Disclosure of Invention
Aiming at the defects in the prior art, the invention aims to provide an intelligent fitting overall solution for a keratoplasty mirror based on artificial intelligence.
The technical purpose of the invention is realized by the following technical scheme:
an intelligent cornea moulding mirror fitting overall solution based on artificial intelligence comprises an input data assembly, an application database assembly, an automatic calling model prediction assembly and a front-end display assembly, wherein a user/doctor inputs visual data of a patient in the input data assembly, the application database assembly automatically inputs and stores the data in an application database, and the model assembly is waited to be called; the automatic calling model prediction component automatically detects that the application database component finishes updating data, then calls the trained model file, and directly predicts the newly stored data; the front-end display component receives the prediction result, namely the film selection parameter result, and displays the result to a software interface for a user/doctor to view.
Further preferably, the system further comprises a calculation result storage component, which stores the result after the prediction calculation, namely, the result is returned to the application data component, a new data result is stored for later use when the model needs to be retrained, and the result data is transmitted to the front-end display component for display.
It is further preferred that the auto-call model prediction component is provided with timing training, i.e. when the new input data reaches a certain amount, the model is necessary to complete new training to obtain better prediction parameters.
More preferably, the automatic calling model prediction module performs the steps of feature screening, model training, and model prediction in sequence after acquiring data.
More preferably, the feature screening specifically performs correlation test on the patient data and the fitting parameter values and distribution test on the data itself, and screens out the influence factors, i.e., the features, having high value.
Further preferably, the model training utilizes the influence factors with high value screened out, assuming as X, and the fitting parameters Y to establish a one-to-one correspondence relationship, and finally obtains an implicit equation: y- > f (X).
In summary, compared with the prior art, the beneficial effects of the invention are as follows:
the method comprises the steps of automatically acquiring selection parameters of the orthokeratology lens by adopting a machine learning algorithm, acquiring rules behind the data in a training process according to the existing data by the algorithm, finding a deterministic relationship between factors influencing the selection parameters of the orthokeratology lens and the selection parameters, and presenting a unified calculation logic.
Detailed Description
The principles and spirit of the present invention will be described with reference to a number of exemplary embodiments. It is understood that these embodiments are given solely for the purpose of enabling those skilled in the art to better understand and to practice the invention, and are not intended to limit the scope of the invention in any way. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art. Reference in the specification to "an embodiment" or "an implementation" may mean either one embodiment or one implementation or some instances of embodiments or implementations.
As will be appreciated by one skilled in the art, embodiments of the present invention may be embodied as a system, apparatus, device, method, or computer program product. Accordingly, the present disclosure may be embodied in the form of: entirely hardware, entirely software (including firmware, resident software, micro-code, etc.), or a combination of hardware and software.
It is to be noted that any number of elements in the figures are provided by way of example and not limitation, and any nomenclature is used for distinction only and not in any limiting sense.
An intelligent cornea moulding mirror fitting overall solution based on artificial intelligence comprises an input data component, an application database component, an automatic calling model prediction component, a calculation result storage component and a front-end display component.
The specific flow is shown in fig. 1, and each component is explained as follows:
1. user input data component: the component sets a page for a front end, and when a user selects an AI-choose interface after logging in a software page, the visual data of the patient, including the length of the eye axis, the naked eye vision, the flat K value, the steep K value and the like, can be input, and the interactive interface can friendly help the user to use the software.
2. Application database component: the component automatically enters data into the application database and saves after the user clicks on the component, waiting for the model component to call.
3. A model prediction component: the component can automatically detect that the application database component finishes updating data, then the component calls out the trained model file and directly predicts the newly stored data; meanwhile, the model component can also be set with timing training, namely when new input data reaches a certain amount, the model needs to complete new training to obtain better prediction parameters and assist doctors in completing the selection of the corneal plastic lens.
4. A result storage component: the component is mainly used for storing the result after model calculation, namely returning the result to the application data component and storing a new data result so as to be used when the model needs to be retrained later. While the component will also pass the result data to the next component.
5. Front end display module: the results of the culling parameters of the results storage component are primarily received and presented to the software interface for viewing by the user.
For the data utilized above, there is actually a orthokeratology mirror fitting table;
field: flat K value, gender, age, steep K value, e value, corneal diameter, naked eye vision value, corneal topography value, fitting parameter value, toric curve, etc.;
the corresponding relation of the fields is as follows: different sample data are corresponding to each other, but it is possible that the fitting parameters are consistent, that is, multiple cases may correspond to one fitting parameter.
After the automatic calling model prediction component acquires data, the steps of feature screening, model training and model prediction are sequentially performed, and data cleaning is performed before feature screening, which is specifically as follows with reference to fig. 2:
step _1. data washing: collected irregular sight data and data under the condition of poor fitting are removed, and meanwhile, part of data needs to be subjected to normalization processing.
Step _2. feature screening: and (4) for the cleaned data, carrying out correlation test and self-distribution test on the data and the fitting parameter values, and screening out influence factors with high value, namely characteristics.
Step _3. model training: establishing a one-to-one corresponding relation by utilizing the screened factors, supposing X, and the fitting parameters Y, and finally obtaining an implicit equation: y- > f (X).
Step _4. model prediction: and obtaining model parameters and files based on the trained model in the last step, so that new data can be predicted, and a new result is obtained to help a user to quickly check and match the selected film.
The above description is intended to be illustrative of the present invention and not to limit the scope of the invention, which is defined by the claims appended hereto.