Disclosure of Invention
In view of the above technical problems, the present invention provides an oral health assessment method and system, a computer readable storage medium, a computer program product and an electronic device for implementing the method.
In a first aspect of the invention, a method for oral health assessment is presented, said method being applied to an oral health diagnostic teaching process,
The method comprises the following steps:
s110, acquiring inspection data of a target oral cavity;
S120, determining an auxiliary database and an auxiliary recognition model based on the test data;
s130, inputting the test data into the auxiliary identification model, and outputting an auxiliary diagnosis result;
Wherein, the inspection data of the target oral cavity obtained in the step S110 includes oral cavity digital inspection image data and/or oral cavity microbiome sampling index data.
The target oral cavity test data obtained in the step S110 is anonymized oral cavity test data;
The step S120 further includes:
prompting a user to determine a target age bracket of a target patient corresponding to a target oral cavity according to the oral cavity test data;
and determining an auxiliary database from a plurality of candidate databases based on the target age bracket and the inspection data, and determining an auxiliary recognition model from a plurality of candidate recognition models, wherein the plurality of candidate recognition models comprise an index data self-learning diagnosis model and an image data recognition diagnosis model.
The plurality of candidate databases includes a plurality of oral health databases of different ages and different types.
The plurality of candidate databases comprise an age group-microbiome sampling index database and an age group-oral digital inspection image database;
The age group-microbiome sampling index database stores a data pair of < microbiome sampling index and oral health evaluation result >, and the age group-oral digital inspection image database stores a data pair of < inspection image characteristic value and oral health evaluation result >;
After the step S130, the method further includes:
when the difference degree between the auxiliary diagnosis result and the original diagnosis result of the target oral cavity is larger than a preset threshold, retraining the auxiliary identification model by using the test data;
The retraining includes:
and taking the test data of the target oral cavity as an updated training sample, taking the test data of the auxiliary database corresponding to the auxiliary diagnosis result as an countermeasure sample, and retraining the auxiliary identification model.
The step S130 specifically includes:
inputting oral digitized test image data into the image data recognition diagnostic model, the image data recognition diagnostic model deriving a first auxiliary diagnostic result based on the age-oral digitized test image database;
And/or the number of the groups of groups,
And inputting the oral microbiome sampling index data into the index data self-learning diagnosis model, wherein the index data self-learning diagnosis model obtains a second auxiliary diagnosis result based on the age group-microbiome sampling index database.
After the first auxiliary diagnostic result and/or the second auxiliary diagnostic result is obtained in the step S130, the method includes the steps of:
S140, acquiring an original diagnosis result of the target patient;
S150, judging whether the first auxiliary diagnosis result and/or the second auxiliary diagnosis result are normal or not based on the original diagnosis result of the target patient;
and when the first auxiliary diagnosis result and/or the second auxiliary diagnosis result are abnormal, updating the auxiliary database based on the inspection data of the target oral cavity, and retraining the auxiliary identification model based on the updated auxiliary database.
Retraining the auxiliary recognition model based on the updated auxiliary database specifically comprises the following steps:
and taking the test data of the target oral cavity as an updated training sample, taking the test data of the auxiliary database corresponding to the first auxiliary diagnosis result and/or the second auxiliary diagnosis result as an countermeasure sample, and retraining the auxiliary identification model.
The oral health assessment method can be realized through various forms of electronic equipment and automation through computer program instructions, wherein the computer program instructions can be stored in different forms of storage media and loaded into the computer electronic equipment for execution.
Thus, in a second aspect of the invention, there is also provided a computer readable storage medium storing computer instructions that, when executed on an electronic device, cause the electronic device to perform the aforementioned oral health assessment method.
In a third aspect of the present invention, a computer device is also presented, the computer device comprising a processor and a memory, the memory for storing instructions, the processor for invoking the instructions in the memory, causing the computer device to perform the aforementioned oral health assessment method.
In a fourth aspect of the invention, a computer program product is also presented, the product comprising a computer program, which, when executed, implements the aforementioned oral health assessment method.
To achieve the foregoing oral health assessment method of the first aspect, in a fifth aspect of the present invention, there is also provided an oral health assessment system comprising:
The system comprises an inspection data acquisition unit, a control unit and a control unit, wherein the inspection data acquisition unit is used for acquiring inspection data of a target oral cavity, and the inspection data of the target oral cavity comprises oral cavity digital inspection image data and/or oral cavity microbiome sampling index data;
an auxiliary unit for determining an auxiliary database and an auxiliary recognition model based on the inspection data;
the diagnosis unit is used for inputting the inspection data into the auxiliary identification model and outputting an auxiliary diagnosis result;
the retraining unit is used for retraining the auxiliary identification model by using the test data when the difference degree between the auxiliary diagnosis result and the original diagnosis result of the target oral cavity is larger than a preset threshold;
The retraining includes:
and taking the test data of the target oral cavity as an updated training sample, taking the test data of the auxiliary database corresponding to the auxiliary diagnosis result as an countermeasure sample, and retraining the auxiliary identification model.
The system further comprises:
And the auxiliary diagnosis result sending unit is used for sending the auxiliary diagnosis result to a medical care end when the original diagnosis result does not exist in the target oral cavity, and the medical care end judges the health state of the target oral cavity based on the auxiliary diagnosis result.
The evaluation system is applied to an oral health diagnosis teaching process, and the target oral cavity test data acquired by the test data acquisition unit are anonymized oral cavity test data;
The auxiliary unit determines an auxiliary database and an auxiliary identification model based on the inspection data, and specifically comprises the following steps:
prompting a user to determine a target age bracket of a target patient corresponding to a target oral cavity according to the oral cavity test data;
and determining an auxiliary database from a plurality of candidate databases based on the target age bracket and the inspection data, and determining an auxiliary recognition model from a plurality of candidate recognition models, wherein the plurality of candidate recognition models comprise an index data self-learning diagnosis model and an image data recognition diagnosis model.
According to the technical scheme, the oral health assessment based on the artificial intelligent auxiliary model can be realized, meanwhile, the defects of artificial intelligent diagnosis can be contrasted and displayed in teaching, and the database and the auxiliary model are continuously retrained in the diagnosis process, so that people and machines supplement each other to jointly promote the oral health assessment effect.
Further advantages of the invention will be further elaborated in the description section of the embodiments in connection with the drawings.
Detailed Description
It should be noted at first that the embodiments of the oral health assessment method mentioned in this section may be implemented by a computer program on an electronic device, a system configured with a memory and a processor, which may be in the form of a physical machine, a virtual machine, a server, a cluster or any combination thereof.
Preferably, the specific form of the electronic equipment can also be a man-machine interaction terminal, and the man-machine interaction terminal can be a desktop terminal with a man-machine interaction interface, an intelligent handheld terminal, a mobile terminal and the like.
Fig. 1 is a schematic flow chart of a method for evaluating oral health according to an embodiment of the present invention.
The method shown in fig. 1 includes steps S110-S130, each of which is implemented as follows:
s110, acquiring inspection data of a target oral cavity;
S120, determining an auxiliary database and an auxiliary recognition model based on the test data;
S130, inputting the test data into the auxiliary identification model, and outputting an auxiliary diagnosis result.
When the embodiment of fig. 1 is embodied for oral health assessment, "auxiliary diagnostic results" should be understood as an "intermediate" result, rather than a final result, because the embodiment of fig. 1 may be understood as an auxiliary health assessment result obtained by assessing the health status of the target oral cavity based on existing test data using computer-aided techniques, but it should be understood by those skilled in the art that the final health assessment needs to be obtained by a doctor based on clinical experience in combination with the above-described auxiliary diagnostic results.
However, due to the convenience and popularity of the A I model, some oral patients, subjects about the oral cavity (teaching) to be trained (trainees, students, etc.), and even actual clinical staff, rely excessively on the existing A I auxiliary diagnosis technology, cannot realize the defect of the A I model, and their own subjective motility is lost, thus reducing the accuracy of oral health assessment.
Therefore, the first technical problem to be solved in the technical scheme of the application is to show the defects of A I models to related users (training staff (students, etc.) to be (teaching) in the oral relevant discipline and actual clinical staff) through comparison, and improve the subjective motility of the users in the A I auxiliary diagnosis process.
At this time, as a first specific application and technical scheme of the present invention, the method is applied to an oral health diagnosis teaching process, the target oral test data obtained in the step S110 is anonymized oral test data, and the anonymization process at least includes eliminating data related to character features in the target oral test data, including name, age, gender, blood type, telephone number, identification number, etc.
Specifically, referring to fig. 2 on the basis of fig. 1, fig. 2 shows a schematic flow chart of a main body of an oral health assessment method according to still another preferred embodiment of the present invention. The preferred embodiment of fig. 2 is to continue to perform the following steps S140 and S150 after "output auxiliary diagnosis result" in step S130 of the embodiment of fig. 1:
S140, acquiring an original diagnosis result of the target patient;
And S150, when the auxiliary diagnosis result is judged to be abnormal based on the original diagnosis result of the target patient, retraining the auxiliary identification model by using the test data.
It can be seen that, in this embodiment, the method is applied to a teaching diagnosis process, the target oral test data obtained in step S110 is anonymized oral test data, which is actually derived from actual clinical practice, such as a certain oral test data A0 of patient a, for which a professional doctor has given a professional diagnosis conclusion B;
However, the test number A0 is input to the auxiliary recognition model, and the output auxiliary diagnosis result is A1;
and when the difference degree between the auxiliary diagnosis result A1 and the original diagnosis result B of the target oral cavity is larger than a preset threshold value, determining that the auxiliary diagnosis result is abnormal.
When the method is specifically applied to the teaching process of oral disease diagnosis, the difference part of the auxiliary diagnosis result and the original diagnosis result can be compared and displayed to training staff (students, students and the like) to be (taught) and actual clinical staff, so that the defect of A I model is displayed, and the defect of insufficient reliability of the diagnosis result of the A I model of related staff is prompted.
Of course, related personnel are required to be reminded, and A I models are not enough, but are a continuous learning and updating process, so that oral health evaluation effects can be improved jointly through man-machine supplement.
For this purpose, the auxiliary recognition model is then retrained using the test data.
Specifically, the retraining includes:
and taking the test data of the target oral cavity as an updated training sample, taking the test data of the auxiliary database corresponding to the auxiliary diagnosis result as an countermeasure sample, and retraining the auxiliary identification model.
Next, with the embodiment of fig. 3, the flow control process of data related to the oral health assessment process is further described. It can be seen that in the following embodiment, the oral health assessment method provided by the application not only utilizes A I models, but also interactively introduces the database for user operation selection matching, so as to avoid the defect caused by excessively relying on A I technology.
Specifically, as can be seen in fig. 3, the inspection data of the target oral cavity obtained in step S110 includes digital inspection image data of the oral cavity and/or sampling index data of the oral microbiome.
As specific examples, the oral digitized test image data includes oral imaging test results such as root tip slices, curved body slices, head shadow measurement slices, cone beam CT, and the like, as well as other oral digitized images and digitized images.
The digital image captures and records every corner and detail of the oral cavity through the forms of high-definition two-dimensional digital photo, precise and fine digital impression, three-dimensional surface scanning data and the like
The images not only provide abundant information about the oral condition of the patient, but also provide reliable visual basis for diagnosis and treatment planning, the two-dimensional digital photo provides a way for a doctor to rapidly examine the current oral health condition of the patient by virtue of visual and convenient characteristics, the digital impression converts the physical form of the traditional impression into digital information, the efficiency of storage, replication and analysis is greatly improved, meanwhile, the requirement of physical storage space is reduced, and the three-dimensional facial sweep data show the three-dimensional structure of the oral cavity in the form of a three-dimensional model, so that the diagnosis and treatment planning can achieve unprecedented accuracy.
Oral digitized imaging techniques are presented in a variety of forms, including high resolution digitized X-ray images, detailed well-defined Cone Beam CT (CBCT) images, comprehensive in-depth conventional CT images, MR I images of fine soft tissue presentation, and dynamic captured ultrasound images, among others.
The oral microbiome sampling index data is mainly embodied by information in a text and table structural form. Bacterial diseases of the oral cavity (such as caries and periodontal disease) are closely related to overgrowth of resident bacteria in the oral cavity under pathogenic conditions, which leads to micro-ecological unbalance of the oral cavity and further opportunity for pathopoiesia. Therefore, the relevant indexes of the bacterial genus which are obviously related to the severity of gingivitis can be used for diagnosing oral health indexes, and particularly can be obtained by preprocessing after oral thallus is sampled to obtain an oral microbiome sampling index data report.
The step S120 further includes:
prompting a user to determine a target age bracket of a target patient corresponding to a target oral cavity according to the oral cavity test data;
It can be seen that in step S120, the method needs to actively interact with the user, and prompts the user to determine the target age group of the target patient corresponding to the target oral cavity according to the oral cavity test data.
The target age group may be a generalized age segment of children, young, middle-aged, or elderly, or may be a specific age segment of 0-3 years, 3-6 years, 6-12 years, 18-30 years. Of course, other age group criteria are also possible.
This is done because the difference in eating and living habits (such as smoking) of different ages causes the feature points of the oral examination data to be different, and if a universal database of all ages is uniformly adopted to execute the subsequent A I model diagnosis, the accuracy is obviously lower.
Thus, multiple candidate databases may be pre-constructed, including multiple oral health databases of different ages and different types.
Specifically, the plurality of candidate databases comprise an age group-microbiome sampling index database and an age group-oral digital inspection image database;
The age group-microbiome sampling index database stores a data pair of < microbiome sampling index and oral health evaluation result >, and the age group-oral digital inspection image database stores a data pair of < inspection image characteristic value and oral health evaluation result >;
for this reason, in order to avoid the defect of A I model diagnosis, in step S120, a professional' S subjective activity needs to be exerted, and the user is prompted to determine the target age bracket of the target patient corresponding to the target oral cavity according to the oral cavity test data.
Further, based on the target age group and the inspection data, an auxiliary database is determined from a plurality of candidate databases, and an auxiliary recognition model is determined from a plurality of candidate recognition models including an index data self-learning diagnosis model and an image data recognition diagnosis model.
The step S130 specifically includes:
inputting oral digitized test image data into the image data recognition diagnostic model, the image data recognition diagnostic model deriving a first auxiliary diagnostic result based on the age-oral digitized test image database;
And/or the number of the groups of groups,
And inputting the oral microbiome sampling index data into the index data self-learning diagnosis model, wherein the index data self-learning diagnosis model obtains a second auxiliary diagnosis result based on the age group-microbiome sampling index database.
The index data self-learning diagnostic model is a pre-trained index diagnostic model, and the basic principle of the diagnostic model is that indexes of bacteria which are obviously related to the severity of oral cavity (gingivitis) are determined, and an oral cavity (gum) severity diagnostic model based on microorganisms is constructed according to the indexes.
The image data identification diagnosis model is also a pre-trained image diagnosis model, and the basic principle of the image diagnosis model is that based on the technologies of deep learning, neural network, data mining and the like, the input digital inspection image data of the oral cavity is preprocessed, then feature extraction and feature fusion are carried out, classification is carried out, a plurality of features to be identified are obtained, and based on the association relation between the features to be identified and abnormal indexes of the oral cavity, the oral cavity health state is determined.
It should be noted that, both the index diagnosis model and the image diagnosis model belong to the mature prior art in the field, and therefore, are not the improvement points of the present invention, and this embodiment is not specifically developed. However, the A I diagnostic models (including index diagnostic models and image diagnostic models) mentioned in the prior art are used for executing the subsequent A I model diagnosis by uniformly adopting a universal database of all ages, the accuracy is obviously lower, and the diagnostic process is not assisted by professional staff, so that the defect of the A I model diagnostic result is covered, and the interactivity between A I and people is also not facilitated.
The invention focuses on introducing interactive operation of professional staff in the A I model diagnosis process to avoid the defect of auxiliary diagnosis of the A I model, so that the defect of artificial intelligent diagnosis can be contrasted and displayed in teaching, and the database and the auxiliary model are continuously retrained in the diagnosis process, so that the human-computer interaction and the oral health assessment effect are jointly improved.
After the first auxiliary diagnostic result and/or the second auxiliary diagnostic result is obtained in the step S130, the method includes the steps of:
S140, acquiring an original diagnosis result of the target patient;
S150, judging whether the first auxiliary diagnosis result and/or the second auxiliary diagnosis result are normal or not based on the original diagnosis result of the target patient;
and when the first auxiliary diagnosis result and/or the second auxiliary diagnosis result are abnormal, updating the auxiliary database based on the inspection data of the target oral cavity, and retraining the auxiliary identification model based on the updated auxiliary database.
Retraining the auxiliary recognition model based on the updated auxiliary database specifically comprises the following steps:
and taking the test data of the target oral cavity as an updated training sample, taking the test data of the auxiliary database corresponding to the first auxiliary diagnosis result and/or the second auxiliary diagnosis result as an countermeasure sample, and retraining the auxiliary identification model.
After describing relevant details of the method embodiments of fig. 1-3, fig. 4-5 are simplified to illustrate functional block diagrams of an oral health assessment system in accordance with two different embodiments, respectively.
In fig. 4, corresponding to the method embodiment of fig. 2, an oral health assessment system is shown, the assessment system comprising:
The system comprises an inspection data acquisition unit, a control unit and a control unit, wherein the inspection data acquisition unit is used for acquiring inspection data of a target oral cavity, and the inspection data of the target oral cavity comprises oral cavity digital inspection image data and/or oral cavity microbiome sampling index data;
an auxiliary unit for determining an auxiliary database and an auxiliary recognition model based on the inspection data;
the diagnosis unit is used for inputting the inspection data into the auxiliary identification model and outputting an auxiliary diagnosis result;
the retraining unit is used for retraining the auxiliary identification model by using the test data when the difference degree between the auxiliary diagnosis result and the original diagnosis result of the target oral cavity is larger than a preset threshold;
The retraining includes:
and taking the test data of the target oral cavity as an updated training sample, taking the test data of the auxiliary database corresponding to the auxiliary diagnosis result as an countermeasure sample, and retraining the auxiliary identification model.
On the basis of fig. 4, the seed oral health evaluation system shown in fig. 5 further includes:
And the auxiliary diagnosis result sending unit is used for sending the auxiliary diagnosis result to a medical care end when the original diagnosis result does not exist in the target oral cavity, and the medical care end judges the health state of the target oral cavity based on the auxiliary diagnosis result.
It should be understood that, herein, the "medical care end determines the health status of the target oral cavity based on the auxiliary diagnosis result" means that the "auxiliary diagnosis result" is only an "intermediate" result, and not a final result, and the final health assessment needs to be obtained by a doctor (medical care end) in combination with the above auxiliary diagnosis result based on clinical experience, thereby finally determining the health status of the target oral cavity.
Preferably, when the system is specifically applied to an oral medicine teaching process, the target oral cavity test data acquired by the test data acquisition unit is anonymized oral cavity test data;
The auxiliary unit determines an auxiliary database and an auxiliary identification model based on the inspection data, and specifically comprises the following steps:
prompting a user to determine a target age bracket of a target patient corresponding to a target oral cavity according to the oral cavity test data;
and determining an auxiliary database from a plurality of candidate databases based on the target age bracket and the inspection data, and determining an auxiliary recognition model from a plurality of candidate recognition models, wherein the plurality of candidate recognition models comprise an index data self-learning diagnosis model and an image data recognition diagnosis model.
Preferably, although not shown, the system further includes a comparison display unit, configured to compare and display the difference between the auxiliary diagnostic result and the original diagnostic result of the target oral cavity to a training person (a student, etc.) to be (teaching) and an actual clinical staff when the difference between the auxiliary diagnostic result and the original diagnostic result of the target oral cavity is greater than a preset threshold, so as to display the defect of A I model, and prompt that the reliability of the diagnostic result of the related person A I model is insufficient.
Other techniques, principles, algorithms or models of the application not specifically developed may be found in the prior art.
Compared with the prior art, the outstanding improvement and beneficial effects of the technical scheme of the application at least comprise:
(1) The technical scheme of the invention can realize the oral health evaluation based on the artificial intelligent auxiliary model and simultaneously can contrast and display the defects of artificial intelligent diagnosis in teaching.
(2) In order to avoid the defect of A I model diagnosis, the technical scheme of the application needs to exert the subjective activity of a professional and prompt a user to determine the target age bracket of a target patient corresponding to the target oral cavity according to the oral cavity test data, in the process, the user can determine a target age bracket according to clinical experience by combining relevant characteristic points of the oral cavity test data, and a plurality of constructed candidate databases comprising a plurality of oral health databases of different age brackets and different types can improve the accuracy compared with the prior art that the training database and the interaction database are all universal databases of all age brackets to execute the subsequent A I model diagnosis;
(3) The database and the auxiliary model are continuously retrained in the diagnosis process, so that the human and the machine supplement each other to jointly promote the oral health evaluation effect.
In the foregoing description, the invention provides several embodiments, each of which may constitute a separate technical solution and may contribute to the prior art and solve the corresponding technical problems. It should be noted that different embodiments may be combined with each other without violating logic, and that each embodiment may solve at least one technical problem, but that each individual embodiment is not required to solve multiple or all technical problems.
Meanwhile, in each specific embodiment of the application, when related processing is required to be performed according to the user information, the user behavior data, the user history data, the user position information and other data related to the user identity or the characteristics, the permission or the consent of the user is obtained first, and the collection, the use, the processing and the like of the data accord with related laws and regulations and standards. In addition, when the embodiment of the application needs to acquire the sensitive personal information of the user, the independent permission or independent consent of the user is acquired through popup or jump to a confirmation page and the like, and after the independent permission or independent consent of the user is definitely acquired, the necessary relevant data of the user for enabling the embodiment of the application to normally operate is acquired.
The foregoing description of implementations of the present disclosure has been provided for illustrative purposes, is not exhaustive, and is not limited to the implementations disclosed. Many modifications and variations will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the various implementations described. The terminology used herein was chosen in order to best explain the principles of each implementation, the practical application, or the improvement of technology in the marketplace, or to enable others of ordinary skill in the art to understand each implementation disclosed herein.