Method and system for dental treatment planning - Patent application

Abstract

The present invention relates generally to methods and systems that can be used to analyze an individual's oral condition. These methods and systems can be used to identify oral conditions or diseases and provide a personalized, comprehensive treatment care plan. More specifically, the present invention relates to methods and systems that use artificial intelligence capabilities to generate a treatment plan for an identified oral condition or disease.

Description

[Technical field]

[0001] The present invention relates generally to methods and systems that can be used to analyze an individual's oral condition. These methods and systems can be used to identify oral conditions or diseases and provide a personalized, comprehensive treatment care plan. More specifically, the field of the invention relates to methods and systems that use artificial intelligence capabilities to generate a treatment plan for an identified oral condition or disease. [Background technology]

[0002] Dental radiographs and digital images are widely used by dentists to detect pathologies and monitor the progress of treatment of an individual's dental condition. Digital dental radiographs and images have the advantages of immediate availability, low radiation dose, and the ability to enhance and reconstruct images, and the use of digital dental images in clinical diagnosis has increased significantly over the past decade.

[0003] However, despite the rapid spread of digital dental imaging, radiographic misdiagnosis is likely to occur even among experienced dentists. Moreover, intra- and inter-examiner agreement between dentists is very low.

[0004] Dental caries, periodontal disease, and missing teeth are some of the most common oral diseases and conditions that affect an individual's quality of life and cause significant health problems. Effective management of an oral condition or disease is highly dependent on decision-making reasoning, which requires a lot of knowledge, such as risk factors, treatment plans, treatment outcomes, and incidence and progression rates. All of this is left to the professional judgment of the healthcare professional who examines the oral condition of an individual. However, the ability of healthcare professionals to always make the right decision is limited by their cognitive functions, such as reasoning ability and memory capacity. These factors lead to discrepancies in decisions made by different healthcare professionals, which often leads to overtreatment or undertreatment, and therefore poses serious problems.

[0005] There is a need to limit uncertainty in inference and variance in dentists' decision-making through the development of clinical decision support systems (CDSS) for the management of oral conditions and diseases, such as dental caries, periodontal disease, and missing teeth.

[0006] WO201994504A1 discloses a machine learning system and method for analyzing and detecting features of dental radiographs, i.e. dental lesions. In particular, object detection is performed based on real face images focused on the oral cavity region, a bounding box is drawn around the oral cavity region, and then a UNet segmentation model determines for each pixel whether it is a tooth or not. This disclosure is not based on using X-ray images to simultaneously detect both tooth numbers and their positions.

[0007] CN109528323A discloses an artificial intelligence based orthodontic method and device. In particular, a Generative Adversarial Network (GAN) model helps to create a 3D reconstruction of an individual's set of teeth. The GAN model helps to demonstrate the impact of orthodontic problems. A CT scan is used to create temporal slices of the oral cavity, unlike an x-ray, which has no temporal component.

[0008] CN105260598A discloses a dental diagnosis and treatment decision support system and a decision-making method for obtaining the most similar recommended cases in diagnosing a disease and automatically providing decision-making data. In particular, a three-stage screening method is adopted to obtain the most similar cases in diagnosis in order to speed up the screening speed. The decision support system utilizes case studies, converts them into numerical representations, and determines matches based on Euclidean distance to the cases, in contrast to those based on medical directives, which are rule-based and more decision-based in nature.

[0009] In these regards, there is a need to provide an automated system and method that employs artificial intelligence capabilities that can analyze dental charts, oral conditions, and generate a comprehensive treatment plan as needed. Summary of the Invention [Problem to be solved by the invention]

[0010] It is a first object of the present invention to provide a computer-implemented method for generating a dental treatment plan for a patient, the method may include the following steps: analyzing patient data including dental images of the patient using an AI model and generating AI predictions regarding detection, numbering, and dental problems of the patient's teeth; populating a dental chart based on the AI ​​predictions and input received from one or more users; generating a completed dental questionnaire using a clinical decision support system (CDSS) based on the populated dental chart and input received from the one or more users; generating a final treatment plan based on the completed dental questionnaire; and displaying the final treatment plan to one or more users.

[0011] Another object is to provide a computer system including a processor configured to perform the methods disclosed herein.

[0012] A further object is to provide a computer readable storage medium containing instructions which, when executed by a computer, cause the computer to carry out the methods disclosed herein.

[0013] Another object is to provide a system for generating a dental treatment plan for a patient, the system may include: an input unit for receiving dental images; a computer readable storage medium configured to store instructions defining an AI model; a first server running a clinical decision support system (CDSS); a second server running the instructions defining the AI ​​model; where the second server is configured to perform operations including: using the AI ​​model to generate AI predictions regarding tooth detection, numbering, and dental issues based on the patient's dental images; and populating a dental chart based on the AI ​​predictions and inputs received from one or more users, where the first server is configured to perform operations including: generating a completed dental questionnaire based on the dental chart and inputs received from the one or more users using the CDSS; and generating a final treatment plan based on the completed dental questionnaire using the CDSS; and an output unit configured to communicate the dental chart, questionnaire, and final treatment plan to a user.

[0014] Further objects of the present invention will become apparent as the description proceeds.

[0015] To the accomplishment of the foregoing and related ends, the invention may be embodied in the form illustrated in the accompanying drawings, however, it is to be noted that the drawings are illustrative only and that changes may be made in the specific constructions illustrated and described within the scope of the appended claims. [Means for solving the problem]

[0016] definition The terms used herein generally have their usual meaning in the art, the context of the disclosure, and the specific context in which each term is used.Specific terms used to describe the disclosure are discussed below or elsewhere in this specification to provide additional guidance to practitioners regarding the description of the disclosure.It will be understood that the same thing can be expressed in multiple ways.

[0017] To that end, alternative language and synonyms may be used for any one or more of the terms discussed herein. Also, no special meaning should be attached to whether a term is detailed or discussed herein. Synonyms for certain terms are provided. The description of one or more synonyms does not exclude the use of other synonyms. The use of examples, including examples of terms discussed herein, anywhere in this specification is for illustrative purposes only and is not intended to further limit the scope and meaning of the disclosure or the terms exemplified. Similarly, the disclosure is not limited to the various embodiments described herein.

[0018] As used herein, the following words and terms have the meanings set forth below.

[0019] The term "oral condition" or "oral disease" includes, but is not limited to, dental caries, periodontal disease, missing teeth, crowns, bridges, implants, fillings, onlays, inlays, veneers, failed or defective restorations (fillings, crowns, bridges, veneers, inlays, onlays), root canal treatment, posts and cores, impacted teeth, partially erupted teeth, unerupted teeth, primary teeth, fractures, periapical lesions, remaining roots, oral and maxillofacial lesions, oral anatomical landmarks, arrangements of the patient's teeth that are undesirable in light of applicable orthodontic standards, and / or combinations thereof. Tooth arrangements may be undesirable for medical, orthodontic, aesthetic, and other reasons, such as overbite, crossbite, open bite, overjet, underbite, etc. The term "Application Programming Interface" or "API" refers to a set of subroutine definitions, protocols, and tools for building software. Generally speaking, it is a well-defined set of ways for different components to communicate with each other. The terms "cloud" or "cloud computing" refer to an information technology (IT) paradigm that enables ubiquitous access to a shared pool of configurable system resources and high-level services that can be rapidly provisioned with minimal administrative effort, often over the Internet.

[0020] The term "convolutional neural network" ("CNN") is a term traditionally used in the art and generally refers to a powerful tool for computer vision tasks, where deep learning CNNs can be formulated to automatically learn mid-level and high-level abstractions taken from raw data such as images. A convolutional layer can contain one or more convolution kernels, where each convolution kernel has an input tensor, with the same input tensor but different coefficients corresponding to different filters. Each convolution kernel in a layer produces a different output map, and therefore the output neurons are different for each kernel. A convolutional network may also contain local or global "pooling" layers that combine the neuron group outputs of one or more output maps. The output combination consists, for example, of taking the maximum or average value of the neuron group outputs for the corresponding output on the output map of the "pooling" layer. The "pooling" layers allow to reduce the size of the output map from one layer to another in the network, while at the same time improving the performance level by increasing the tolerance to small deformations and transformations of the input data. Below is an example of a pseudocode algorithm that shows the operation of a basic CNN.

number

[0021] The term "machine learning" ("ML") refers to applications of artificial intelligence (AI) that give systems the ability to learn and improve from experience without being explicitly programmed. Machine learning focuses on developing computer programs that can access and use data to teach themselves.

[0022] The term "deep learning" ("DL") is a subfield of machine learning that concerns algorithms inspired by the structure and function of the brain, known as artificial neural networks. An often-cited benefit of deep learning models is their ability to automatically extract features from raw data, also known as feature learning.

[0023] The term "module" refers to a self-contained unit, e.g., an assembly of electronic components and associated wiring, or a piece of computer software, that performs a defined task by itself and can be linked with other such units to form a larger system.

[0024] The term "server" includes any suitable hardware and / or software system, mechanism, or component that processes data, signals, or other information. A server may include a general-purpose central processing unit (CPU), a multiprocessing unit, dedicated circuitry implementing a particular function, or other system. A process need not be limited to a geographic location, nor need it be time-bound. For example, a server may perform functions in "real-time," "offline," "batch mode," etc. Some of the processing may be performed at different times and places by another (or the same) processing system. Examples of server systems include clients, end-user devices, routers, switches, network storage, etc. A computer is any server that communicates with a memory. A memory is any suitable processor-readable storage medium suitable for storing instructions to be executed by a server, such as random access memory (RAM), read-only memory (ROM), hard disk drive (HDD), solid-state disk (SSD), or other tangible medium.

[0025] Unless otherwise specified, the terms "comprising" and "comprise," and grammatical variations thereof, are intended to express "open" or "inclusive" language, which includes the enumerated elements and also permits the inclusion of additional, unenumerated elements.

[0026] Throughout this disclosure, certain embodiments may be disclosed in a range format. It should be understood that the description in range format is merely for convenience and brevity and should not be construed as an inflexible limitation on the scope of the disclosed range. Thus, the description of a range should be considered to have specifically disclosed all possible subranges as well as individual numerical values ​​within that range. For example, the description of a range of 1 to 6 should be considered to have specifically disclosed subranges such as 1 to 3, 1 to 4, 1 to 5, 2 to 4, 2 to 6, 3 to 6, etc., and also discloses individual numerical values ​​within that range, such as 1, 2, 3, 4, 5, and 6. This applies regardless of the breadth of the range.

[0027] The invention illustratively described herein can be suitably practiced in the absence of any element or limitation not specifically disclosed herein. Thus, for example, terms such as "comprise," "include," "containing," and the like, are to be interpreted expansively and without limitation. Moreover, the terms and expressions used herein are used as terms of description and not of limitation, and in the use of such terms and expressions, there is no intention to exclude equivalents of the features shown and described or portions thereof, but it is recognized that various modifications are possible within the scope of the claimed invention. [Brief description of the drawings]

[0028] Various other objects, features and attendant advantages of the present invention will be better understood when considered in conjunction with the accompanying drawings, in which like reference characters designate the same or similar parts throughout the several views and will be fully understood, wherein:

[0029] [Figure 1] FIG. 1 illustrates various computing hardware embodiments of the systems and methods disclosed herein.

[0030] [Diagram 2]FIG. 2 illustrates an embodiment of the structure of modules used in the second server disclosed herein and their operational relationships with each other and with computer system hardware.

[0031] [Diagram 3] FIG. 3 is a representative workflow of the general steps and processes encompassed in the methods disclosed herein for generating a patient treatment plan.

[0032] [Figure 4] FIG. 4 is an exemplary flow chart illustrating the system and operational relationships between the computing hardware, software components, and input from the user regarding dental charts, dental questionnaires, and treatment planning options.

[0033] [Diagram 5] FIG. 5 illustrates an embodiment of the structure of modules used to train a deep learning CNN as disclosed herein and their operational relationships to each other and to computer system hardware.

[0034] [Figure 6A] FIG. 6A is a diagram illustrating an embodiment of a CDSS and a first server disclosed herein and their operational relationship to each other. [Figure 6B] FIG. 6B is an exemplary flow chart illustrating the operational relationships of the CDSS between the computational hardware, software components, and input from the user and how they interact with the CDSS.

[0035] [Figure 7A] FIG. 7A is a representative screenshot of the dashboard provided to the user via the output unit, showing an exemplary presentation of the questionnaire and final treatment plan. [Figure 7B]FIG. 7B is a representative screenshot of the dashboard provided to the user via the output unit, showing an exemplary presentation of the questionnaire and final treatment plan. [Figure 7C] FIG. 7C is a representative screenshot of the dashboard provided to the user via the output unit, showing an exemplary presentation of the questionnaire and final treatment plan. [Figure 7D] FIG. 7D is a representative screenshot of the dashboard provided to the user via the output unit, showing an exemplary presentation of the questionnaire and final treatment plan. [Figure 7E] FIG. 7E is a representative screenshot of the dashboard provided to the user via the output unit, showing an exemplary presentation of the questionnaire and final treatment plan. [Figure 7F] FIG. 7F is a representative screenshot of the dashboard provided to the user via the output unit, showing an exemplary presentation of the questionnaire and final treatment plan.

[0036] [Figure 8] FIG. 8 is an example of an X-ray image that is input to the methods and systems disclosed herein.

[0037] [Figure 9] FIG. 9 is an example of a dental chart that may be populated with the methods and systems disclosed herein. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0038] In the following description, reference is made to the accompanying drawings that form a part hereof, and in which is shown by way of illustration specific examples in which the invention may be practiced It is to be understood that other embodiments may be utilized and structural changes may be made without departing from the scope of the present invention.

[0039] Disclosed herein are computer-implemented methods and systems for analyzing the condition of a patient's oral cavity to assist in providing a diagnosis and treatment plan for said patient. The methods and systems disclosed herein provide knowledge of the condition of the oral cavity to the patient in a visually simple manner, reducing the risk of "undetected" dental problems, allowing the patient to seek treatment as early as possible. As can be appreciated, early detection and treatment of dental problems and oral conditions or diseases can prevent costly treatment in the future as a result of delayed detection. Thus, the diagnosis and treatment plan generated is not based solely on the analysis of a dental professional (such as a dentist), but can vary widely depending on the training, capabilities, and educational sophistication of that professional.

[0040] The methods and systems disclosed herein aim to provide an end-to-end solution, from diagnosis and dental charting to devising treatment care for patients, by providing insights into the state of a patient's oral cavity using developed and trained artificial intelligence capabilities. The employed artificial intelligence capabilities provide insights by analyzing the patient's dental images and are able to infer clinically significant information that may have been overlooked by the manual efforts of dental professionals.

[0041] Artificial intelligence (AI) can be used in conjunction with CDSS to generate medical treatment plan options for patients. In this regard, the patient's oral condition can be analyzed with an AI model to identify oral conditions and diseases, and the output can be input into a CDSS to assist in diagnosis and generation of personalized treatment plan options.

[0042] It is understood that the performance of software-only systems in analyzing patient data (such as dental images) often falls short of the level required for accurate results and prevention of misinterpretations. Thus, in addition to employing artificial intelligence capabilities, the methods and systems disclosed herein can leverage the knowledge and input of dental professionals. In particular, to facilitate the decision-making process and the generation of a comprehensive treatment plan, the methods and systems disclosed herein can utilize input from dental professionals (i.e., users) in performing certain updates, decisions, and selections that cannot be performed or performed well by software components alone. In this regard, the methods and systems disclosed herein represent a symbiotic human-machine approach in analyzing a patient's oral condition, combining the respective advantages of computer and human data processing while minimizing the human involvement required. To that end, a diagnosis and treatment plan for a patient can be created on a combined artificial intelligence and human intelligence basis.

[0043] As will be appreciated, the systems and methods described herein can be implemented on a computer using a combination of both hardware and software. Various aspects can be implemented on a programmable computer, each computer including one or more input units, data storage media, hardware processors, and output units or communication interfaces. Use of terms such as server, service, unit, module, interface, portal, platform, or other system of computing devices is deemed to represent one or more computing devices with at least one processor configured to execute software instructions stored on a computer-readable storage medium. For example, a server may include one or more computers operating as a web server, database server, or other type of computer server in a manner that performs the roles, responsibilities, or functions described.

[0044] In one embodiment, the systems and methods described herein may include one or more input units for inputting at least one dental image, one or more computer-readable storage media configured to store instructions defining the AI ​​model and the CDSS, one or more servers that execute the instructions defining the AI ​​model and the CDSS, one or more databases, one or more cloud storage systems, and one or more output units.

[0045] FIG. 1 illustrates a representative embodiment of computing hardware that can be used in the methods and systems disclosed herein. In one embodiment, the methods and systems disclosed herein generally can include an input unit for receiving dental images from a user and inputting them into a first server. The second server can communicate with the first server to access and execute an AI model stored as software instructions in one or more computer-readable storage media. The AI ​​model can communicate with both a database and a cloud storage system via the first server and the second server. The first server can also be configured to include and execute a CDSS. It can be configured to include an output unit to communicate with a user via the first server.

[0046] In one embodiment, users include one or more individuals, one or more patients, one or more dental or clinical professionals, one or more physicians, and other stakeholders / parties.

[0047] In one embodiment, the input unit of the system may include any computing device or input module capable of receiving input from one or more databases, cloud storage systems, or image capture devices. In one embodiment, dental images may be input by a user via the input unit or by integrating an image capture device into the computer system. For example, the computer system may be integrated with an x-ray machine to load x-rays into the computer system. The input dental images may be displayed and reviewed via the output unit (i.e., a user interface).

[0048] In one embodiment, the first server may be a web-based database server, such as an Integrated Dental Management System (IDMS) server.

[0049] In one embodiment, the second server can be an AI server. The AI ​​server can be any conventionally used virtual machine with CPU power capable of executing the AI ​​model disclosed herein for inference of input dental images and providing AI predictions for creating dental charts. FIG. 2 illustrates a representative embodiment of the structure of the second server deploying the AI ​​model using various communication links for operational relationships between software and hardware components of the computer system. In particular, the second server can access or communicate with the first server, a database via the first server, and a cloud storage system for analyzing dental images using the AI ​​model. In one embodiment, the second server can include a pre-processing module, a post-processing module, and an application programming interface (API) endpoint module for deploying or executing the AI ​​model to provide AI predictions. The pre-processing module can convert the dental images into a suitable format that the AI ​​model can use. The post-processing module converts the raw output from the AI ​​model into a suitable format that the first server can use and determines where all the AI ​​predictions for the teeth belong.

[0050] In one embodiment, the database serves as a local memory store and can be a persistent memory cache. The database can also be a relational, NoSQL, or other disk storage based database. Data input and output to the database can be routed by one or more internal APIs. Data stored in the database can include, but is not limited to, AI predictions output by AI models, a model registry of trained AI models, model deployment logs for deployed AI models, audit logs of AI predictions for individual dental images, etc. As described below, the AI ​​predictions can be in the form of bounding box data.

[0051] In one embodiment, the cloud storage system can store data from the first and second servers via links based on communication spectrums such as WiFi, Bluetooth, Sigfox, Lora, IoT, cellular, etc. One or more APIs enable data transfer between the cloud storage system and the first and / or second servers. The one or more APIs can communicate with the cloud storage system to route data and requests. In one embodiment, the data stored in the cloud storage system includes, but is not limited to, raw dental images, TF (TensorFlow) record files, AI model weights, AI model log files for tracking, and patient data. In this regard, the TF record files refer to dental images that indicate bounding box data and are processed by an AI model optimized for use with the Tensorflow framework. The patient data includes the patient's medical history, dental history, social history, and biodata (e.g., age, gender, etc.). The AI ​​model weights can refer to mathematical values ​​derived in mathematical functions of the AI ​​model after training of the AI ​​model.

[0052] In one embodiment, the output unit can communicate or display to the user terminal the output of one or more steps of the method disclosed herein in addition to the final treatment plan result. In this regard, the output can be displayed via the first server, more specifically, the IDMS server, whereby one or more APIs communicate with the first server upon request and display the output via the output unit. In one embodiment, the output unit can communicate or display to the user a dashboard representing the output of one or more steps of the method disclosed herein in addition to the final treatment plan result.

[0053] In one embodiment, the output unit may include a Graphical User Interface (GUI). In one embodiment, the GUI may have the functionality to load and display patient data, dental charts, dental questionnaires, treatment plan options, and final treatment plans. In one embodiment, the output unit may enable a user to selectively configure or update outputs from the system related to patient data, dental charts, dental questionnaires, and treatment plan options. In one embodiment, the GUI may include a dashboard display system that enables a user to selectively configure or update outputs from the system. The dashboard may include an interactive database of menu items that a user may access using touch screen technology or actuatable buttons to select items to display and to select the desired display items and the appearance of the displayed items.

[0054] The architectural design of the computer system disclosed herein provides APIs for creating, accessing, and using data obtained from a user and processing dental images via the AI ​​model and CDSS, as well as training the AI ​​model. In one embodiment, the APIs utilized in the system disclosed herein include an endpoint API, an object detection API, an inference API, a training API, and a CDSS API. As will be appreciated, one or more additional APIs can link the database and cloud storage system to transfer data to and from the server.

[0055] Additionally, the computer systems disclosed herein may include additional components. For example, the system may include one or more communication channels or interconnection mechanisms, such as a bus, controller, network, etc., that interconnect the components of the system. Various embodiments of operating system software provide an operating environment for various software executing on the computer system and manage various functions of the components of the computer system. The communication channels enable communication with various other computing entities over a communication medium. The communication medium provides information, such as program instructions, or other data in the communication medium. The communication medium includes wired or wireless methodologies implemented with electrical, optical, radio frequency, infrared, acoustic, microwave, Bluetooth, or other transmission mediums.

[0056] In one embodiment, a system for generating a dental treatment plan for a patient is provided, the system may include: one or more input units for receiving dental images; one or more computer readable storage media configured to store instructions defining an AI model and a CDSS; one or more servers for executing instructions defining at least one AI model and a CDSS; a database; a cloud storage system; and an output unit configured to communicate results to a user, where the one or more servers may be configured to perform operations including: generating AI predictions based on the dental images using the AI ​​model; populating a dental chart based on the AI ​​predictions and inputs received from the one or more users; generating a completed dental questionnaire using the CDSS based on the dental chart and inputs received from the one or more users using the CDSS; generating a final treatment plan based on the completed dental questionnaire using the CDSS.

[0057] In another embodiment, a system for generating a dental treatment plan for a patient is provided, the system including: an input unit for receiving dental images; a computer readable storage medium configured to store instructions defining an AI model and a CDSS; a first server executing instructions defining a CDSS for populating a dental chart based on the AI ​​predictions and user input and generating a completed dental questionnaire and final treatment plan based on the user input; a second server executing instructions defining an AI model for generating the AI ​​predictions; a database; a cloud storage system; and an output unit configured to communicate results to a user.

[0058] FIG. 3 illustrates an exemplary workflow and steps involved in a method that may be performed on a computer system disclosed herein to generate a treatment plan for a patient.

[0059] Analysis of a patient's oral condition and generation of a treatment plan begins at step 101, where dental images of the patient to be processed are input to a first server. The input dental images of the patient are communicated from the first server to a user via an output unit such as a graphical user interface.

[0060] The patient's dental image may be a radiograph, such as an X-ray image. In one embodiment, the X-ray image may be an occlusal image, a periapical image, or an orthopantogram / panoramic image.

[0061] In step 102, in response to a user request, the first server can instruct the second server to execute the AI ​​model to process the dental image to generate an AI prediction. The AI ​​prediction can provide a prediction regarding tooth number, dental problem, and restoration based on the combined output or inference results of the AI ​​models. For example, the AI ​​prediction can be based on the following components: 1) the x-min, y-min coordinates, and width and height of the bounding box (x-min and y-min are the coordinates of the upper left corner of the bounding box); 2) the probability that an object exists within the bounding box; and, 3) the class probability of the object detected within the bounding box. An exemplary pseudocode for this process prior to sending to step 103 is shown below:

number

[0062] In step 103, the AI ​​predictions are sent from the second server to the first server. The first server populates the dental chart based on the received AI predictions. In step 104, the populated dental chart can be communicated to a user via a user interface for the purpose of checking the dental chart and the information displayed therein. The user can selectively edit the dental chart if the dental chart needs to be changed or updated with new clinical information, patient data, examination results, or if discrepancies need to be corrected. The final updated dental chart can be reviewed and verified by the user for further processing by the first server.

[0063] The user interface can include elements that allow the user to confirm or reject the generated AI predictions, as well as add optional tooth information.

[0064] A user interface displaying a dental chart can display detected teeth by number in addition to the dental problems indicated by visual elements and labels on the dental chart. The position and dimensions of the visual elements can be visually distinguishable depending on the type of dental problem detected. For example, non-pathological conditions are marked in a first color on the image and pathological conditions are marked in another color. The user interface can include user interface elements that allow a user to add or remove visual elements based on a physical review and evaluation of the patient's oral cavity.

[0065] A standard symbol-numbered dental chart can be displayed, including multiple teeth representing a typical arrangement of a complete dentition of adult or primary teeth, as appropriate, and mapped or correlated to the tooth position, number, and detected dental problems. The dental chart can be color-coded to indicate the location of the dental problem (e.g., surface or tooth level). Additionally, teeth that appear on the standard chart but are not detected (e.g., missing teeth) can also be displayed as black. Thus, the symbolically depicted teeth on the dental chart can be correlated with regions of interest identified by bounding boxes and other visual elements. Teeth that are detected but otherwise have no dental problems (i.e., "normal" teeth) can be omitted from the dental chart or can be included and displayed as normal.

[0066] In one embodiment, the dental chart may be accompanied by a table listing the AI ​​predictions for each tooth.

[0067] The user can review the entered dental chart and the information therein, correct it if necessary, and then submit the final dental chart.

[0068] In step 105, based on the final, verified dental chart, the first server instructs the CDSS to execute and generate a partially completed dental questionnaire. In this context, "partially" may refer to an incomplete questionnaire where one or more questions are missing answers. A questionnaire may be partially complete because some of the answers are automatically extracted from both the patient data stored in the cloud storage system and the final dental chart.

[0069] The first server can send the partially completed dental questionnaire, including both answered and unanswered questions, to the user. In step 106, the user can check the answered questions and complete the unanswered questions to provide the completed dental questionnaire to the first server for further processing.

[0070] In one embodiment, the questions can be in the form of single and / or multiple choice questions related to the location of the dental problem, symptoms of the dental problem, patient biodata, and treatment options. In one embodiment, the questions can be in the form of "yes" or "no" questions, in addition to single and / or multiple choice questions.

[0071] In one embodiment, the questionnaire can be presented to the user via a user interface or dashboard with separate interactive sections (tabs) for the patient biodata, each dental problem shown in the dental chart, and treatment options. In one embodiment, the questionnaire can include a section for general patient biodata and a section for each dental problem. Each of these sections can include a set of pre-populated questions depending on the dental problem to be answered by the user. The general patient biodata section can include questions regarding the patient's medical history, dental history, social history, general dental risk factors, and symptoms the patient is currently exhibiting (e.g., fever, bad breath, bleeding gums, general pain, etc.). The section for each dental problem can include a representation of the populated dental chart showing the tooth number corresponding to that dental problem, the location of the dental problem at the surface level or tooth level (e.g., mesial, occlusal, distal, buccal, lingual), and one or more questions related to the symptoms of the dental problem.

[0072] In one embodiment, questions related to symptoms of dental problems can be presented one after the other in succession, with question number 2 being displayed only after an answer to question 1 is selected. Thus, the CDSS can be a rule-based framework in which questions are generated in the form of a directed graph.

[0073] In step 107, the first server may instruct the CDSS to process the completed dental questionnaire and generate a preliminary diagnostic assessment of the patient's oral condition with one or more treatment plan options.

[0074] The diagnostic assessment and one or more treatment plan options are communicated and confirmed by the user via an output unit (e.g. a dashboard). In particular, for each identified diagnostic assessment, the CDSS can suggest one or more treatment plans, orthodontic appliances, or a combination of both. Diagnostic assessments and treatment plans can be generated separately for each tooth where a problem was identified and for the whole mouth.

[0075] In step 108, the user selects a treatment plan option for each tooth for which a dental problem has been identified. The selection of the treatment plan option may be at the user's discretion or may involve consultation between the user and the patient to determine the optimal treatment plan. In this regard, the user may consider patient preferences or dentist-specified preferences when selecting a treatment plan option. For example, a patient may filter the proposed treatment and appliance results based on cost, risks and benefits, side effects, personal preferences, pain, dietary difficulties, and the relative aesthetics of the treatment. Similarly, a dentist may filter the proposed treatments, ideally taking into account the patient's overall condition, including oral condition, health status, socio-economic status, diet and behavior, environmental conditions, and the risks and prognosis of the treatment options.

[0076] In step 109, the treatment plan options selected by the user for each identified dental problem are input to a first server. The first server instructs and executes the CDSS to process the selected treatment plan options to generate a final treatment plan and sequence of steps for that treatment. The final treatment plan is communicated to the user via an output unit.

[0077] The treatment sequence may take the form of one or more phases, including but not limited to an "emergency phase," a "control phase," a "reevaluation phase," a "definition phase," and a "maintenance phase." An "emergency phase" relates to the need for emergency treatment for an identified dental problem, although if emergency treatment is not required, the treatment plan may not include an "emergency phase."

[0078] Thus, in one embodiment, a computer-implemented method for generating a dental treatment plan for a patient is provided, comprising the following steps: analyzing dental images of the patient using an AI model to generate AI predictions; populating a dental chart based on the AI ​​predictions and input received from one or more users; generating a completed dental questionnaire using a CDSS based on the populated dental chart and input received from the one or more users; generating a final treatment plan based on the completed dental questionnaire; and displaying the final treatment plan to one or more users.

[0079] In another embodiment, a computer-implemented method for generating a dental treatment plan for a patient is provided, the method can include the steps of: inputting dental images; transmitting the dental images from a first server to a second server and instructing the second server to run an AI model to generate an AI prediction; instructing the first server to populate a dental chart based on the AI ​​prediction and input from a user; instructing the first server to run a CDSS to generate a completed dental questionnaire based on the validated dental chart and input from the user; instructing the first server to run the CDSS to generate a final treatment plan based on the completed dental questionnaire; and displaying the final treatment plan to a user.

[0080] In one embodiment, a computer program is provided comprising instructions which, when executed by a computer, cause the computer to carry out the methods disclosed herein.

[0081] In one embodiment, a computer system is provided that includes a processor configured to perform the methods disclosed herein.

[0082] In one embodiment, a computer readable storage medium is provided that includes instructions that, when executed by a computer, cause the computer to perform the methods disclosed herein.

[0083] 4 is a representative flow chart of the system and operational relationships between the computing hardware, software components, and input from the user regarding the population and validation of dental charts, generation of completed dental questionnaires, treatment planning options, and subsequent final treatment planning. In particular, FIG. 4 further defines the method disclosed in FIG. 3 by including intermediate steps.

[0084] In step 201, a patient's dental image is input to a first server for a first processing. In step 201a, the input patient's dental image is communicated to a user by the first server via an output unit and also sent to a second server. In step 201b, in response to a user's request, the first server instructs the second server to process the dental image through an AI model.

[0085] In step 202, the AI ​​model processes the dental images and generates AI predictions as output.

[0086] In step 203, the AI ​​predictions are sent from the second server to the first server. The first server populates the dental chart based on the received AI predictions.

[0087] In step 204, the populated dental chart is communicated to a user via a user interface for the purpose of reviewing and checking the dental chart and the information presented therein. In step 204a, the user can selectively edit the dental chart if the dental chart needs to be changed or updated with new clinical information, patient data, examination results, or if discrepancies need to be corrected. In step 204b, the updated final dental chart is received by the first server for further processing. In step 204c, the updated dental chart is communicated to the user via a user interface for further review and checking. If no further changes or updates are required in the updated dental chart, the user can review and verify the final dental chart for further processing by the first server. In step 204d, upon the user's request, the first server can instruct and execute the CDSS to process the final dental chart.

[0088] In step 205, the CDSS generates a partially completed dental questionnaire and sends it to the first server. In step 205a, the partially completed dental questionnaire is received by the first server and can be communicated to the user via a user interface for review and checking.

[0089] In step 206, the user can check answered questions, complete unanswered questions, and submit the completed dental questionnaire to the first server for further processing. In step 206a, in response to a user request, the first server can instruct the CDSS to further process the completed dental questionnaire.

[0090] In step 207, the CDSS processes the completed dental questionnaire and generates a preliminary diagnostic assessment of the patient's oral condition with one or more treatment plan options. In step 207a, the preliminary diagnostic assessment and treatment plan options are received by the first server and communicated to the user via a user interface for review and checking.

[0091] In step 208, the user reviews the preliminary diagnostic assessment and selects a treatment plan option. In step 208a, at the user's request, the first server can instruct the CDSS to further process the selected treatment plan option. In step 208b, the CDSS processes the selected treatment plan option and generates a final treatment plan and sequence of steps for that treatment.

[0092] In step 209, the final treatment plan and sequence of steps are received by the first server and communicated to the user via the user interface. In step 209a, the final treatment plan and sequence of steps can be reviewed by the user.

[0093] Artificial Intelligence (AI) Model Artificial intelligence image recognition can be used to analyze a patient's dental images, such as orthopantomogram images, to determine the condition of the oral cavity. The artificial intelligence models disclosed herein can combine image recognition and localization to detect objects and incorporate the analysis results into an automated tooth charting system. The analysis results and output of the AI ​​models disclosed herein can provide an AI prediction. The AI ​​prediction can be based on meeting a confidence threshold.

[0094] To generate the AI ​​predictions, the AI ​​model can automatically perform object detection on the input dental images to detect and identify tooth localization for tooth numbering and classification related to dental issues. In one embodiment, the AI ​​model can further perform keypoint detection. Tooth location and tooth number can be determined by forming a bounding box around the tooth. Tooth classification can be based on object detection within the bounding box. For object detection, the AI ​​model mainly focuses on extracting features that help detect the bounding box region and the possibility of an object being present in that region, and the possibility that the object is related to the dental issue itself.

[0095] The AI ​​prediction is based on components including: 1) the bounding box's x-min, y-min coordinates, and width and height (x-min and y-min are the coordinates of the bounding box's top-left corner); 2) the probability that an object exists within the bounding box; and 3) the class probability of the object detected within the bounding box.

[0096] In this regard, a bounding box may define the boundaries of each tooth captured in the dental image. A number may then be assigned to each detected tooth. The numbers may be assigned according to FDI notation, the Universal Numbering System, Palmer notation, or any other suitable dental notation. It is understood that the numbering of the teeth may be based on an adult dental chart, although a primary dental chart and suitable notation may be used for younger patients.

[0097] The bounding boxes formed around the teeth can indicate different arrangements of the teeth in the dental image, for example a first bounding box around the anterior teeth, a second bounding box around the left molar teeth, a third bounding box around the right molar teeth, a fourth bounding box around all teeth, etc. Additionally, the bounding boxes can include data regarding the location of the detected objects (e.g. dental problems) (surface level or tooth level (mesial, occlusal, distal, buccal, lingual, etc.)).

[0098] Thus, the AI ​​prediction as an output of the AI ​​model can be one or more probabilities. In one embodiment, the output of the AI ​​model can act as a classifier and assign a probability for detecting a dental problem to an identified tooth in the dental image. The threshold and probability value range (i.e., range from 0 to 1) for each classification and dental problem can be modified and set appropriately.

[0099] In one embodiment, the classification may refer to dental problems including, but not limited to, caries, periodontal problems, missing teeth, crowns, bridges, implants, fillings, onlays, inlays, veneers, failed or defective restorations (fillings, crowns, bridges, veneers, inlays, onlays), root canal treatments, posts and cores, impacted teeth, partially erupted teeth, unerupted teeth, primary teeth, fractures, periapical lesions, remaining roots, oral and maxillofacial lesions, oral anatomical landmarks, undesirable patient tooth positioning according to applicable orthodontic standards, and / or combinations thereof. Tooth positioning may be undesirable for medical, orthodontic, aesthetic, and other reasons, such as overbite, crossbite, open bite, overjet, underbite, etc. In one embodiment, the classification is caries, periodontal disease areas, and restorations. Each dental problem may be at a different stage, progression, and / or severity of progression.

[0100] Thus, the AI ​​model can perform tooth detection and numbering, as well as detection and classification of non-pathological and pathological dental problems from input dental images.

[0101] In one embodiment, the AI ​​model includes one or more deep learning CNNs for object detection, where each deep learning CNN can be used to detect a different dental problem. In one embodiment, each deep learning CNN can be used to detect objects that are dynamic in size (e.g., missing teeth, cavities, periodontal disease, dental restorations or fillings) or static in size (e.g., tooth position / number, impacted teeth).

[0102] In one embodiment, the AI ​​model may include two or more deep learning CNNs for object detection, where each deep learning CNN may operate in parallel, rather than sequentially or sequentially, to process the same input dental image separately. In this regard, the dental image may be input to each deep learning CNN and processed simultaneously to provide separate output predictions, which may be combined to form a unified AI prediction. For example, a first deep learning CNN may be used to detect tooth position / number, a second deep learning CNN may be used to detect caries, a third deep learning CNN may be used to detect restorations, a fourth deep learning CNN may be used to detect baby teeth, a fifth deep learning CNN may be used to detect impacted teeth, and an Nth deep learning CNN may be used to detect any of the other dental issues.

[0103] In one embodiment, the machine learning CNN is a deep learning CNN. In one embodiment, the AI ​​model can include two or more deep learning CNNs. In another embodiment, the AI ​​model can include two deep learning CNNs.

[0104] The first deep learning CNN can be used to detect objects whose size is dynamic. The second deep learning CNN can be used to identify objects whose size is static. In one embodiment, both the first and second deep learning CNNs are object detection CNN models.

[0105] The first deep learning CNN can have an architecture that predicts objects based on three points that form the bounding box of each object for localization and classification. The three points can be used to form the bounding box of each object: the top left corner, the bottom right corner, and the center. The first deep learning CNN works based on the concepts of center pooling and cascaded corner pooling. Center pooling refers to the horizontal and vertical maxima of the feature map; cascaded corner pooling refers to the identification of the boundary maximum, the interior maximum, and the sum of these two values.

[0106] The second deep learning CNN can have a different architecture than the first deep learning CNN, where the first deep learning CNN predicts objects as points rather than predicting objects by regions or boxes. The second deep learning CNN can employ a Bidirectional Feature Pyramid Network (BiFPN), unlike other CNN architectures such as Faster-RCNN. In particular, the second deep learning CNN can have an architecture that is wide enough to capture finer details and deep enough to extract larger semantics from the image itself (as small objects become larger in larger images). The size of the input image can be important, and so there can be a heuristic-based scaling mechanism within the second deep learning CNN. To accommodate these requirements, the second deep learning CNN can have a compound scaling architecture.

[0107] In a preferred embodiment, the deep learning CNNs include EfficientDet and CenterNet. The EfficientDet model can be used to identify objects whose size is static, and the CenterNet model can be used to detect objects whose size is dynamic.

[0108] EfficientDet relies on an anchor mechanism to determine prior regions of interest for detection, which can be tedious as the anchors need to be properly configured to accommodate different sizes. In contrast, CenterNet detects objects by points, which means fewer hyperparameters to tune and more dynamism in predictions, which is beneficial when the model recognizes object classes that may vary in size during out-of-sample prediction. It is understood that CNNs other than EfficientDet and CenterNet can be employed in the AI ​​models disclosed herein that perform the same functions and results in object detection.

[0109] The deep learning CNN disclosed herein can be trained using a dataset of patient information representing various oral structures and oral conditions (health or dental problems). Because CNNs can be trained using deep learning techniques, CNNs can be applied to the systems and methods disclosed herein to accurately and sensitively identify tooth locations and classes, and objects that may be indicative of symptoms of dental problems. In particular, the deep learning CNN can be trained using large datasets obtained from a cohort of patients. A machine learning package or platform can be used to train the CNN. The dataset can include orthopantomogram recordings of patients combined with examination results from clinical experts that help validate the CNN's analysis results.

[0110] 5 illustrates an exemplary embodiment of the structure of modules that can be used to train a deep learning CNN and their operational relationships to each other and to specific hardware of a computer system. In particular, training a CNN involves the use of a preprocessing extension module, a TF record creation module, a data checker module, a data generator module, and the TF object detection API.

[0111] The preprocessing extension module allows the dental images to be manipulated to improve object detection, such as cropping out unwanted parts of the image. The TF record creation module allows the dental images to be converted into a protobuf file (one file containing all images). This file is optimized within the TensorFlow framework when loading / processing the data. The data checker module can be used to ensure quality labels (e.g. tooth number labels) by performing certain heuristic checks. The data generator module allows the labels in the XML file to be converted into a tabular format for easier use in TF record sections and analysis. The object detection API is an open source library by search platforms such as Google, where many popular object detection model architectures are coded and available for use.

[0112] The modules of the training system and the API are connected to the database and the cloud storage system via bidirectional links, respectively, and can transmit data therebetween. In one embodiment, the modules of the training system can be stored as software instructions on a computer-readable storage medium.

[0113] In one embodiment, the deep learning CNN can be retrained or optimized periodically, rather than continuously, and thus can be updated and retrained in a batch process as needed.

[0114] After training of the deep learning CNN, the output AI prediction results are preferably validated using a cross-validation technique on a blind dataset. In particular, the deep learning CNN can be processed on a validation set of dental images, which is a different image dataset that was not used for training and is separate from the training dataset. The performance of the deep learning CNN using the validation dataset can be compared to the training dataset to determine the accuracy of the deep learning CNN disclosed herein. In this regard, a validation dataset can be used for each CNN used, and the results will indicate which CNN is best to use for detecting dental problems.

[0115] The outputs obtained from the AI ​​model related to the detected tooth number and tooth problem form the AI ​​predictions. These AI predictions are sent from the second server to the first server to populate the dental chart. In one embodiment, the raw outputs from the AI ​​model can be in tensor format, which is then converted to JSON and returned to the first server. In particular, the output results from each CNN of the AI ​​model can be combined along with their probabilities and localities to form the final raw AI model output.

[0116] The AI ​​predictions can be used to automatically populate the dental chart in a post-processing step. This automatic population of the dental chart can use metrics such as Intersection over Union (IOU), which in particular refers to the percentage of the overlapping area of ​​two bounding boxes relative to the total area of ​​the union of the two boxes, and algorithms that assign a detected classification (e.g., caries, restoration) to each tooth that appears in the dental chart.

[0117] Clinical Decision Support Systems (CDSS) Clinical decision support systems (CDSS) are a branch of expert systems (ES) that leverage medical knowledge engineering. Using ES design principles, CDSSs can simulate the diagnostic and treatment process that is typically performed by medical professionals. The goal is to assist clinical professionals in solving complex medical problems and making diagnoses.

[0118] In the traditional structure of a decision support system, a decision maker or user arrives at a solution by interpreting and understanding the information and problem at hand. The process can be complicated when there are multiple decision makers, and even more so when the information available is subjective, objective, or a combination of both.

[0119] In the specific context of a dental clinic, the problem and the solution concern the most suitable dental treatment for the patient. The decision made by the decision maker depends on the problem itself, which affects not only the information related to the problem but also the criteria adopted by the decision maker. In most clinical situations, the patient also acts as the decision maker, and information on financial costs and aesthetic demands may affect the final decision outcome. Dentists have limitations and differences in cognitive functions, such as recalling and applying possible risk factors and assessments, which may lead to potential differences in decisions made by different dentists or even by the same dentist at different times. The CDSS disclosed here can minimize such differences. Specifically, rather than a subjective analysis or diagnosis based on the individual clinical expert managing the patient, the CDSS disclosed here represents the collected expertise from a dental expert committee consisting of professors, dental specialists, and experienced dentists. Thus, the risk of undertreatment, overtreatment, and / or negligence can be reduced, greatly benefiting the patient.

[0120] Traditionally, CDSSs can be built solely on the basis of a data-driven approach using historical data. Although this approach is unbiased and relatively cheaper than a knowledge-based approach alone, the main drawback is that the data-driven approach is limited to a specific type of treatment plan. For the successful implementation of a decision support system, it is important to consider knowledge-based data and information within the domain and technical field.

[0121] Thus, in one embodiment, the CDSS employed is a rule-based or knowledge-based CDSS, i.e., the CDSS disclosed herein does not employ a data-driven approach, but is only a knowledge-based CDSS for generating dental questionnaires, diagnoses, and treatment planning options.

[0122] The CDSS employed in the methods and systems disclosed herein can advantageously provide an automated diagnostic and treatment plan option system, whereby a user can receive objective and reliable treatment plan options based on the patient's needs. In this regard, the CDSS disclosed herein can be configured and designed to provide an expert system for the decision-making process using the specific characteristics of each patient. In one embodiment, the CDSS disclosed herein can be in the form of a software application or program representing a data processing method.

[0123] In one embodiment, the CDSS includes a rule-based framework in the form of a directed graph. A directed graph can include one or more nodes / vertices as basic units and edges pointing in a particular direction as connectors between the nodes, e.g., A>B means you can go from A to B but not vice versa. So the CDSS and rule-based or knowledge-based approach are in the form of graph rules coded into the software. These rules dictate which questions to present to the user next based on the answers to previous questions, i.e. the rules that generate the diagnostic and treatment plan opsins.

[0124] In one embodiment, the CDSS may include a rules engine with a rule-based framework in the form of a directed acyclic graph. The rule-based framework is a JavaScript front-end framework that can store rules in an XML file or a graph database. The JavaScript front-end parses the rules upon load, and then, depending on user selections in the JavaScript front-end, JavaScript traverses the graph and returns output based on the logic expressed in the rules.

[0125] In one embodiment, the CDSS disclosed herein can be constructed and developed using graph data structures that simulate knowledge-based data in a domain.

[0126] The rule-based or knowledge-based framework of a CDSS can be derived from information found in industry-recognized medical literature, such as textbooks and journals. Thus, this knowledge-based approach can incorporate academic studies and research, clinical expertise, and industry best practices, as well as medical protocols for providing treatment to patients for specific oral conditions. Clinical expertise can be obtained from renowned professors and experts.

[0127] The knowledge-based approach of CDSS focuses on creating a knowledge description language that can be combined with a reasoning engine to perform diagnostic reasoning.

[0128] The CDSS disclosed herein empowers users (e.g., clinical professionals) to intuitively and systematically collect and analyze a patient's personal and clinical data to create a comprehensive, individualized care and treatment plan based on comprehensive step-by-step protocols or rule-based graphs. The advantage of using a CDSS over machine learning tools (e.g., supervised learning) is clinical interpretability, where each decision step is clearly identified and traceable, unlike other "black box" algorithms.

[0129] FIG. 6A is an exemplary embodiment of a CDSS disclosed herein in association with a first server. The first server may include a user interface component for a user to view and, if necessary, update a patient's social, medical, and dental history, biodata, and dental chart. The first server user interface component may include an IDMS web page. The patient's social history, medical history, dental history, biodata, and dental chart may be managed and extracted by a database module. The database module may be a MySQL® database module. The first server may include a data model to organize the data received from the database module and standardize their interrelationships.

[0130] The CDSS may include a CDSS user interface component that allows the user to view the generated questionnaire, treatment plan options, and final treatment plan and to enter or select information therein. The CDSS user interface component may include HTML web pages. It may utilize a rules engine that uses rules in a JavaScript® framework. Additionally, it may include a medical rules module that allows rules to be defined in Extensible Markup Language (XML).

[0131] An API controller may be included to handle requests between the first server and the CDSS, more specifically, requests from the CDSS user interface and the first server's data model, where the data model simply manages data structures based on specified logic.

[0132] Figure 6B shows the flow chart and operational relationships between the computational components of the CDSS, as well as inputs from the user and how they interact with the CDSS.

[0133] In step 301a, the patient's social, medical, dental, and biodata can be viewed and updated by a user through a user interface of the first server. Similarly, in step 301b, the user can update and confirm the populated dental chart through a user interface of the first server. The user can then request the CDSS to process the inputs from steps 301a and 301b.

[0134] In step 302, the CDSS user interface receives input from steps 301a and 301b. In step 303, the CDSS rules engine requests medical rules to be extracted from the medical rules module for processing. In step 304, the medical rules module receives the request from the rules engine and in step 305 sends the rules to the rules engine.

[0135] In step 306, the rules engine automatically answers the question using the information received in steps 301a and 301b based on the processed rules. The rules engine generates HTML code containing the question and answer to send to the CDSS user interface in step 307.

[0136] In step 308, the CDSS user interface is populated with the responses from step 307 and generates the partially completed questionnaire and treatment plan options.

[0137] In step 309, the user may check answered questions and answer any unanswered questions to complete and submit the dental questionnaire. Upon completion of the questionnaire, treatment plan options are generated from which the user can choose.

[0138] In step 310, the rules engine processes the completed questionnaire and selected treatment plan options to generate a final treatment plan including a sequence of steps. The final treatment plan is rendered and sent to the CDSS user interface for user review in step 311. The final treatment plan is received and stored by the first server user interface.

[0139] User Interface (Dashboard) As shown in Figures 7A-F, the information and output from the AI ​​model and the CDSS can be displayed to a user via a first server and a user interface or dashboard, which includes a first server user interface and a CDSS user interface, as shown in Figures 6A and B.

[0140] In particular, the user interface / dashboard can display information through separate categories related to patient data, dental problems or conditions, and treatment plans. This dashboard can be provided to the user in the form of a web-accessible page identified by a URL or as an application executed on a computer. The dashboard can be manipulated to present a partially filled-out questionnaire and allow the user to enter and complete questions.

[0141] As shown in Figure 7A, the functional tab labeled "General" is associated with a patient information display area that includes options to select a number of patient symptoms and risks, as well as associations between the patient's biodata and the patient's medical, social, and dental history.

[0142] Figures 7B-E show separate tabs with dedicated questions for dental issues identified by the AI ​​model and presented in the final dental chart. Each tab can be provided for each identified dental issue with pre-set selections and a set of questions related to the symptoms of the dental issue for the user to complete and confirm. Once the user completes the symptom-related questions, a diagnosis and one or more treatment options are provided, from which the user can select one. For example, in Figure 7B, the tab for detected caries is shown with a representative dental chart showing that the caries is associated with tooth number 23, the caries can be differentiated as moderate caries or root caries, followed by the location of the caries (mesial, occlusal, distal, buccal, lingual), then the user can select the symptom of the caries (pain) in numbered questions 1 and 2, and choose from a diagnosis and three treatment options. Figures 7C and 7D show tabs for detected periodontal disease, which can be classified as generalized (whole mouth) or localized (per tooth), and each tab includes one or more questions about the location of the disease (left, right, upper, lower), size (<3mm, >5mm), and symptoms (ulcer, swelling, pain) that, when answered, provide a diagnosis and two treatment options. Figure 7E shows tabs for one or more detected missing teeth, which provide questions about the symptoms of the missing teeth (alignment, dimensions), and provide a diagnosis and four treatment options.

[0143] Once a treatment option for each dental problem has been selected and all necessary questions have been answered, the completed questionnaire is further processed by the CDSS to generate a final (complete) treatment plan including a sequence of treatment steps. The sequence of treatment steps may be in the form of one or more phases including but not limited to an "emergency phase", a "control phase", a "reassessment phase", a "definition phase", and a "maintenance phase". The "emergency phase" pertains to the need for emergency treatment of the identified dental problem, whereas the treatment plan may not include an "emergency phase" if emergency treatment is not required.

[0144] An example final treatment plan is shown in Figure 7F along with the sequence of treatment steps. As will be appreciated, the steps included in the final (complete) treatment plan will vary depending on the questions answered. For example, in Figure 7B, if the user selects option 2 instead of option 1, a reassessment step will appear in the final treatment plan.

[0145] Thus, in one embodiment, the order of treatment steps may be: 1. acute phase; 2. control phase; 3. reassessment phase; 4. determination phase; and 5. maintenance phase. In another embodiment, the order of treatment steps may be: 1. control phase; 2. reassessment phase; 3. determination phase; and 4. maintenance phase. In another embodiment, the order of treatment steps may be: 1. acute phase; 2. control phase; 3. determination phase; and 4. maintenance phase.

[0146] The emergency phase (UP) involves patients who present with fever, swelling, pain, bleeding, or infection and require urgent treatment.

[0147] The control phase (CP) can be designed to a) eliminate active disease such as caries and inflammation; b) remove conditions that impede maintenance; c) eliminate potential causes of disease; d) institute preventive dental activities such as i) management of gingival and periodontal infections, ii) management of caries risk, and iii) management of carious teeth.

[0148] The reassessment phase (RP) refers to a waiting phase until inflammation subsides and healing occurs. Thus, during this phase, home care habits are strengthened, willingness to undergo further treatment is assessed, and initial treatment and pulpal response are reassessed before definitive treatment begins.

[0149] The Definitive Phase (DP) refers to the phase of correcting and resolving specific dental problems. This phase may require periodontal surgery, oral surgery, and replacement of missing teeth. In one embodiment, management of missing teeth is in this Definitive Phase.

[0150] The maintenance phase (MP) may include regular recall periods for confirmation and subsequent examinations that a) may reveal the need for adjustments to prevent future breakdowns and b) provide an opportunity to intensify home care. This phase indicates that patients should be examined every 3, 6 or 12 months, depending on the stage and severity of the dental problems identified. For example, i) if the patient has stage 3 and stage 4 periodontitis or high caries risk, examinations every 3 months are indicated, ii) if the patient has moderate caries risk and stage 1 or 2 periodontitis, examinations every 6 months are indicated, and iii) if the patient has low caries risk or healthy periodontitis, examinations every 12 months are indicated.

[0151] Depending on the treatment recommendation, the user can combine the CDSS and its recommendations with their own judgment and experience to make a decision. Furthermore, the output of the CDSS can be checked and verified for consistency with the decision map using visual checks and standard tests (e.g., unit tests, integration tests) to ensure robustness. Furthermore, the CDSS has been validated by a group of experts to test the comprehensiveness and accuracy of the generated diagnoses and recommended treatment plans.

[0152] Working Example The following non-limiting examples are provided for illustrative purposes only to facilitate a more complete understanding of the representative embodiments currently under discussion. These examples are intended to be merely a subset of all possible contexts in which the components of the system and steps of the methods disclosed herein may be combined. As such, these examples should not be construed as limiting any of the embodiments described herein, including with respect to the types and amounts of the components of the system and / or the methods and their uses.

[0153] In a dental office, a patient may come in with one or more dental problems. The dentist requests an x-ray (orthopantomogram) to get a better understanding of the patient's dental condition.

[0154] The x-rays and other patient data can be fed into the computer system disclosed herein. In this regard, the system can be integrated into the clinic's Patient Management System (PMS). The system provides an output within seconds indicating the patient's oral problems. When a patient comes in complaining of pain, dentists tend to prioritize only the painful areas on the x-rays and overlook other dental problems.

[0155] By feeding x-ray images into the AI ​​model of the system disclosed herein, a comprehensive analysis is performed that could potentially detect multiple oral issues within seconds. As a result, dental charts are automatically populated with AI predictions about oral conditions that dentists can immediately review with their patients. For example, a patient may come into the clinic with a toothache, but analysis by the system may detect four to five additional interrelated issues.

[0156] The detected dental problems are combined with further clinical examination by the dentist and further processed by the CDSS disclosed herein. The CDSS generates a questionnaire for the dentist to complete, which the dentist completes by selecting treatment plan options. Once the treatment option selection and completion of the questionnaire are complete, a final treatment plan is generated for all detected dental problems in the appropriate sequence.

[0157] The correct sequence of treatment steps is important, as the success rate of treatment may be determined by the correct sequence. For example, implants are more likely to succeed if gum / bone issues are addressed first. If an implant is placed in a setting of poor oral hygiene, including gum or bone issues, the implant is more likely to fail. Thus, the systems and methods disclosed herein generate the correct sequence for continuing with gum treatment followed by implant placement.

[0158] The analysis of patient data (X-ray images) using AI models and generation of treatment plans using CDSS are completed during the patient's first consultation. This means that a patient visits the dental clinic for a potential oral condition or disease and is provided with a complete comprehensive treatment plan to resolve the patient's dental condition. Furthermore, through the comprehensive treatment plan, serious dental problems can be detected early or even prevented.

[0159] Thus, the systems and methods disclosed herein aim to improve the standard of dental practice across the industry. Instead of only receiving information and advice on the main oral symptoms, patients will receive a comprehensive, holistic, yet individualized diagnosis and treatment plan for their oral condition. Patients will be confident of a more accurate diagnosis and the best treatment options, while avoiding errors based on personal judgment and bias. Furthermore, early detection of dental diseases will be possible. This will help patients manage their treatment costs while at the same time increasing awareness and knowledge of good dental care in the community.

[0160] Example 1 FIG. 8 illustrates an exemplary x-ray image input into the systems and methods disclosed herein to analyze the patient's oral condition and generate a treatment plan according to the exemplary method steps outlined in FIG.

[0161] Initially, dentists took full-mouth X-rays during patient examinations. The X-rays were uploaded to the system and displayed to the dentist via a user interface for review. The system then analyzed the X-ray images using AI models at the dentist's request.

[0162] The results of the AI ​​model analysis were used to populate a dental chart with tooth numbers and noted dental issues, along with additional relevant information and details. The dental chart was displayed to the dentist via a user interface, who updated the dental chart to make any changes upon further examination of the patient.

[0163] The final populated dental chart is shown in Figure 9, accompanied by a table populated by the AI ​​predictions, as shown in Table 1 below. [Table 1]

[0164] Based on the final dental chart, a partially filled dental questionnaire was generated and the dentist had to answer questions about each tooth and symptom. The questionnaire was displayed to the dentist via a user interface where the dentist could fill out the questionnaire.

[0165] The dental questionnaire will include the following multiple choice questions regarding the problem with tooth number 46, as shown in Table 2 below: As can be appreciated, for other teeth identified as having problems, the questionnaire may vary depending on the problem and the details associated with it. [Table 2]

[0166] Once the questionnaire has been completed and submitted, a diagnostic assessment and one or more treatment plan options are generated by the CDSS and displayed to the dentist via the user interface. Table 3 below summarizes the information provided to the user regarding tooth number 46 and missing teeth 26 and 47. [Table 3]

[0167] After the dentist selected the treatment plan options for each tooth and the identified dental problems, the CDSS generated a final treatment plan with a set of treatment steps that considered the problems identified in all teeth for a comprehensive treatment plan and sequence of steps. The treatment steps were divided into phases and consisted of the emergency phase, control phase, reevaluation phase, confirmation phase, and maintenance phase, as shown in Table 4 below. [Table 4]

[0168] Further details regarding how to use and operate the present invention are apparent from the above description, and therefore no further details regarding how to use and operate the present invention will be provided.

[0169] With respect to the above description, it is believed that the optimum dimensional relationships of the parts of the present invention, including variations in size, material, shape, form, function, and operation, assembly, and use, will be readily apparent to those skilled in the art, and all equivalent relationships shown in the drawings and described in the specification are intended to be encompassed by the present invention.

[0170] As such, the foregoing is considered as merely illustrative of the principles of the invention. Further, since numerous modifications and changes will readily occur to those skilled in the art, it is not desired to limit the invention to the exact construction and operation as illustrated and described, and therefore, all suitable modifications and equivalents may be employed within the scope of the invention.

[0171] The foregoing has described the principles, embodiments, and modes of operation of the present invention. However, the invention should not be construed as limited to the particular embodiments discussed herein. It is to be understood that the above-described embodiments are to be considered as illustrative rather than limiting, and that changes can be made thereto by those skilled in the art without departing from the scope of the invention as defined by the following claims.

[0172] The invention has been described broadly and generically herein, and each of the narrower species and subgeneric groupings falling within the generic disclosure also form part of the invention, including the generic description of the invention with any disclaimer or negative limitation.

[0173] Other embodiments are within the scope of the following claims and non-limiting examples. Furthermore, when features or aspects of the invention are described in terms of a Markush group, those skilled in the art will recognize that the invention is also described in terms of any individual members or subgroups of members of the Markush group.

Claims

1. 1. A computer-implemented method for generating a dental treatment plan for a patient, comprising the steps of: receiving patient data including dental images of the patient; analyzing dental images of the patient using the AI ​​model to detect, number, and generate AI predictions regarding the patient's teeth and dental problems; populating a dental chart based on the AI ​​predictions and inputs received from one or more users; generating a completed dental questionnaire using a clinical decision support system (CDSS) based on the input dental chart, patient data, and input received from one or more users; generating a final treatment plan based on the completed dental questionnaire; and Displaying the final treatment plan to one or more users. The computer-implemented method comprising:

2. AI predictions are based on the following components: The minimum x, y coordinates of the bounding box with width and height; the probability that the object is in the bounding box; and Class probabilities of objects detected within the bounding box The computer-implemented method of claim 1 ,

3. 2. The computer-implemented method of claim 1, wherein the AI ​​model includes a first deep learning CNN for detecting objects that are dynamic in size and a second deep learning CNN for identifying objects that are static in size.

4. 4. The computer-implemented method of claim 3, wherein the first deep learning CNN includes a CenterNet model for detecting objects with dynamic size and the second deep learning CNN includes an EfficientDet model for identifying objects with static size.

5. The computer-implemented method of claim 1 , wherein the CDSS is a rule-based or knowledge-based CDSS.

6. The computer-implemented method of claim 5 , wherein the rule-based CDSS includes a framework in the form of a directed graph.

7. 20. The system of claim 18, wherein the knowledge-based CDSS includes a framework in the form of software-coded graph rules focused on creating a knowledge description language for making diagnostic inferences.

8. A computer system comprising a processor configured to carry out the method according to claims 1 to 7.

9. A computer program comprising instructions which, when executed by a computer, cause the computer to carry out the method according to claims 1 to 7.

10. A computer readable storage medium comprising instructions which, when executed by a computer, cause the computer to carry out the method according to claims 1 to 7.

11. 1. A system for generating a dental treatment plan for a patient, comprising: an input unit for receiving patient data including dental images; a computer readable storage medium configured to store instructions defining an AI model; a first server executing a clinical decision support system (CDSS); a second server that executes instructions that define the AI ​​model; Wherein the second server: Using the AI ​​model to generate AI predictions for tooth detection, numbering, and dental problems based on the patient's dental images; and populating a dental chart based on the AI ​​predictions and inputs received from one or more users; Wherein the first server: generating a completed dental questionnaire based on the dental charts and inputs received from one or more users using the CDSS; and using the CDSS to generate a final treatment plan based on the completed dental questionnaire; and An output unit configured to communicate dental charts, questionnaires, and final treatment plans to the user. The system comprising:

12. 10. The system of claim 1, wherein the AI ​​model combines image recognition and localization for object detection in input dental images to detect and identify tooth locations for numbering and classification of teeth related to dental problems.

13. The system of claim 12 , wherein tooth location and tooth numbering are determined by forming a bounding box around the tooth.

14. AI predictions are based on the following components: The x min, y min coordinates and the width and height of the bounding box; the probability that the object is in the bounding box; and Class probabilities of objects detected within the bounding box The system of claim 13, based on

15. The system of claim 14 , wherein a bounding box defines a boundary for each tooth captured in the dental image and assigns a number to each detected tooth.

16. 12. The system of claim 11, wherein the AI ​​model includes a first deep learning CNN for detecting objects that are dynamic in size and a second deep learning CNN for identifying objects that are static in size.

17. 17. The system of claim 16, wherein the first deep learning CNN includes a CenterNet model for detecting objects with dynamic size, and the second deep learning CNN includes an EfficientDet model for identifying objects with static size.

18. The system of claim 11 , wherein the CDSS is a rule-based or knowledge-based CDSS.

19. 20. The system of claim 18, wherein the rule-based CDSS comprises a directed graph framework.

20. 20. The system of claim 18, wherein the knowledge-based CDSS includes a framework in the form of software-coded graph rules focused on creating a knowledge description language for making diagnostic inferences.

Images