CN115331805B - Mental disease diagnosis method based on natural language processing and computer equipment - Google Patents
Mental disease diagnosis method based on natural language processing and computer equipment Download PDFInfo
- Publication number
- CN115331805B CN115331805B CN202210785536.3A CN202210785536A CN115331805B CN 115331805 B CN115331805 B CN 115331805B CN 202210785536 A CN202210785536 A CN 202210785536A CN 115331805 B CN115331805 B CN 115331805B
- Authority
- CN
- China
- Prior art keywords
- label
- open
- user
- model
- question
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/20—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/16—Devices for psychotechnics; Testing reaction times ; Devices for evaluating the psychological state
- A61B5/165—Evaluating the state of mind, e.g. depression, anxiety
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/35—Clustering; Classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/237—Lexical tools
- G06F40/242—Dictionaries
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Public Health (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Biomedical Technology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Databases & Information Systems (AREA)
- Psychiatry (AREA)
- Pathology (AREA)
- Social Psychology (AREA)
- Surgery (AREA)
- Hospice & Palliative Care (AREA)
- Biophysics (AREA)
- Educational Technology (AREA)
- Developmental Disabilities (AREA)
- Heart & Thoracic Surgery (AREA)
- Child & Adolescent Psychology (AREA)
- Molecular Biology (AREA)
- Psychology (AREA)
- Animal Behavior & Ethology (AREA)
- Computational Linguistics (AREA)
- Veterinary Medicine (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Artificial Intelligence (AREA)
- Epidemiology (AREA)
- Primary Health Care (AREA)
- Medical Treatment And Welfare Office Work (AREA)
- Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
Abstract
The application provides a mental disease diagnosis method based on natural language processing, which comprises the following steps: obtaining replies of users to questions in the analog video inquiry process, wherein the questions comprise a duration question, an open question, a Kensing question and a past medical history question; extracting labels of replies of users aiming at problems based on a preset model; and confirming the probability of the mental diseases of the user according to the label extraction result. According to the mental disease diagnosis method, the sign labels of different types of problems are extracted based on the preset model, and the accuracy of label recognition degree is improved by mining association of each label in anticipation. The diagnosis method of the mental diseases is according to the diagnosis results, scientific and objective, and has universality relatively, so that the mental disease diagnosis of a large range of people can be realized, and the diagnosis requirements of a large number of mental disease patients under the condition of limited resource conditions are further met.
Description
Technical Field
The invention relates to the technical field of disease diagnosis, in particular to a mental disease diagnosis method based on natural language processing, computer equipment and a storage medium.
Background
In the related art, the diagnosis of depression anxiety is mainly implemented by manual inquiry, and the talking content is mainly to inquire about the current state of the patient, such as whether there is sleep disorder, duration, etc., and then diagnose the mental diseases such as depression, anxiety, etc. according to the manifestation of symptoms and duration of symptoms. The manual inquiry method is effective in diagnosing depression and anxiety, but due to limited professional resources, the requirements of all patients can not be met in time, and the patients are not aware of the insufficient cognition of the diseases, so that the patients are not aware of the need of the diagnosis or have the feeling of pubic disease and are unwilling to visit, the diagnosis is not in time, and more serious illness is caused.
Disclosure of Invention
In view of the above, embodiments of the present invention provide a mental disease diagnosis method, a computer apparatus, and a storage medium based on natural language processing.
The embodiment of the invention provides a mental disease diagnosis method based on natural language processing, which comprises the following steps:
obtaining replies of users to questions in the analog video inquiry process, wherein the questions comprise a duration question, an open question, a Kensing question and a past medical history question;
extracting labels from replies of users aiming at the problems based on a preset model;
and confirming the probability of the mental diseases of the user according to the label extraction result.
In this way, in the mental disease diagnosis method according to the embodiment of the present application, by extracting the symptom label of different types of problems based on the predetermined model, and by mining the association of each label in anticipation, the accuracy of the label recognition degree is improved. The diagnosis method of the mental diseases is according to the diagnosis results, scientific and objective, and has universality relatively, so that the mental disease diagnosis of a large range of people can be realized, and the diagnosis requirements of a large number of mental disease patients under the condition of limited resource conditions are further met.
In some embodiments, the extracting the label from the reply of the user to the question based on the predetermined model includes:
and matching in a pre-created word list in a regular matching mode according to the reply of the user to the past medical history problem so as to extract medical history labels.
Thus, by identifying the tags in a regular matching manner, a disease description that has been typical of the past medical history of the user can be identified.
In some embodiments, the extracting the label from the reply of the user to the question based on the predetermined model includes:
and according to the reply of the user to the Kennel question, carrying out Kennel label extraction based on a pre-trained Kennel model.
Thus, the affirmative-negative model is used for judging the relatively simple affirmative and negative answers of the user, and extracting relevant labels according to the reply symptoms of the user.
In some embodiments, the extracting the label from the reply of the user to the question based on the predetermined model includes:
and extracting an open model label based on a pre-trained open problem model according to the extracted medical history label, the Kennel label and the reply of the user to the open problem.
Therefore, in consideration of the problem of Kennel and the association and influence of past medical history and the current mental state of the user, the extracted medical history labels and Kennel labels are matched with the reply of the user, the labels of the open problem reply are extracted, the association between corpora is enhanced, and the accuracy of the label recognition degree is improved.
In some embodiments, the extracting the open model label based on the pre-trained open problem model according to the extracted medical history label, the affirmative label and the reply of the user to the open problem comprises:
performing secondary extraction on the extracted medical history label and the extracted affirmative label through a label word list;
and extracting the medical history label, the Kennel label and the reply of the user to the open type problem after the secondary extraction based on a pre-trained open type problem model.
Therefore, the extracted medical history labels and the Kennel labels are subjected to secondary extraction of the label word list and are filtered, so that the extracted labels can be suitable for a subsequent problem model.
In some embodiments, the open question includes a plurality of the post-secondary extraction medical history labels, affirmative labels, and user replies to the open question, and the open model label extraction is performed based on a pre-trained open question model, including:
and inputting the medical history label after the secondary extraction, the Kennel label, the open model label extracted according to the reply of the user to the previous open problem and the reply of the user to the current open problem into the open problem model so as to extract the open model label aiming at the current open problem.
Thus, the open questions are mainly used for extracting labels of relevant expressions of mental diseases from answers of users, a plurality of open questions are associated with each other, the extracted labels and replies of the current questions are used as inputs of a model together, the labels of replies of the current questions are output, relevance among corpus is enhanced, and accuracy of label extraction is improved.
In some embodiments, the extracting the open model label based on the pre-trained open problem model according to the medical history label after the second extracting, the affirmative label and the reply of the user to the open problem comprises:
and under the condition that the current problem is the first open problem, inputting the medical history label, the Kennel label and the reply of the user to the current open problem after the second extraction into the open problem model under the condition that the current problem is the first open problem so as to extract the open model label to the current open problem.
Thus, when the current problem is the first open problem, the reply to the current problem is combined with the extracted affirmative label and the medical history label to be used as input together, so that the accuracy of label extraction is improved.
In some embodiments, the extracting the label from the reply of the user to the question based on the predetermined model includes:
and according to the reply of the user to the time problem, extracting a disease duration label based on a pre-trained duration model.
Thus, the duration of the illness is one of the criteria for diagnosis of the illness, and can be extracted from the relevant description of the user through a duration model.
In some embodiments, the determining the probability of the user having a mental disorder according to the result of the tag extraction includes:
and confirming the probability of the mental diseases of the user according to the extracted open model label and the disease duration label.
In this way, the open model label and the continuous duration label of the illness state, which are combined with the Ken-NO label and the medical history label, are used as input together to output the probability of finally suffering from the related illness, so that the help is provided for further diagnosis and treatment of the user.
The present invention provides a computer device comprising a memory and a processor, the memory having stored therein a computer program which, when executed by the processor, implements a method as claimed in any one of the preceding claims.
Therefore, the computing equipment of the method and the device can extract the symptom labels of the problems of different types based on the preset model, and improve the accuracy of the label recognition degree by mining the association of each label in anticipation. The diagnosis method of the mental diseases is according to the diagnosis results, scientific and objective, and has universality relatively, so that the mental disease diagnosis of a large range of people can be realized, and the diagnosis requirements of a large number of mental disease patients under the condition of limited resource conditions are further met.
The present invention provides a non-transitory computer readable storage medium for a computer program which, when executed by one or more processors, causes the processors to perform the method.
Therefore, in the method, the problem of different types is subjected to symptom label extraction based on the preset model, and the accuracy of label recognition degree is improved by mining association of each label in anticipation. The diagnosis method of the mental diseases is according to the diagnosis results, scientific and objective, and has universality relatively, so that the mental disease diagnosis of a large range of people can be realized, and the diagnosis requirements of a large number of mental disease patients under the condition of limited resource conditions are further met.
Additional aspects and advantages of embodiments of the application will be set forth in part in the description which follows and, in part, will be obvious from the description, or may be learned by practice of the application.
Drawings
The foregoing and/or additional aspects and advantages of the invention will become apparent and readily appreciated from the following description of the embodiments, taken in conjunction with the accompanying drawings, in which:
FIG. 1 is a schematic flow chart of a mental disorder diagnosis method according to certain embodiments of the present invention;
FIG. 2 is a schematic flow chart of a mental disorder diagnosis method according to some embodiments of the present invention.
Detailed Description
Embodiments of the present invention are described in detail below, examples of which are illustrated in the accompanying drawings, wherein like or similar reference numerals refer to like or similar elements or elements having like or similar functions throughout. The embodiments described below by referring to the drawings are illustrative and intended to explain the present invention and should not be construed as limiting the invention.
In recent years, the incidence of mental diseases has increased year by year, and the variety of diseases has also increased, such as dysphoria, anxiety, depression, phobia, obsessive compulsive disorder, etc. In the case of depression, which is a mental disorder characterized mainly by a remarkable and persistent mood disorder, patients mainly show loss of interest, low mood, and insufficient energy, and may be accompanied by anxiety, self-crime, low self-evaluation, slow thinking, difficulty in attention, memory loss, appetite loss, weight loss, or sleep abnormality. Most of mental diseases can be cured in early stage, people cannot realize that the people tend to or have the mental diseases, and the diagnosis of the mental diseases is mainly dependent on professionals at present, the quantity of professionals is limited, and the mental diseases of a large range of people are difficult to diagnose, so that hidden danger of the diseases is found.
Referring to fig. 1 and 2, the present application provides a mental disease diagnosis method based on natural language processing, including:
s10: obtaining replies of users to questions in the analog video inquiry process, wherein the questions comprise a duration question, an open question, a Kensing question and a past medical history question;
s20: extracting labels of replies of users aiming at problems based on a preset model;
s30: and confirming the probability of the mental diseases of the user according to the label extraction result.
The present application also provides a computer device by which the mental disorder diagnosis method of the present application can be implemented. The computer device comprises a memory and a processor, wherein the memory stores a computer program, and the processor is used for acquiring a reply of a user aiming at a problem in the simulated video inquiry process, extracting a label of the reply of the user aiming at the problem based on a preset model and confirming the probability of the user suffering from mental diseases according to the label extraction result. The computer device in the present application may be a medical diagnostic device having processing capabilities.
In particular, mental disease diagnosis in the present application is realized based on natural language processing. The medical scale is widely applied to clinical practice, provides an important reference basis for standard diagnosis and treatment of diseases, and greatly improves diagnosis efficiency and accuracy. In the application, a user needs to fill in a medical scale for replying to the diagnosis of mental diseases, and then pertinently selects a question for carrying out a simulated video inquiry from a pre-constructed question library according to the reply of the user to the medical scale. Analog video interrogation is the process of simulating a question at a hospital doctor using an artificial intelligence based computer device. The medical scale to which the user replies can provide the user with an abnormality in which aspects there may be psychological problems such as depression, anxiety, etc. Therefore, the questions can be selected for video inquiry according to the scale scores.
The question library comprises a plurality of preset questions, the labels which are to be extracted from each question correspond to the questions one by one, and the label content is the core thought which can be expressed when the user answers the questions. In the flow, the problems are divided into different types according to the content of the labels, and the labels are extracted by different methods, including an open model, a Kensin model, a vocabulary matching mode and the like. Correspondingly, questions include duration questions, open questions, kennel questions, and past history questions. According to the answers of the user to each type of questions, the labels related to the disease symptoms can be finally extracted by utilizing the corresponding model trained in advance, and the labels are combined with each other to finally obtain the probability of the user of having related mental diseases.
It will be appreciated that different types of questions may obtain information from different angles to the user, such as information that may not be available through affirmative questions, such as the specific form of mental illness, then open questions may be designed and relevant information obtained from the open questions.
In summary, in the mental disease diagnosis method and the computer device according to the embodiments of the present application, the symptom label extraction is performed on the problems of different types based on the predetermined model, and the accuracy of the label recognition degree is improved by mining the association of each label in the expectation. The diagnosis method of the mental diseases is according to the diagnosis results, scientific and objective, and has universality relatively, so that the mental disease diagnosis of a large range of people can be realized, and the diagnosis requirements of a large number of mental disease patients under the condition of limited resource conditions are further met.
In certain embodiments, S40 comprises:
s41: and matching in a pre-created word list in a regular matching mode according to the reply of the user to the past medical history problem so as to extract medical history labels.
In some embodiments, the processor is configured to match in a pre-created vocabulary by means of regular matching for medical history tag extraction based on user responses to past medical history questions.
Specifically, a vocabulary may be created in advance, and the vocabulary includes names of existing diseases. The vocabulary matching can extract labels in a regular matching mode, and is mainly used for identifying past disease history and typical psychological disease description of users. For example, the problems are: is you have what physical illness before? User answer: i have asthma. Asthma is in the vocabulary, then through the mode of regular matching just can be accurate draw user's medical history label: [ 'history of asthma' ]. The matching of the vocabulary can provide higher accuracy. And the vocabulary can be updated at any time to realize faster extraction of medical history labels, and the method is more flexible than a method for extracting labels by using a model.
It can be appreciated that, when a large amount of training data is required to be utilized to extract the tag by using the model, the tag beyond the range of the training data cannot be extracted in the subsequent process of extracting the tag by using the model. And diseases are added or deleted, the labeling and model training are needed to be carried out again, and the iteration speed is low. And the vocabulary can be updated at any time.
Thus, by identifying the tags in a regular matching manner, a disease description that has been typical of the past medical history of the user can be identified.
In certain embodiments, S40 comprises:
s42: and according to the reply of the user to the Kennel question, carrying out Kennel label extraction based on a pre-trained Kennel model.
In some of these approaches, the processor is configured to perform a ken-n label extraction based on a pre-trained ken-n model based on a user's reply to a ken-n question.
Specifically, a Kennel model is used to determine simple positive and negative answers by the user. For example, the problem may be: do you experience heavy frustration? User answer: the last few years car accidents led to the fact that the left-hand response is now not sensitive. The affirmative-negative model will determine that the answer is affirmative and generate the label "experience significant frustration". The core construction mode of the model can be a general GRU model mode.
Wherein the tag can be foreseen to be constructed according to a mental or psychological disease diagnostic manual, such as a depression type tag: "mood disorder", "insomnia", "loss of interest or pleasure", etc.; anxiety-related labels: dyspnea or palpitation, restlessness, etc. Corresponding labels are extracted from the questions according to positive or negative answers of the user.
Thus, the affirmative-negative model is used for judging the relatively simple affirmative and negative answers of the user, and extracting relevant labels according to the reply symptoms of the user.
In certain embodiments, S40 comprises:
s43: and extracting an open model label based on a pre-trained open problem model according to the extracted medical history label, the Kennel label and the reply of the user to the open problem.
In some of these approaches, the processor is configured to perform open model label extraction based on a pre-trained open problem model based on the extracted medical history labels, the Kennel labels, and the user's responses to the open problem.
In particular, open questions, i.e. whether they cannot be simply answered, require a more detailed description by the user. The open question model is used to extract from the user answers the vocabulary of labels for relevant manifestations of the mental disorder being diagnosed, for example for depression or anxiety, the labels of the relevant labels may be hypopneas, hypo-mood, etc.
For example, for the question "what affects your daily life is the current puzzlement. The open problem model matches all possible labels, the preset labels can form n-dimensional vectors, and the result returned after model identification is the probability for each label in the matrix, for example, the returned result is [ A:0.1, B:0.2, C:0.6, D:0.8 ]. Setting a threshold value of 0.5, that is, a returned result higher than the threshold value, confirming that the sign probability corresponding to the label is larger, and giving the label to the user, the final returned result is [ C, D ]. If the user answers to the question: "he will put me on a negative emotional stress for some time and then not put any interest in things, including long-term insomnia, and then have some effect on the rest time of the next day. "based on the answer and the set threshold, through the open question model, the label [ 'interest decline', 'mood fall' ], the other label thresholds are all lower than the set value, so no return is made. These labels are one of the manifestations of mental illness and are finally summarized together with the labels extracted from other questions to participate in judging whether the user has a certain mental illness.
The open problem model has higher requirements on accuracy, can be used for migration training based on the Bert model, and is added with a sigmoid function at an output layer. In the model design, since questions for related diseases have been determined, there is a causal relationship between each question in the user's answer, and the input design of the model is to extract a label according to the questions that the user has answered, such as a medical history label, a Kennel label, etc., and output as a label extracted according to the user's answer to the current question.
For example, for the case where the user fills out a medical scale, 3 open questions are designed for the user. For each problem, when the label extraction is performed by using the model, the label extraction is performed according to the serial number sequence of the problem. The input of the open question model consists of two parts, one part being the labels extracted from the previous questions and the other part being the answers to the current questions. The extracted tag for the current question is input as part of the next question along with other tags of the previous questions, and the cycle is repeated until the final tag is obtained.
Therefore, in consideration of the problem of Kennel and the association and influence of past medical history and the current mental state of the user, the extracted medical history labels and Kennel labels are matched with the reply of the user, the labels of the open problem reply are extracted, the association between corpora is enhanced, and the accuracy of the label recognition degree is improved.
In certain embodiments, S43 comprises:
s430: extracting the extracted medical history label and the label of Kennel for the second time through a label word list;
s431: and extracting the open model label based on the pre-trained open problem model by using the medical history label, the Kennel label and the reply of the user aiming at the open problem after the secondary extraction.
In some embodiments, the processor is configured to perform a secondary extraction of the extracted medical history tag, the ken tag via the tag vocabulary, and perform an open model tag extraction based on a pre-trained open problem model for the extracted medical history tag, the ken tag, and a user's reply to the open problem.
Specifically, the tag vocabulary is used for extracting the extracted medical history tag and the ken tag for the second time, so that the content which is not suitable for being input as the open problem model in the medical history tag and the ken tag is filtered. It will be appreciated that the open problem model may be subjected to extensive data training to extract labels, and if its inputs are subsequently updated, such as by adding or subtracting from the disease history vocabulary, the open problem model also needs to be trained by reusing the updated data, which would otherwise result in a reduced accuracy in extracting labels using the open problem model.
That is, in actual use, the tag vocabulary may have a smaller range than the range of the affirmative question tag and the medical history tag set, and thus secondary filtering is required. In the case where no training material for the model or the table of medical history is actually updated, the step may be omitted, and the extracted label for the model and the extracted label for the medical history may be directly input as part of the open-ended problem model.
Therefore, the extracted medical history labels and the Kennel labels are subjected to secondary extraction of the label word list and are filtered, so that the extracted labels can be suitable for a subsequent problem model.
In certain embodiments, S431 comprises:
s4310: and inputting the medical history label after the secondary extraction, the Kennel label, the open model label extracted according to the reply of the user to the previous open problem and the reply of the user to the current open problem into the open problem model so as to extract the open model label aiming at the current open problem.
In some embodiments, the processor is configured to input the second extracted medical history label, the affirmative label, the open model label extracted from the user's reply to the previous open question, and the user's reply to the current open question into the open question model to extract the open model label for the current open question.
Specifically, the affirmative label and the medical history label after the secondary extraction of the label vocabulary and the label extracted by the reply to the previous open question are used as the input of the label extraction, and the label extraction is performed through the open question model.
For example, a total of 3 open questions are presented for the current user:
what can you help you?
What is the cause of the current trouble?
Is there a current affliction to what impact you have on their daily lives?
The open problem model is used one by one for the 3 open problems for label extraction.
Taking labels of the second and third questions as an example, for the first question, the user answers "recent depression", and the labels extracted according to the open question model are [ 'no harm to own mind', 'no harm to own behavior', 'negative emotional stress' ].
For the second question, the user answers "I and girlfriend hands away". "
At this time, the input of the open question model is the label extracted from the previous question [ 'no harm to own mind', 'no harm to own behavior', 'negative emotional stress', ] and the answer of the current question "i and girl friends have hands away". Can extract the label [ 'marital love' ]
And by analogy, extracting the label of the third problem, and returning the third problem to the trouble and poor sleep. ". The model input of the label extraction at this time is the label extracted according to the previous question [ 'no harm to own mind', 'no harm to own behavior', 'negative emotion pressure', 'marital love', 'answer' trouble, sleep not good. The label can be extracted as [ 'insomnia' ].
The subsequent open problems can be processed according to the above method, and the accuracy of extracting the tags of the subsequent problems is improved mainly by weighting the tags.
The number of open questions, the content of questions, user replies and extracted labels in this embodiment are schematically described, and are mainly used for describing the form of cyclic extracted labels.
Thus, the open questions are mainly used for extracting labels of relevant expressions of mental diseases from answers of users, a plurality of open questions are associated with each other, the extracted labels and replies of the current questions are used as inputs of a model together, the labels of replies of the current questions are output, relevance among corpus is enhanced, and accuracy of label extraction is improved.
In certain embodiments, S431 comprises:
s4311: under the condition that the current problem is the first open problem, the medical history label, the Kennel label and the reply of the user aiming at the current open problem after the secondary extraction are input into an open problem model so as to extract an open model label aiming at the current open problem.
In some embodiments, the processor is configured to input the second extracted medical history tag, the affirmative tag, and the user's reply to the current open question into the open question model to extract an open model tag for the current open question if the current question is the first open question.
Specifically, with the above-mentioned example, the description is continued of the tag extraction for the first open problem.
For the first problem, there is no label extracted from the prior open problem model. Thus, the labels included in the input of the model are relevant labels extracted from the Kennel model and the medical history vocabulary. In addition, the input of the model includes the user's answer to the first question. For example, the user answers "recent depression somewhat" for the first question. "extracted affirmative no label and medical history label are [ ' ideas without hurting oneself ', ' behaviors without hurting oneself ], and extracted label is [ ' negative emotional stress ' ].
The questioning process and the sequence of extracting the labels are not consistent, and the questioning can ask questions according to the sequence of questions designed by professional consultants, ask main complaints first, ask reasons and influences, and ask questions such as disease history. In the case of recognition, whether a question is affirmative or negative is preferentially recognized, and whether the question is not substantially correlated with another question is judged, so that a label extracted from the question is input as a part of an open question model. The order may then be to identify past medical history, major and long-term diseases are likely to have an impact on a person's mental state, and similarly, the extracted tags may be entered as part of an open question model. And then matching the previously extracted labels with the answers of the users, and sequentially extracting the labels from each open question.
Thus, when the current problem is the first open problem, the reply to the current problem is combined with the extracted affirmative label and the medical history label to be used as input together, so that the accuracy of label extraction is improved.
In certain embodiments, S40 comprises:
s44: and according to the reply of the user to the time problem, extracting the condition duration label based on a pre-trained duration model.
In some embodiments, the processor is configured to perform condition duration label extraction based on a pre-trained duration model according to user replies to the time questions.
Specifically, the duration classification model is used to judge the duration of symptoms described by the user, such as weeks, days, or months, and in the diagnosis of mental diseases, a certain duration is required for the duration of certain symptoms, such as 2 weeks, 4 weeks, etc., according to the related medical professional diagnosis manual. In the judgment of the final disease probability, the duration is used as an input parameter, and the duration is used as a final judgment condition together with other labels describing symptoms. The core construction mode of the duration model can be a general GRU model mode
Thus, the duration of the illness is one of the criteria for diagnosis of the illness, and can be extracted from the relevant description of the user through a duration model.
In certain embodiments, S50 comprises:
s51: and confirming the probability of the mental diseases of the user according to the extracted open model label and the disease duration label.
In some embodiments, the processor is configured to confirm the probability of the user having a mental disorder based on the extracted open model label and the disorder duration label.
Specifically, in the judgment of the final disease probability, the duration is used as an input parameter, and the final judgment condition is used together with other labels describing symptoms. The extracted duration label and the open model label are used as the input of a final diagnosis probability model. For example, 100 labels are set, it can be represented by a 100-dimensional vector, and if some labels exist for the user, it is represented by 1 at the vector of the corresponding label, otherwise it is represented by 0. And splicing 2 dimensions to represent the duration of symptoms, namely a duration label, and finally obtaining a 102-dimensional vector. Using the activation function sigmoid at the fully connected layer, the final output is an m-dimensional vector for the number of disease categories, e.g., diagnosing depression and anxiety, and the outputs ([ P1, P2 ]) represent the likelihood of depression and anxiety, respectively, for that user. .
In this way, the open model label and the continuous duration label of the illness state, which are combined with the Ken-NO label and the medical history label, are used as input together to output the probability of finally suffering from the related illness, so that the help is provided for further diagnosis and treatment of the user.
Embodiments of the present application also provide a computer-readable storage medium. One or more non-transitory computer-readable storage media containing computer-executable instructions that, when executed by one or more processors, cause the processors to perform the method of any of the above embodiments.
As such, the present invention provides a non-transitory computer-readable storage medium storing a computer program that, when executed by one or more processors, causes the processors to perform a mental disorder diagnosis method.
Those skilled in the art will appreciate that all or part of the processes in the methods of the above embodiments may be implemented by a computer program, and the program may be stored in a non-volatile computer readable storage medium, where the program, when executed, may include processes in the embodiments of the methods described above. The storage medium may be a magnetic disk, an optical disk, a Read-Only Memory (ROM), or the like.
The foregoing examples represent only a few embodiments of the present application, which are described in more detail and are not thereby to be construed as limiting the scope of the present application. It should be noted that it would be apparent to those skilled in the art that various modifications and improvements could be made without departing from the spirit of the present application, which would be within the scope of the present application. Accordingly, the scope of protection of the present application is to be determined by the claims appended hereto.
Claims (9)
1. A method for diagnosing a mental disorder based on natural language processing, the method comprising:
obtaining replies of users to questions in the analog video inquiry process, wherein the questions comprise a duration question, an open question, a Kensing question and a past medical history question;
extracting labels from replies of users aiming at the problems based on a preset model;
confirming the probability of the mental diseases of the user according to the label extraction result;
the label extraction of the reply of the user to the problem based on the predetermined model comprises the following steps:
extracting a medical history label according to the reply of the user to the past medical history problem;
extracting a Kelly label according to the reply of the user to the Kelly question;
performing secondary extraction on the extracted medical history label and the extracted affirmative label through a label word list;
and extracting the medical history label, the Kennel label and the reply of the user to the open type problem after the secondary extraction based on a pre-trained open type problem model.
2. The diagnostic method of claim 1, wherein the label extraction of the user's reply to the question based on a predetermined model comprises:
and matching in a pre-created word list in a regular matching mode according to the reply of the user to the past medical history problem so as to extract medical history labels.
3. The diagnostic method of claim 2, wherein the label extraction of the user's reply to the question based on a predetermined model comprises:
and according to the reply of the user to the Kennel question, carrying out Kennel label extraction based on a pre-trained Kennel model.
4. The diagnostic method of claim 1, wherein the open question comprises a plurality of the post-secondary extraction medical history labels, affirmative labels, and user responses to the open question, the open model label extraction based on a pre-trained open question model comprising:
and inputting the medical history label after the secondary extraction, the Kennel label, the open model label extracted according to the reply of the user to the previous open problem and the reply of the user to the current open problem into the open problem model so as to extract the open model label aiming at the current open problem.
5. The diagnostic method of claim 4, wherein the performing open model label extraction based on a pre-trained open problem model based on the post-secondary extraction medical history label, the positive negative label, and the user's reply to the open problem comprises:
and under the condition that the current problem is the first open problem, inputting the second extracted medical history label, the Kennel label and the reply of the user to the current open problem into the open problem model so as to extract the open model label to the current open problem.
6. The diagnostic method of claim 1, wherein the label extraction of the user's reply to the question based on a predetermined model comprises:
and according to the reply of the user to the duration problem, extracting the condition duration label based on a pre-trained duration model.
7. The diagnostic method of claim 6, wherein said determining the probability of the user having a mental disorder based on the result of the tag extraction comprises:
and confirming the probability of the mental diseases of the user according to the extracted open model label and the disease duration label.
8. A computer device, characterized in that it comprises a memory and a processor, in which memory a computer program is stored which, when executed by the processor, implements the method of any of claims 1-7.
9. A non-transitory computer readable storage medium of a computer program, characterized in that the method of any of claims 1-7 is implemented when the computer program is executed by one or more processors.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210785536.3A CN115331805B (en) | 2022-07-04 | 2022-07-04 | Mental disease diagnosis method based on natural language processing and computer equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210785536.3A CN115331805B (en) | 2022-07-04 | 2022-07-04 | Mental disease diagnosis method based on natural language processing and computer equipment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115331805A CN115331805A (en) | 2022-11-11 |
| CN115331805B true CN115331805B (en) | 2023-05-19 |
Family
ID=83916714
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210785536.3A Active CN115331805B (en) | 2022-07-04 | 2022-07-04 | Mental disease diagnosis method based on natural language processing and computer equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115331805B (en) |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107194158A (en) * | 2017-05-04 | 2017-09-22 | 深圳美佳基因科技有限公司 | A kind of disease aided diagnosis method based on image recognition |
Family Cites Families (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108986908B (en) * | 2018-05-31 | 2023-04-18 | 平安医疗科技有限公司 | Method and device for processing inquiry data, computer equipment and storage medium |
| CN109192300A (en) * | 2018-08-17 | 2019-01-11 | 百度在线网络技术（北京）有限公司 | Intelligent way of inquisition, system, computer equipment and storage medium |
| CN109509551A (en) * | 2018-11-01 | 2019-03-22 | 新博卓畅技术（北京）有限公司 | A kind of common disease intelligent diagnosing method and system |
| US11393590B2 (en) * | 2019-04-02 | 2022-07-19 | Kpn Innovations, Llc | Methods and systems for an artificial intelligence alimentary professional support network for vibrant constitutional guidance |
| US20210005316A1 (en) * | 2019-07-03 | 2021-01-07 | Kenneth Neumann | Methods and systems for an artificial intelligence advisory system for textual analysis |
| CN110504028A (en) * | 2019-08-22 | 2019-11-26 | 上海软中信息系统咨询有限公司 | A kind of disease way of inquisition, device, system, computer equipment and storage medium |
| CN111462112A (en) * | 2020-04-23 | 2020-07-28 | 上海联影医疗科技有限公司 | Method and system for judging image quality |
| CN111540440B (en) * | 2020-04-23 | 2021-01-15 | 深圳市镜象科技有限公司 | Psychological examination method, device, equipment and medium based on artificial intelligence |
| CN112582060A (en) * | 2020-12-14 | 2021-03-30 | 首都医科大学 | Multimedia auxiliary inquiry system for depression preliminary screening |
| CN114093520A (en) * | 2021-07-14 | 2022-02-25 | 北京好欣晴移动医疗科技有限公司 | Classification method, device and system for information related to treatment of sleep disorder |
| CN113903431A (en) * | 2021-10-13 | 2022-01-07 | 平安国际智慧城市科技股份有限公司 | Psychological disorder prediction method, device, equipment and storage medium |
| CN113705226B (en) * | 2021-10-29 | 2022-02-11 | 上海柯林布瑞信息技术有限公司 | Medical text entity labeling method and device |
| CN114613489A (en) * | 2022-02-21 | 2022-06-10 | 哈尔滨工业大学（深圳） | Deep reinforcement learning-based inquiry method and system |
-
2022
- 2022-07-04 CN CN202210785536.3A patent/CN115331805B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107194158A (en) * | 2017-05-04 | 2017-09-22 | 深圳美佳基因科技有限公司 | A kind of disease aided diagnosis method based on image recognition |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115331805A (en) | 2022-11-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11120895B2 (en) | Systems and methods for mental health assessment | |
| US20240170109A1 (en) | Systems and methods for mental health assessment | |
| Dharwadkar et al. | A medical chatbot | |
| CN110570941B (en) | System and device for assessing psychological state based on text semantic vector model | |
| CN111259111B (en) | Medical record-based decision-making assisting method and device, electronic equipment and storage medium | |
| Sriram et al. | A chatbot mobile quarantine app for stress relief | |
| Dudzik et al. | Context in human emotion perception for automatic affect detection: A survey of audiovisual databases | |
| Marriwala et al. | A hybrid model for depression detection using deep learning | |
| Fergadiotis et al. | Algorithmic classification of five characteristic types of paraphasias | |
| Smink et al. | Towards text mining therapeutic change: A systematic review of text-based methods for Therapeutic Change Process Research | |
| Baki et al. | A multimodal approach for mania level prediction in bipolar disorder | |
| CN115691786A (en) | Electronic medical record-based ophthalmologic disease information extraction method and auxiliary diagnosis device | |
| Abaho et al. | Assessment of contextualised representations in detecting outcome phrases in clinical trials | |
| Danner et al. | Advancing mental health diagnostics: GPT-based method for depression detection | |
| CN117828355A (en) | Emotion quantitative model training method and emotion quantitative method based on multi-modal information | |
| Avishka et al. | Mobile app to support people with dyslexia and dysgraphia | |
| CN115331805B (en) | Mental disease diagnosis method based on natural language processing and computer equipment | |
| CN116662515A (en) | Search type multi-round dialogue method and device, storage medium and electronic equipment | |
| CN116959754A (en) | Feature extraction method of structured interview recording transcribed text based on intention slots | |
| CN114328864A (en) | Ophthalmic question-answering system based on artificial intelligence and knowledge graph | |
| CN114283912A (en) | Medical record filing method based on RTHD and artificial intelligence and cloud platform system | |
| Vuyyuru et al. | A Transformer-CNN Hybrid Model for Cognitive Behavioral Therapy in Psychological Assessment and Intervention for Enhanced Diagnostic Accuracy and Treatment Efficiency | |
| CN114582449A (en) | Electronic medical record named entity standardization method and system based on XLNet-BiGRU-CRF model | |
| CN113761899A (en) | Medical text generation method, device, equipment and storage medium | |
| Liu et al. | Stone needle: A general multimodal large-scale model framework towards healthcare |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |