaluation of the unstructured text data, to obtain an empirical quantification score, includes: The unstructured text data is input into the evaluation agent; Obtain the experience quantification score output by the evaluation agent; The evaluation agent is used to perform the experience dependency evaluation on the input unstructured text data based on preset evaluation knowledge, and output the experience quantification score.

4. The method according to claim 1, characterized in that, The step of performing a rule-dependent assessment of the patient's physical condition based on the patient's medical record data to obtain the rule assessment results includes: According to preset judgment rules, judgment event data and cumulative event data are identified in the patient's medical record data; If the existence of the determination event data is detected, the critical illness assessment result is generated based on the determination event data.

5. The method according to claim 4, characterized in that, After identifying judgment event data and cumulative event data in the patient's medical record data according to preset judgment rules, the method further includes: The accumulated event data is subjected to quantitative scoring to obtain a rule-based quantitative score; If the rule-quantified score reaches the cumulative score threshold, the critical illness assessment result is generated based on the cumulative event data.

6. The method according to claim 1, characterized in that, The risk assessment feedback on the patient's physical condition includes: From the patient case data, obtain empirical evidence data associated with the empirical assessment results and rule evidence data associated with the rule assessment results; If a conflict is found between the empirical evidence data and the rule-based evidence data, a severe illness risk assessment is performed on the empirical evidence data and the rule-based evidence data to obtain a severe illness assessment result, and the risk assessment feedback is performed based on the severe illness assessment result.

7. The method according to claim 6, characterized in that, The step of determining a conflict between the empirical evidence data and the rule-based evidence data, and then making a severe illness risk assessment on the empirical evidence data and the rule-based evidence data to obtain a severe illness assessment result, includes: The experience evaluation results, the rule evaluation results, the experience evidence data, and the rule evidence data are input into the adjudication agent; Obtain the critical illness decision result output by the decision-making agent; The adjudication agent is used to make a severe illness risk adjudication on the input empirical evidence data and rule evidence data according to the preset adjudication logic, and output the severe illness adjudication result.

8. A device for assessing a patient's physical condition, characterized in that, include: The experience assessment module is used to conduct an experience-dependent assessment of the patient's physical condition based on the patient's case data, and obtain the experience assessment results. The rule evaluation module is used to perform rule-dependent evaluation of the patient's physical condition based on the patient's medical record data, and obtain the rule evaluation result. The assessment report module is used to provide risk assessment feedback on the patient's physical condition when at least one of the experience assessment results and the rule assessment results includes a critical illness assessment result.

9. A device for assessing a patient's physical condition, characterized in that, include: One or more processors; Memory, used to store one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the patient physical condition assessment method as described in any one of claims 1-7.

10. A storage medium containing computer-executable instructions, characterized in that, The computer-executable instructions, when executed by a computer processor, are used to perform the patient physical condition assessment method as described in any one of claims 1-7.

Images