GEHLOT ANITA (IN)
GUPTA LOVI RAJ (IN)
BUDDHI DHARAM (IN)
DAS PRABIN KUMAR (IN)
ROY BISHWAJYOTI (IN)
| US20210098113A1 | 2021-04-01 | |||
| US20200019823A1 | 2020-01-16 | |||
| CN109903839A | 2019-06-18 |
| Claims: 1. A web based framework for easy and fast artificial intelligence software/model deployment comprising: a. at least one central custom server which is intended to parent the artificial intelligence models/software in it and process the input in the artificial intelligence models/software after checking user’s authentication; b. a deployment of an artificial intelligence models/software for image classification between infected and healthy CT scan images; c. a web application for providing authorized access to the corresponding authorities to accept the CT scan image reports through a user interface and provide the input to the deployed artificial intelligence models/software remotely; and d. an internet connectivity. 2. The system of claim 1, wherein the artificial intelligence models and software are deployed into the central custom server. 3. The system of claim 1, wherein the authorized users receive access to deployed models and software using serialized or non-serialized transaction. 4. The system of claim 1, wherein the web application is an interface between the authorized users and the central custom server. 5. The system of claim 1, wherein the central custom server receives user request through the internet connectivity in the form of an object through an Application Programming Interface (API). 6. The system of claim 1, wherein the user is authenticated by the central custom server, and allows the object to send into the input field feed of artificial intelligence model and software in the central custom server. 7. The system of claim 1, wherein the central custom server after execution of the received feed, sends reply to query to the corresponding user with the output. 8. The system of claim 1, wherein the user’s webpage displays the result for further consideration. 9. A method for user side viewing of web based framework comprising: a. visiting the Uniform Resource Locator link provided to the users; b. creating an account to get an access to an interface; c. assigning credentials to the user after verification by the admin groups; d. login into the account using the unique credentials; e. after login user provides input to the deployed artificial intelligence models/software; f. checking the patient’s list showing on the dashboard; g. adding the new patient by click on the add patient button on top right corner of dashboard; h. entering required details for creating a separate instance which stores, sort and arrange all the information about the patient’s diagnosis; i. adding scan image by click on the add scan button on top right corner of dashboard; j. generating a unique diagnosis ID whenever a new CT scan is added to the patient’s account; and k. displaying the predicted result after uploading a new CT scan report. 10. The method of claim 9, wherein the authorized user is able to delete any record. 11. A method for server side viewing of web based framework comprising: a. receiving the request for authorization; b. extracting the UID and KEY from string; c. checking whether user is authenticated or not; d. allowing the received object after authentication; e. providing an empty access gate; f. pre-processing of received input; g. feeding into the artificial intelligence model/software; h. generating an output after processing the input; i. converting the output into an object; j. finding the query; k. sending reply to query by adding header-footer; 1. receiving an acknowledgement from the web interface; and m. marking task as completed in logs. 12. The method of claim 11, wherein the central custom server rejects the request if user is not authenticated and stores the UID and KEY of that user. |
[001] The present invention relates to a web-based framework, more particularly, to a web-based framework for easy and fast deployment of artificial intelligence model and software to increase the scalability and effectiveness of the AI models and software.
BACKGROUND
[002] COVID-19 is a contagious disease which transmits faster from one person to another, and social distancing is one of the solutions to limit corona virus from spreading. Also, to control the corona virus, it is important to diagnose it in early stages. In case of COVID-19, CT scans are considered as effective way of diagnosis. However, it is a considerable slow process when done manually.
[003] There are some issues related to already existing systems such as extra setup is required for implementation and deployment of any machine learning model; in rural areas, the upgradation of equipment or system specifications is quite difficult; in case of any technical issue or power cut, the system becomes unusable.
[004] Moreover, technical skills are required for operating and using software and physical access is mandatory to monitor lists, status and progress of patients in hospital. The usage of modem age technologies such as Artificial Intelligence has proven the solution to increase the speed for diagnosis and makes the process easy. However, the pre-requisites for usage of AI model in a system is quite adverse and cannot be implement on normal computer systems. Also, it effects the scalability of AI software deployment because neither every hospital around the country has such high specification systems nor every hospital have proper resources for buying high specification systems.
[005] Patent number KR101818074B1, discloses an artificial intelligence based automatic medical diagnosis method and system thereof which uses decision making support system based on artificial intelligence diagnosis to enhance reliability. However, this invention includes user interface which inculcate multi-level interpretation of medical image.
[006] Patent number JP2003116838A, discloses a medical image diagnostic equipment which prevent display processing and reduces user dissatisfaction because of processing delay. However, this invention deals with medical image processing and displaying systems.
[007] Patent number US10499845B2, discloses a method and device for analysing an image which process the image prior to identifying the lesion in the image. However, this invention uses to identify the image of lesion on the skin.
[008] Patent number US2008246617A1, discloses an apparatus, system and method which uses sensor to monitor the reactions of a person in motion and physical and mental characteristics vectors are used to build up database which further executes an algorithm. However, this invention uses self learning algorithm for classification which is different from present invention.
[009] Patent number US2017352158A1, discloses systems, methods for obtaining a medical interpretation for medical images in which images uploaded on the portal are accessed by medical imaging expert. However, this invention requires human intervention for accessing the medical images.
[0010] Patent number US2019189263A1, discloses methods and systems for automatically generation of medical images based of cognitive classification which uses deep learning methodology to classify image studies. However, this invention is different from present invention.
[0011] Patent number EP3629898A1, discloses an automated lesion detection, segmentation and require high resolution images for accurate and timely diagnosis. However, this invention is different from present invention which is a framework. [0012] Patent number US2020126668A1, discloses a method and system for fast access to advanced visualization of medical scans over the internet which displays the requested virtual view to the remote user. However, this invention is having a more latency time than present invention.
[0013] Patent number KR20200015280A, discloses method for predicting pneumonia which uses artificial neural network based on plurality of learning data. However, this invention uses multi configuration input nodes for predicting the probability of pneumonia.
[0014] The above mentioned prior art states that there is a need for a web based framework which increases the scalability and effectiveness of the artificial intelligence models/software by removing physical access barrier and providing user-friendly interface.
[0015] The present invention addresses the above mentioned short comings of the prior art.
SUMMARY
[0016] The summary as given below.
[0017] The present invention is a web-based framework which simplifies the deployment of machine learning models by providing implementation in minimal setups along with some addon features.
[0018] In one implementation, a central custom server provides back up which is intended to parent all the Artificial Intelligence (AI) models and software in it.
[0019] In one implementation, a software that is used by the hospitals is installed/deployed into the central custom server once, after that all the authorised hospitals is given classified access to the software deployed in the central custom server using either serialized or non-serialized transaction access depending on the traffic and other parameters. [0020] In another implementation, a web application is developed as a user interface which provides an interface between authorized users and the central custom server.
[0021] In yet another implementation, a user interface is capable of accepting CT (Computed Tomography) scan image reports and sends it securely to the central custom server through API in the form of an object.
[0022] In another implementation, a user is authenticated by the central custom server which allows the object to get into the input field of the AI model/software that is already hosted in the central custom server.
[0023] In another implementation, after execution of the received feed, the central custom server automatically provides reply to query to corresponding user with the output prediction montage.
[0024] In another implementation, the output is projected into the user’s webpage for further consideration.
[0025] In another implementation, a web application is act as an interface between the authorized users and the central custom server.
[0026] In another implementation, the present invention improves the scalability and effectiveness of the deployed artificial intelligence models.
[0027] In yet another implementation, the web-based framework is implemented on existing systems also that are using in medical field and require no extra setup.
BRIEF DESCRIPTION OF THE DRAWINGS
[0028] The foregoing detailed description of embodiments is better understood when read in conjunction with the attached drawings. For better understanding each component is given a specific number which is further illustrated as reference number for the components used with the figure.
[0029] Figure 1, illustrates the web-based framework, in accordance with an embodiment of the present subject matter.
[0030] Figure 2, flowchart illustrates the user side working, in accordance with an embodiment of the present subject matter. [0031] Figure 3, flowchart illustrates the server side working, in accordance with an embodiment of the present subject matter.
[0032] Figure 4, illustrates the new user registration login page, in accordance with an embodiment of the present subject matter.
[0033] Figure 5, illustrates the existing user login page, in accordance with an embodiment of the present subject matter.
[0034] Figure 6, illustrates the patient list and details on dashboard after login, in accordance with an embodiment of the present subject matter.
[0035] Figure 7, illustrates the add patient on dashboard after login, in accordance with an embodiment of the present subject matter.
[0036] Figure 8, illustrates the individual patient details on dashboard, in accordance with an embodiment of the present subject matter.
[0037] Figure 9, illustrates the addition of new CT scan on dashboard, in accordance with an embodiment of the present subject matter.
[0038] Figure 10 (a), 10 (b) illustrates diagnosis results on dashboard, in accordance with an embodiment of the present subject matter.
[0039] The figures depict an embodiment of the present disclosure for the purpose of illustration and understanding only.
DETAILED DESCRIPTION
[0040] Some embodiments of this disclosure, illustrating all its features, will now be discussed in detail.
[0041] In one implementation, the present invention increases the scalability and effectiveness of the Artificial Intelligence (AI) models and software to a great extent.
[0042] The disclosure herein is a web based framework which simplifies the deployment of machine learning models and implement in minimal setups. [0043] In one embodiment, a central custom server which stores all the artificial intelligence models/software in it.
[0044] In one embodiment, a software is installed or deployed in the central custom server and after that all the authorized hospitals receive the classified access to deployed software using either serialized or non-serialized transaction depending on the amount and priority of the incoming task request.
[0045] In one embodiment, the framework initiates its computing methodology as a serial which attend the task request one after another, and in a non-serialized or parallel methodology multi-thread the resources of the server which perform multiple tasks simultaneously.
[0046] In one embodiment, a web application is developed for the authorized users which provides a simple User Interface (UI) to accept CT scan image reports through just few clicks and sends it securely to the central access gate through the internet connectivity in the form of an object through an Application Programming Interface (API).
[0047] In another embodiment, authorized access is provided to the corresponding authorities using which they provide the input to the deployed AI models/software remotely.
[0048] In another embodiment, the authentication of user is verified by the central custom server, and allows the object to get into the input field feed of artificial intelligence model/software which is already hosted in the central custom server.
[0049] In another embodiment, the central custom server automatically replies the query with the output prediction montage to the corresponding user after the execution of the received feed.
[0050] In another embodiment, the output is projected to the user’s webpage for further consideration.
[0051] In another embodiment, the web application is act as interface between the authorized users and the central custom server. [0052] In one embodiment, new user is registered by creating account using the URL (Uniform Resource Locator) link provided to them and after successful registration, admin groups verify the details provided by user.
[0053] In another embodiment, once admin groups verify the details, the user gets a status of verified authorized user and account is activated automatically.
[0054] In another embodiment, all the verified authorized users are able to login into their account using the unique credentials through interface.
[0055] In another embodiment, after successfully login into the account, a dashboard is opened which shows the list of the patients that are being diagnosed using the same system before.
[0056] In yet another embodiment, user is able to view details related to patient and diagnosis result after clicking on that patient’s name showing on the dashboard.
[0057] In yet another embodiment, an add patient button on top right comer on the dashboard, which is used to add patient’s details.
[0058] In another embodiment, an interface is opened after clicking on patient’s name displayed on the dashboard which further contains the unique diagnosis ID that is automatically assigned whenever a new CT scan image is added to that patient’ s account.
[0059] In yet another embodiment, a unique diagnosis ID helps to keep track of reports and results of individual patients.
[0060] In another embodiment, new CT scan is uploaded by the authorized user for diagnosis purpose.
[0061] In another embodiment, any unnecessary record is deleted by the authorized user. [0062] In another embodiment, after uploading a new CT scan report, the authorized user receives the predicted result from the deployed artificial intelligence model/software.
[0063] In another embodiment, the number of output parameters directly depends on the artificial intelligence model deployed in the central custom server.
[0064] In another embodiment, for testing the framework, a basic
Artificial Intelligence (AI) model trained with CNN is deployed to classify between infected and healthy CT scan image.
[0065] In yet another embodiment, AI model is trained with any appropriate algorithm, model or network.
[0066] In yet another embodiment, the image is transferred using
Advanced Encryption Standard (AES) encryption and maintain the user’s privacy.
[0067] In another embodiment, the web based CT scan diagnostic application is developed and deployed in the central server. Instead of deploying AI models in the system of each hospital, it is deployed in the central server and remote access is provided to the authorized users/hospitals where they are registered through the web user interface and add patient details. After adding patient details, users are able to upload their CT scan which is further processed in the central server and diagnosis result is reflected on the user interface of the application. Moreover, the application provides the ability to monitor the past results of the corresponding patient also.
[0068] In an exemplary embodiment, a web application with simple user interface accepts the CT scan image reports and sends securely to the central custom server in the form of an object through an Application Programming Interface. The user is verified through the central custom server and the object is sent into the input field feed of the artificial intelligence model/software in the central custom server and after execution of the received feed, the server automatically send reply to the query with the output prediction montage to the corresponding user which further displays on the user’s webpage for further consideration.
[0069] Referring to figure 1, illustrates the web based framework in which a web application is act as an interface to accept CT scan image reports and sends securely to the central custom server in the form of an object through an API (Application Programming Interface).
[0070] Again, referring to figure 1, after verification by the central custom server, the object is sent into the input field feed of the artificial intelligence model/software in the central custom server and after execution of the received feed, the server automatically replies that query with the output prediction montage to the corresponding user which is displayed on the user’s webpage for further consideration.
[0071] Referring to figure 2, flowchart illustrates the steps that are used on user side in which user visit the Uniform Resource Locator (URL) provided to them for login as a new user or existing user. If user is new then user is registered by providing required information on the URL link and create their account. All the existing users uses their unique credentials assigned to them for login into their accounts.
[0072] Again, referring to figure 2, after successfully logged in into the account, details of patients is checked on the dashboard. Also, user is able to add new patient as well as upload CT scan image report through the add scan button on the dashboard and results are displayed on the webpage.
[0073] Referring to figure 3, flowchart illustrates the steps that are used on server side in which an image diagnosis request is received by a server and UID and KEY information is extracted from string. If user is found authenticated, the received object is allowed to provide an empty access gate for pre processing the image. Further, the image is feed into the artificial intelligence model/software which converts an output into an object to find the query and reply is sent by adding header and footer, and acknowledgement is received from the web interface. The task is marked as completed in logs.
[0074] Again, referring to figure 3, an image diagnosis request is received by a server and UID and KEY information is extracted from string. If user is found not authenticated one, the request is rejected by the server and UID and KEY is stored in the server.
[0075] Referring to figure 4, illustrates the new user registration login page which appears when a user is registered and create their account using the URL link provided to them.
[0076] Referring to figure 5, illustrates the existing user login page which is used by verified authorized users to login into their accounts using the unique credentials.
[0077] Referring to figure 6, illustrates the patient list and details on dashboard after successfully logged in into the dashboard. The list of patients is showed on the dashboard who are diagnosed using the system before and details are available there.
[0078] Referring to figure 7, illustrates the add patient on dashboard after clicking on the add patient button on top right corner as shown in figure 6. By entering details, a separate instance is created which is used to store, sort, and arrange all the information about that patient’s diagnosis for future consideration.
[0079] Referring to figure 8, illustrates the individual patient details on dashboard with unique ID that is automatically assign whenever a new CT scan is added to patient’s account. This helps the user to keep track and sort the reports and results of individual patients.
[0080] Again, referring to figure 8, the authorized user is able to add new
CT scan for diagnosis and delete any record of no use.
[0081] Referring to figure 9, illustrates addition of new CT scan on dashboard through clicking add scan button on top right corner as shown in figure 8. This allows the user to upload the CT scan report after confirm the patient’s name.
[0082] Referring to figure 10 (a), 10 (b) illustrates the diagnosis results from the deployed artificial intelligence model/software after uploading a new CT scan report.
[0083] Some of the embodiments may be further upgraded upon the study performed further.
Next Patent: SYSTEM AND PROCESS TO HEAT PREFORMED PLASTIC PIPES