Medical decision-making assistance system and method

Abstract

A medical decision-making assistance system stores a natural language processing model, a diagnosis and treatment recording database, and an intelligent decision-making module implemented by using the machine learning technology. The medical decision-making assistance system is used to generate a diagnosis data and a voice data related to the current patient during the diagnosis process of a current patient (including the patient himself / herself or his / her accompanying patient or his / her condition narrator), select a target historical diagnosis and treatment record corresponding to the current patient from the diagnosis and treatment record database, input the speech data into the natural language processing model to obtain the oral text data, and input the diagnosis data, the oral text data, and the target history diagnosis and treatment record into the intelligent decision-making module to obtain a decision-making assistance result, and to output the decision-making assistance result, wherein the decision-making assistance result indicates the medical reference information related to the symptoms of the current patient.

Description

[Technical Field]

[0001] This invention relates to an auxiliary system, and more particularly to a medical decision support system suitable for application in medical procedures. This invention also relates to a medical decision support method implemented by the medical decision support system. [Previous Technology]

[0002] In life, it is inevitable that people will seek treatment at the hospital due to health problems. However, when doctors diagnose patients, it takes a considerable amount of time to review all of the patient's medical history and individual circumstances. On the other hand, considering only part of the patient's medical history may lead to an incomplete medical decision. Therefore, how to provide intelligent decision-making assistance in the medical field has become a topic worthy of discussion. [Summary of the Invention]

[0003] One of the objectives of this invention is to provide a medical decision support system that can improve the incompleteness and inconvenience of the prior art, and avoid diagnosticians from making imperfect or erroneous diagnostic decisions due to limitations in their own knowledge and experience.

[0004] The medical decision support system of the present invention includes a storage unit, an input unit, an output unit, and a processing unit electrically connected to the storage unit, the input unit and the output unit. The storage unit stores a natural language processing model, a medical record database containing multiple historical medical records, and an intelligent decision-making module implemented using machine learning technology. The historical medical records correspond to multiple patients respectively. The processing unit is configured to: generate diagnostic data indicating the current patient during the diagnosis process of a current patient based on the operation received by the input unit; receive voice data related to the current patient via the input unit; select a target historical medical record corresponding to the current patient from the historical medical records based on the diagnostic data indicating the current patient; input the voice data into the natural language processing model to obtain dictated text data output by the natural language processing model; input the diagnostic data, the dictated text data, and the target historical medical record into the intelligent decision-making module to obtain an auxiliary decision result output by the intelligent decision-making module based on the diagnostic data, the dictated text data, and the target historical medical record; and control the output unit to output the auxiliary decision result, wherein the auxiliary decision result indicates medical reference information related to the symptoms of the current patient.

[0005] In some embodiments of the medical decision support system of the present invention, the input unit includes a recording device, and the voice data is generated by the recording device recording during a diagnosis performed by a diagnostician on the current patient. Furthermore, the spoken text data includes at least a patient's spoken portion that presents the content of the current patient's speech.

[0006] In some embodiments of the medical decision support system of the present invention, the oral text data further includes a diagnostic narration portion that presents the content of the diagnostician's narration.

[0007] In some embodiments of the medical decision support system of the present invention, the intelligent decision module is implemented using the random forest algorithm and contains multiple decision models, wherein each decision model includes multiple decision trees.

[0008] Another objective of the present invention is to provide a medical decision support method that can improve the incompleteness and inconvenience of the prior art, and avoid diagnosticians from making imperfect or erroneous diagnostic decisions due to limitations in their own knowledge and experience.

[0009] The medical decision support method of the present invention is implemented by a medical decision support system, which includes a storage unit, an input unit, an output unit, and a processing unit electrically connected to the storage unit, the input unit, and the output unit. The storage unit stores a natural language processing model, a medical record database containing multiple historical medical records, and an intelligent decision-making module implemented using machine learning technology. The historical medical records correspond to multiple patients respectively. The medical decision support system includes: a processing unit generating diagnostic data indicating the current patient based on operations performed on an input unit during the diagnosis process of a current patient; and receiving voice data related to the current patient via the input unit; the processing unit selecting a target historical medical record corresponding to the current patient from the historical medical records based on the diagnostic data indicating the current patient; the processing unit inputting the voice data into a natural language processing model to obtain dictated text data output by the natural language processing model; the processing unit inputting the diagnostic data, the dictated text data, and the target historical medical record into an intelligent decision-making module to obtain an auxiliary decision result output by the intelligent decision-making module based on the diagnostic data, the dictated text data, and the target historical medical record; and the processing unit controlling the output unit to output the auxiliary decision result, wherein the auxiliary decision result indicates medical reference information related to the symptoms of the current patient.

[0010] In some embodiments of the medical decision support method of the present invention, the input unit includes a recording device, and the voice data is generated by the recording device recording during a diagnosis performed by a diagnostician on the current patient. Furthermore, the spoken text data includes at least a patient's spoken portion that presents the content of the current patient's speech.

[0011] In some embodiments of the medical decision support method of the present invention, the oral text data further includes a diagnostic oral portion that presents the content of the diagnostician's oral statement.

[0012] In some embodiments of the medical decision support method of the present invention, the intelligent decision module is implemented using the random forest algorithm and includes multiple decision models, wherein each decision model includes multiple decision trees.

[0013] The advantage of this invention is that the medical decision support system is equivalent to an expert system that can be applied to the medical diagnosis process, and can provide more comprehensive intelligent decision support based on the patient's past medical records.

Implementation Method

[0015] Before the present invention is described in detail, it should be noted that similar elements are represented by the same reference numerals in the following description. Before the present invention is described in detail, it should be noted that unless otherwise defined, "electrical connection" as used in this patent specification refers to "wired electrical connection" in which multiple electronic devices / devices / components are interconnected through conductive materials, and "radio connection" in which one-way / two-way wireless signal transmission is performed through wireless communication technology. On the other hand, "electrical connection" as used in this patent specification also refers to "direct electrical connection" formed by multiple electronic devices / devices / components being directly interconnected, and "indirect electrical connection" formed by multiple electronic devices / devices / components being indirectly interconnected through other electronic devices / devices / components.

[0016] Referring to FIG1, one embodiment of the medical decision support system 1 of the present invention includes, for example, a storage unit 11, an input unit 12, an output unit 13, and a processing unit 14 electrically connected to the storage unit 11, the input unit 12 and the output unit 13.

[0017] In this embodiment, the storage unit 11 is implemented as a hard disk for storing digital data. However, in other embodiments, the storage unit 11 may also be, for example, flash memory, other types of computer-readable recording media, or a combination of various different computer-readable recording media. Therefore, the specific implementation of the storage unit 11 is not limited to this embodiment.

[0018] The input unit 12 includes, for example, a recording device, a keyboard, and a mouse, but is not limited thereto. The recording device may be used, for example, to record a diagnosis performed by a diagnostician (e.g., a physician) on a patient (e.g., a patient), while the keyboard and mouse may be used, for example, for the diagnostician to input relevant diagnostic data during the diagnosis process, such as the patient's self-reported symptoms, the diagnostician's assessment of the patient's symptoms, the diagnostician's recommended tests for the patient, the prescriptions written by the diagnostician, and the date the diagnostician recommends the patient for a follow-up appointment, but is not limited thereto.

[0019] In this embodiment, the output unit 13 is implemented as a display, for example, but is not limited thereto.

[0020] In this embodiment, the processing unit 14 is implemented as a central processing unit (CPU). However, in other embodiments, the processing unit 14 may also be implemented as a plurality of CPUs electrically connected to each other, or as a control circuit board including a CPU. Therefore, the specific implementation of the processing unit 14 is not limited to this embodiment.

[0021] Further, in this embodiment, the storage unit 11 and the processing unit 14 may be jointly included in a cloud server (not shown), and the input unit 12 and the output unit 13 may be jointly included in a computer device (not shown) for operation by the diagnostician and electrically connected to the cloud server via a network. In other words, the medical decision support system 1 in this embodiment may be implemented as the cloud server and the computer device electrically connected to the cloud server. However, in other embodiments, the storage unit 11 and the processing unit 14 may also be jointly included in the computer device, for example, along with the input unit 12 and the output unit 13. Therefore, the medical decision support system 1 may also be implemented as a single computer device, and is not limited to this embodiment.

[0022] In this embodiment, the storage unit 11 stores, for example, a natural language processing model D1, a medical record database D2 containing multiple historical medical records, and a smart decision-making module D3 implemented using machine learning technology.

[0023] In this embodiment, the natural language processing model D1 is used, for example, to convert speech data into text data, and to define a patient's oral statement and a diagnostician's oral statement based on the voiceprint features of the speech data. Furthermore, the natural language processing model D1 is also used to perform natural language understanding on the patient's oral statement and the diagnostician's oral statement, and to mark multiple diagnostic keywords related to medical behavior in the patient's oral statement and the diagnostician's oral statement. The diagnostic keywords may indicate, for example, the symptoms described by the patient, the disease suspected by the diagnostician, the tests and treatments recommended by the diagnostician, etc., but are not limited thereto.

[0024] In this embodiment, the historical medical records in the medical record database D2 correspond to multiple patients who have previously visited medical institutions. For each historical medical record, the record includes, for example, the patient's basic personal information (e.g., ID number, date of birth, gender, and age), as well as the patient's past medical records, including diagnostic records, test records, treatment records, and medical expense payment records, but is not limited thereto.

[0025] In this embodiment, the intelligent decision-making module D3 is implemented, for example, using the random forest algorithm and the Analytic Hierarchy Process (AHP), and the intelligent decision-making module D3 includes multiple decision models. In this embodiment, each decision model includes, for example, a decision tree group (i.e., multiple decision trees) and / or multiple hierarchical analysis modules each targeting a single disease. Moreover, in this embodiment, these decision models are, for example, a symptom decision model D31 for predicting the disease suffered by the patient, a test decision model D32 for predicting suitable test items for the patient, a treatment method decision model D33 for predicting suitable treatment methods for the patient, and a medical expense decision model D34 for predicting the medical expenses and suitable payment methods for the patient's current visit, but are not limited thereto.

[0026] More specifically, in this embodiment, the symptom decision model D31 includes, for example, multiple symptom decision trees, and each symptom decision tree is constructed, for example, based on the symptoms of multiple patients (e.g., but not limited to: frequent urination, drowsiness) and the doctor's disease diagnosis results for the symptoms (e.g., but not limited to: diabetes), but is not limited thereto. The test decision model D32 includes, for example, multiple test decision trees, and each test decision tree is constructed, for example, based on the symptoms of multiple patients and the test items determined by the doctor based on their symptoms (e.g., but not limited to: abdominal X-ray, upper gastrointestinal endoscopy), but is not limited thereto. The treatment decision model D33 includes, for example, multiple treatment decision trees, and each treatment decision tree is constructed, for example, based on the diseases suffered by multiple patients and the treatment methods adopted by the doctor for their diseases (e.g., specific prescriptions, surgeries, or treatment courses), but is not limited thereto. The medical expense decision model D34 includes, for example, multiple medical expense decision trees, and each medical expense decision tree is constructed based on, for example, the medical records of multiple patients and their corresponding payment records, but is not limited to this.

[0027] It should be further noted that, in other embodiments, the intelligent decision-making module D3 may, for example, include only one or more of the symptom decision model D31, the examination decision model D32, the treatment method decision model D33, and the medical cost decision model D34. Furthermore, each decision model may be implemented as a single decision tree, rather than necessarily as a group of decision trees. Additionally, the decision tree described in this embodiment may be constructed using algorithms in the prior art. Since the construction details of the decision tree are not the focus of this patent specification, their details are not described here.

[0028] The following exemplarily illustrates how the medical decision support system 1 of this embodiment implements a medical decision support method during the process of a diagnostician diagnosing a current patient.

[0029] First, in step S1, during the diagnosis process performed by the diagnostician on the current patient, the processing unit 14 generates diagnostic data corresponding to the current patient based on the operation received by the input unit 12, and the processing unit 14 also receives voice data related to the current patient through the input unit 12 during the diagnosis process.

[0030] More specifically, in this embodiment, the diagnostic data may be generated by the processing unit 14 based on the operation of the keyboard and mouse of the input unit 12 by the diagnostician. The diagnostic data may include, for example, identification information indicating the current patient and diagnostician's recommendations. The identification information may be, for example, the current patient's ID number, and the diagnostician's recommendations may include, for example, one or more of the following: the diagnostician's analysis and judgment of the current patient's symptoms, the list of medications prescribed by the diagnostician, the recommended tests by the diagnostician, the recommended treatment methods by the diagnostician, and the recommended next follow-up appointment time by the diagnostician, but are not limited thereto. On the other hand, the voice data may be the recording result generated by the recording device of the input unit 12 during the diagnosis process, recording both parties continuously. Therefore, the voice data may present, for example, the conversation between the current patient and the diagnostician during the diagnosis process in voice form. It is worth mentioning that, through this voice data, the data collected by the input unit 12 is the complete information of the diagnostic process, and not just the information input by the diagnostician. Therefore, it is different from the simplified and summarized diagnostic conclusion data of general medical auxiliary decision-making systems, which are based on the diagnostician's subjective judgment and selection of symptoms and input into the system.

[0031] After the processing unit 14 obtains the diagnostic data and the voice data, the process proceeds to step S2, for example.

[0032] In step S2, for the diagnostic data, the processing unit 14, for example, selects a target historical medical record corresponding to the current patient from the historical medical records in the medical record database D2 based on the identity information contained in the diagnostic data, thereby obtaining the current patient's past medical records, including diagnoses, tests, treatments, and medical expense payments. On the other hand, for the voice data, the processing unit 14, for example, inputs the voice data into the natural language processing model D1 to obtain dictated text data output by the natural language processing model D1 corresponding to the voice data. Furthermore, in this embodiment, the dictated text data includes, for example, a patient's dictated portion presenting the current patient's dictated content in text form, and a diagnostician's dictated portion presenting the diagnostician's dictated content in text form.

[0033] After the processing unit 14 selects the target historical medical record and obtains the oral text data, the process proceeds to step S3.

[0034] In step S3, when the processing unit 14 receives an analysis command from the input unit 12, the processing unit 14 inputs the diagnostic data, the spoken text data, and the target historical medical record into the intelligent decision-making module D3 according to the analysis command, so as to obtain an auxiliary decision-making result output by the intelligent decision-making module D3 based on the diagnostic data, the spoken text data, and the target historical medical record. The analysis command may be generated, for example, by the manual operation of the keyboard or mouse of the input unit 12 by the patient, and the auxiliary decision-making result may indicate medical reference information related to the symptoms of the current patient. Furthermore, the medical reference information may include, but is not limited to, "the matching rate and ranking of the current condition description, the ranking of the best medical advice recommendation, the ranking of the best medication, the ranking of the best surgical procedure, and the ranking of the best follow-up medical or postoperative care decisions," etc.

[0035] More specifically, in this embodiment, the auxiliary decision-making result includes, for example, a symptom analysis result output by the symptom decision model D31, a test item analysis result output by the test decision model D32, a treatment method analysis result output by the treatment method decision model D33, and a medical cost analysis result output by the medical cost decision model D34. The symptom analysis result, for example, indicates the possible disease of the current patient, and the symptom analysis result is generated by the symptom decision model D31 based on the target's historical medical records, including past diagnostic records, test records, treatment records, and symptoms indicated by the patient's verbal description, but is not limited thereto. The analysis results of the test items may indicate one or more suggested test items related to the current patient's symptoms. Furthermore, the analysis results may be generated by the test decision model D32 based on the current patient's past diagnostic records, test records, treatment records, symptoms indicated by the patient's verbal description, and the suspected disease indicated by the diagnostic person's verbal description, but are not limited thereto. Similarly, the analysis results of the treatment methods may indicate one or more suggested treatment methods related to the current patient's symptoms. Furthermore, the analysis results may be generated by the treatment method decision model D33 based on the current patient's past diagnostic records, test records, treatment records, symptoms indicated by the patient's verbal description, and the suspected disease indicated by the diagnostic person's verbal description, but are not limited thereto. The medical cost analysis results, for example, indicate a predicted amount and a suggested payment method for the predicted amount (e.g., same-day payment, installment payment, or deferred payment). Moreover, the medical cost analysis results are generated by the medical cost decision model D34 based on the current patient's past treatment and medical cost payment records, as well as the diagnostician's recommended tests and / or treatments for the current patient, but are not limited thereto.

[0036] It should be noted that in other embodiments, the symptom analysis results may include, for example, only one or more of the symptom analysis results, the test results, the treatment method analysis results, and the medical cost analysis results, and are not limited to this embodiment.

[0037] After the processing unit 14 obtains the auxiliary decision result, the process proceeds to step S4.

[0038] In step S4, the processing unit 14 controls the output unit 13 to output the auxiliary decision result for reference by the diagnostician and the current patient. On the other hand, the auxiliary decision result can also be further used to optimize and adjust the nodes of each decision tree in the intelligent decision module D3. Specifically, the processing unit 14 controls the output unit 13 to output the auxiliary decision result in a manner that, for example, automatically arranges the optimal medical decision path diagram, the lowest medical cost / insurance premium consumption decision path diagram, the best medical privilege decision path diagram, or other optimal combinations of various path diagrams, based on the current patient's payment habits and the medical needs reflected in the relevant question-and-answer statements, using a mind map model.

[0039] The above is an example of how the medical decision support system 1 of this embodiment implements the medical decision support method.

[0040] In summary, by implementing this medical decision support method, the medical decision support system 1 is equivalent to an expert system that can be applied to the medical diagnosis process, and can provide more comprehensive intelligent decision support based on the patient's past medical records, thus achieving the purpose of this invention.

[0041] However, the above description is only an embodiment of the present invention and should not be construed as limiting the scope of the present invention. Any simple equivalent changes and modifications made in accordance with the scope of the patent application and the contents of the patent specification shall still fall within the scope of the patent of the present invention. [Simplified Explanation of the Diagram]

[0014] Other features and effects of the present invention will be clearly presented in the embodiments with reference to the drawings, wherein: FIG1 is a block diagram of an embodiment of the medical decision support system of the present invention; and FIG2 is a flowchart illustrating how the medical decision support method is implemented in this embodiment.

Claims

1. A medical decision support system, comprising: a storage unit storing a natural language processing model, a medical record database containing multiple historical medical records, and an intelligent decision-making module implemented using machine learning technology, wherein the historical medical records correspond to multiple patients; an input unit; an output unit; and a processing unit electrically connected to the storage unit, the input unit, and the output unit; wherein, The processing unit is configured to: generate diagnostic data indicating the current patient during the diagnosis process of a current patient based on the operation received by the input unit; receive voice data related to the current patient via the input unit; select a target historical medical record corresponding to the current patient from the historical medical records based on the diagnostic data indicating the current patient; input the voice data into the natural language processing model to obtain dictated text data output by the natural language processing model; input the diagnostic data, the dictated text data, and the target historical medical record into the intelligent decision-making module to obtain an auxiliary decision result output by the intelligent decision-making module based on the diagnostic data, the dictated text data, and the target historical medical record; and control the output unit to output the auxiliary decision result, wherein the auxiliary decision result indicates medical reference information related to the symptoms of the current patient.

2. The medical decision support system as described in claim 1, wherein, The input unit includes a recording device, and the audio data is generated by the recording device recording during a diagnosis performed by a diagnostician on the current patient. The spoken text data includes at least a patient's spoken portion that presents the content of the current patient's speech.

3. The medical decision support system as described in claim 2, wherein, The oral transcript also includes a portion of the diagnostic report that presents the content of the diagnosis.

4. The medical decision support system as described in claim 1, wherein, This intelligent decision-making module is implemented using the random forest algorithm and contains multiple decision models, each of which includes multiple decision trees.

5. A medical decision support method, implemented by a medical decision support system, the medical decision support system comprising a storage unit, an input unit, an output unit, and a processing unit electrically connected to the storage unit, the input unit, and the output unit, wherein the storage unit stores a natural language processing model, a medical record database containing multiple historical medical records, and a smart decision-making module implemented using machine learning technology, and the historical medical records correspond to multiple patients respectively; the medical decision support system comprises: the processing unit generating diagnostic data indicating the current patient based on the operation received by the input unit during the diagnosis of a current patient, and receiving voice data related to the current patient through the input unit; the processing unit selecting a target historical medical record corresponding to the current patient from the historical medical records based on the current patient indicated by the diagnostic data; and the processing unit inputting the voice data into the natural language processing model to obtain dictated text data output by the natural language processing model; The processing unit inputs the diagnostic data, the oral written data, and the target's historical medical records into the intelligent decision-making module to obtain an auxiliary decision-making result output by the intelligent decision-making module based on the diagnostic data, the oral written data, and the target's historical medical records; and the processing unit controls the output unit to output the auxiliary decision-making result, wherein... The decision support results indicate medical reference information relevant to the symptoms of the current patient.

6. The medical decision support method as described in claim 5, wherein, The input unit includes a recording device, and the audio data is generated by the recording device recording during a diagnosis performed by a diagnostician on the current patient. The spoken text data includes at least a patient's spoken portion that presents the content of the current patient's speech.

7. The medical decision support method as described in claim 6, wherein, The oral transcript also includes a portion of the diagnostic report that presents the content of the diagnosis.

8. The medical decision support method as described in claim 5, wherein, This intelligent decision-making module is implemented using the random forest algorithm and contains multiple decision models, each of which includes multiple decision trees.

Images