Method and system for optimized prescription management using ai recommendations

Abstract

An optimal prescription management system utilizing artificial intelligence recommendations may include memory and a processor for storing instructions. The system receives symptom information, collects prescription data corresponding to the symptom information from a database, analyzes the effectiveness of prescriptions based on the collected prescription data, selects top prescriptions based on the analyzed effectiveness to determine prescription priorities, generates a recommendation list of a set number of prescriptions according to the determined prescription priorities, and displays the generated recommendation list.

Description

Technology Field

[0001] The present invention relates to a prescription management system used in school health rooms, and more specifically, to a method and system that utilizes artificial intelligence technology to improve the reliability of prescriptions by school nurses and provide optimal prescriptions. Background Technology

[0003] In traditional school health rooms, the primary method used was for school nurses to personally assess patients' symptoms and prescribe medication. In particular, newly appointed nurses often lack experience in treating a wide range of symptoms, making it difficult to clearly determine which medication to use for specific conditions. Consequently, a learning curve is essential until experience is accumulated, and this process can lead to a decline in the consistency and reliability of prescriptions.

[0004] The existing system focused on simple patient record management, lacking the ability to analyze the correlation between symptoms and prescriptions or to refer to the prescription patterns of local health teachers. Additionally, there is no system to evaluate the efficacy of prescriptions after the fact, making it difficult to verify whether a specific prescription actually contributed to symptom improvement. The problem to be solved

[0006] The present invention aims to solve the problem of inconsistency in prescriptions and reduced reliability caused by the lack of experience of new school nurses in school infirmaries.

[0007] The existing method relies on individual doctors' experience to determine the correlation between symptoms and prescriptions, making it difficult for new doctors, in particular, to select appropriate prescriptions during the learning curve period. Furthermore, the lack of a system to evaluate the efficacy of prescriptions retrospectively has led to the problem of drugs with high recurrence rates for the same symptoms being continuously used. In addition, the existing system, which relies on manual records and analysis, has limitations in that it hinders work efficiency and impedes rapid response in emergency situations or the outbreak of infectious diseases. The present invention aims to solve these problems. means of solving the problem

[0009] An optimal prescription management system utilizing artificial intelligence recommendations may include memory and a processor for storing instructions. The system receives symptom information, collects prescription data corresponding to the symptom information from a database, analyzes the effectiveness of prescriptions based on the collected prescription data, selects top prescriptions based on the analyzed effectiveness to determine prescription priorities, generates a recommendation list of a set number of prescriptions according to the determined prescription priorities, and displays the generated recommendation list.

[0010] The system can be replaced by electronic devices (e.g., terminals, computers). The operations of the system executed by instructions can be classified as methods. Effects of the invention

[0012] The present invention enables new health teachers to refer to the prescription patterns of experienced teachers in the region in real time through an artificial intelligence-based recommendation algorithm, thereby shortening the learning curve caused by lack of experience and improving prescription reliability. By applying a multi-weighted recommendation system that combines symptom frequency and utility scores, the scientific standardization of prescriptions can be achieved by prioritizing the presentation of prescriptions that have shown the highest efficiency for the same symptom over the past three months.

[0013] A utility evaluation algorithm that automatically analyzes patient revisit data after prescription dynamically updates recommendation priorities and continuously improves the long-term efficacy of specific drugs. By automating the process from patient symptom input to prescription records through integration with kiosk systems, it reduces processing time and supports real-time notifications and reporting procedures to the Office of Education in the event of an infectious disease outbreak, thereby preventing legal risks in advance.

[0014] Furthermore, the automated generation and detailed recording system of health logs provides clear supporting evidence in the event of complaints, while the emergency recording function via a mobile app enables rapid medical intervention by tracking changes in the condition of patients on the move in real time. This integrated system optimizes the entire operation of school health rooms and can maintain a consistent level of quality in student health management. Brief explanation of the drawing

[0016] FIG. 1 is a diagram illustrating an optimal prescription management system utilizing artificial intelligence recommendation according to one embodiment. FIG. 2 is a diagram illustrating the learning of a neural network according to one embodiment. FIG. 3 is a diagram illustrating the configuration of an artificial intelligence model according to one embodiment. Figure 4 is a block diagram showing the configuration of a system providing an artificial intelligence-based optimal prescription according to one embodiment. FIG. 5 is a block diagram showing the configuration and function of a health system and the process of linking with other systems according to one embodiment. Figure 6 shows a flowchart of an optimal prescription management method utilizing artificial intelligence recommendation according to one embodiment. FIG. 7 is a flowchart illustrating an optimal prescription management method utilizing artificial intelligence recommendations according to one embodiment. FIG. 8 is a flowchart illustrating an optimal prescription management method utilizing artificial intelligence recommendation according to one embodiment. FIG. 9 illustrates a user interface (UI) that provides a prescription method to enable optimal prescription in an optimal prescription management system utilizing artificial intelligence recommendation according to one embodiment. Specific details for implementing the invention

[0017] The embodiments will be described in detail below based on the attached drawings. However, since various modifications are possible to these embodiments, the scope of the patent application is not limited by them. The specific structural or functional descriptions of the embodiments are presented merely for illustrative purposes and may be modified in various forms.

[0018] Unless specifically defined otherwise, all terms used herein have the meaning generally understood by those skilled in the art. Terms defined in commonly used dictionaries must be consistent in meaning within the context of the relevant technology, and if not explicitly defined, they are not interpreted in an overly formal sense. Additionally, when describing drawings, the same reference numeral is assigned to identical components regardless of drawing symbols, and redundant descriptions are omitted. Embodiments may be implemented in various forms of products, such as personal computers, laptops, tablets, smartphones, televisions, smart home appliances, intelligent vehicles, kiosks, and wearable devices.

[0020] FIG. 1 is a diagram illustrating an optimal prescription management system utilizing artificial intelligence recommendation according to one embodiment.

[0021] As shown in FIG. 1, an optimal prescription management system (400) utilizing artificial intelligence recommendations may be composed of multiple user terminals (110-1, ���), a server (120), and a database (130). In one embodiment, the database (130) is configured separately from the server (120), but the database (130) may be included within the server (120). For example, the server (120) may include multiple artificial intelligences for performing machine learning algorithms. According to another embodiment, the user terminals (110-1, ���), the server (120), and the database (130) may be connected to and communicate with each other through a network (N).

[0022] The network (N) may support wireless or wired communication between user terminals (110-1, 쪋), a server (120), and a database (130). For example, the network may perform wireless communication through methods such as LTE (long-term evolution), LTE-A (LTE Advanced), CDMA (code division multiple access), WCDMA (wideband CDMA), WiBro (Wireless BroadBand), WiFi (wireless fidelity), Bluetooth, NFC (near field communication), GPS (Global Positioning System), or GNSS (global navigation satellite system). The database (130) may store various data, and the stored data may include software (e.g., programs) as data collected, processed, and used by components of multiple user terminals (110-1, 쪋) or the server (120). The database (130) may include volatile and / or non-volatile memory.

[0023] In the present invention, Artificial Intelligence (AI) refers to a technology that mimics human learning ability, reasoning ability, and perceptual ability and implements them in a computer, and may include concepts such as machine learning and symbolic logic. Machine Learning (ML) is an algorithmic technology that classifies or learns the characteristics of input data on its own. Artificial intelligence technology can perform judgments or predictions by analyzing input data through machine learning algorithms and learning the results. Furthermore, a technology that mimics the cognitive and judgment functions of the human brain by utilizing machine learning algorithms can also be understood as falling within the category of artificial intelligence.

[0024] Machine learning refers to the process of training neural network models based on experience with data processing, through which computer software can improve its own data processing capabilities. Neural network models are constructed by modeling the correlations between data, and these correlations can be expressed by various parameters. The core of machine learning lies in optimizing the model's parameters by extracting and analyzing features from given data to derive correlations between them, and repeating this process. For example, a neural network model can learn the mapping (correlation) between inputs and outputs for data provided as input-output pairs, and can also learn relationships by deriving regularities between data even when only input data is provided. Artificial intelligence learning models, or neural network models, are designed to implement the structure of the human brain on a computer and can include multiple network nodes with weights that mimic neurons in a neural network. These multiple network nodes can be interconnected, simulating the synaptic activity where neurons exchange signals through synapses. The multiple network nodes of an artificial intelligence learning model are located in layers of different depths and can exchange data according to convolutional connections. Examples of artificial intelligence learning models include artificial neural networks and convolutional neural networks (CNNs). In one embodiment, artificial intelligence learning models can be machine learned according to methods such as supervised learning, unsupervised learning, and reinforcement learning.Algorithms for machine learning that can be used include Decision Trees, Bayesian Networks, Support Vector Machines, Artificial Neural Networks, Ada-boost, Perceptrons, Genetic Programming, and Clustering.

[0025] CNNs are a type of multilayer perceptron designed to use minimal preprocessing. CNNs consist of one or more convolutional layers and general artificial neural network layers placed on top of them, additionally utilizing weights and pooling layers. Thanks to this structure, CNNs can effectively process two-dimensional input data. Compared to other deep learning structures, CNNs demonstrate superior performance in the fields of image and speech and can be trained using standard backpropagation. A convolutional network is a neural network composed of a set of nodes with bounded parameters. As the amount of available training data increases and computational power improves, the combination of advancements in algorithms such as discriminative linear units and dropout training has significantly improved many computer vision tasks. Currently, the volume of available datasets is so vast for many tasks that outfitting is not critical, and increasing the network size can improve test accuracy.

[0027] FIG. 2 is a diagram illustrating the learning of a neural network according to one embodiment.

[0028] As shown in FIG. 2, the learning device can train a neural network (123) to process review responses received from a plurality of user terminals (110-1, ���) item by item. Additionally, the learning device can train a neural network (123) to extract user stay history from user movement path information. According to one embodiment, the learning device may be a separate entity from the server (120), but this is not limited thereto.

[0029] The neural network (123) includes an input layer (121) into which training samples are input and an output layer (125) into which training outputs are generated, and can be trained based on the difference between the training output and the label. Here, the label is defined according to the item corresponding to the review response and can be defined for the user's dwell history based on movement path information. The neural network (123) is connected by multiple groups of nodes and is defined by weights between the connected nodes and an activation function that activates the nodes.

[0030] The learning device can train the neural network (123) using the Gradient Descent (GD) technique or the Stochastic Gradient Descent (SGD) technique. The learning device can use a loss function designed based on the output and label of the neural network. The learning device can calculate the training error by utilizing a predefined loss function. The loss function can predefine the label, output, and parameters as input variables, where the parameters can be set as weights within the neural network (123). For example, the loss function can be designed in the form of Mean Square Error (MSE) or entropy, and various techniques or methods can be applied in designing the loss function.

[0031] The learning device can find weights that affect the training error through the backpropagation technique. Here, weights represent the relationships between nodes within the neural network (123). The learning device can use the SGD technique utilizing labels and outputs to optimize the weights found by the backpropagation technique. For example, the learning device can update the weights of the loss function defined according to the labels, outputs, and weights through the SGD technique.

[0033] FIG. 3 is a diagram illustrating the configuration of an artificial intelligence model according to one embodiment.

[0034] According to one embodiment, the artificial intelligence model may include an input layer, a hidden layer, and an output layer.

[0035] The input layer is the layer associated with the values ​​input into the artificial intelligence model. In the hidden layer, feature maps can be output by performing multiply-accumulate (MAC) operations and activation operations on the input values. The MAC operation is a process of multiplying the input value by its corresponding weight and then summing the resulting products. The activation operation is a process of inputting the result of the MAC operation into an activation function to output a final result; activation functions can be of various types. For example, activation functions may include sigmoid functions, tangent functions, ReLU functions, Leaky ReLU functions, Max Out functions, and / or ELU functions, but their types are not limited. The hidden layer may consist of at least one layer. For example, if the hidden layer consists of a first hidden layer and a second hidden layer, the first hidden layer performs MAC operations and activation operations based on the input values ​​of the input layer to output a feature map, and the feature map resulting from the first hidden layer can be used as the input value of the second hidden layer. The second hidden layer can perform MAC operations and activation operations according to the feature map resulting from the first hidden layer. The output layer may be a layer related to the result of the operations performed in the hidden layer.

[0036] In one embodiment, the learning model learns syllable (character) patterns that are frequently combined and used in a given corpus to automatically recognize the boundaries of compound words and named entities, and integrates object information from a first UI source with object information rendered in a browser to generate a learning object information file. By utilizing this learning object information file, data for training a deep learning network can be generated, and data can be received from various domains of a support system and standardized into an integrated format based on at least one standardization method corresponding to each domain. Data of a specific domain can be learned and inferred, information to be transmitted for standardization in that domain can be determined, and post-processing can be performed on data collected from various domains.

[0037] The first UI source includes an XML file, and the training object information file includes an input JSON file for feature learning and an output JSON file that serves as label data during training. This output JSON file is a file containing HTML DOM Tree information implemented in compliance with web standards. Various domains include at least one of a RAN (radio access network), a transport, or a core, and post-processing may include a correlation function.

[0039] Figure 4 is a block diagram showing the configuration of a system providing an artificial intelligence-based optimal prescription according to one embodiment.

[0040] A system (400) according to one embodiment may include a processor (420) and a memory (430), and some parts of the schematic configuration may be omitted or changed. According to one embodiment, the system (400) may operate as a server or a terminal. The processor (420) may be composed of one or more processors and may be structured to perform operations or data processing that control and / or communicate with each component of the system (400). The memory (430) may store information related to the method or store a program in which the method is implemented. The memory (430) may be volatile memory or non-volatile memory, and may store various file data, and the stored file data may be updated according to the operation of the processor (420).

[0041] According to one embodiment, the processor (420) can execute a program and control the system (400). The program code executed by the processor (420) can be stored in memory (430). The operation of the processor (420) can be performed by loading instructions stored in memory (430). The system (400) can be connected to an external device (e.g., a personal computer or a network) via an input / output device (not shown in the drawing) to exchange data.

[0042] According to one embodiment, when storing a health log, the system (400) may encrypt and store the student's personal information using the AES-256 encryption algorithm, and store symptom and treatment information as a hash value containing a digital signature. Generative artificial intelligence can automatically generate a health log summary report containing statistics by grade, class, and symptom on a daily basis by utilizing a GPT-based natural language processing model, and in particular, if two or more suspected cases of infectious disease occur in one class, the relevant content can be highlighted at the top of the report.

[0043] According to one embodiment, an automatic notification can be sent to the school nurse through a machine learning-based anomaly detection algorithm if the frequency of a specific symptom increases by more than 50% compared to the previous year, or if a student revisits three or more times with the same symptom.

[0044] According to one embodiment, the system (400) may store a student's personal information by encrypting it with the AES-256 encryption algorithm when storing a health log, which ensures the confidentiality of the data by using the strongest 256-bit key length among the Advanced Encryption Standards (AES). In addition, symptom and treatment information may be stored as a hash value containing a digital signature, which can be implemented by applying a hash function and adding a digital signature to ensure data integrity and authentication. Data stored in this manner can be safely protected from unauthorized external access or tampering.

[0045] According to one embodiment, the system (400) can automatically generate a health log summary report including statistics by grade, class, and symptom on a daily basis by utilizing a GPT-based natural language processing model. In this process, the generative artificial intelligence can analyze data recorded in the health log to identify statistical trends and convert them into natural language to write a report. In particular, if two or more suspected symptoms of an infectious disease occur in one class, the system (400) can highlight the relevant content at the top of the report, which enables a rapid response by immediately notifying the user if the frequency of a specific symptom is high.

[0046] According to one embodiment, the system (400) can send an automatic notification to the school nurse through a machine learning-based anomaly detection algorithm if the frequency of a specific symptom increases by more than 50% compared to the previous year, or if a student revisits three or more times with the same symptom. This algorithm can distinguish between normal patterns and anomalies by learning past data, and when an anomaly is detected, it generates a notification in real time to support the school nurse in responding quickly. This function may be useful for early detection of an epidemic of an infectious disease or a specific disease and for taking preventive measures.

[0048] FIG. 5 is a block diagram showing the configuration and function of a health system and the process of linking with other systems according to one embodiment.

[0049] According to one embodiment, the system (400) may include a main function (510) for a health log program (pro / mini) (e.g., Smile Health), a function through a kiosk (520), additional functions (512, 522), and a shopping mall management function (530). The main function (510) may include functions such as student patient registration, symptom registration, prescription registration, local prescription, checking past records, bed management, and item rental and return management, and may be composed of core functions for systematically handling various tasks occurring in the health room.

[0050] According to one embodiment, patient registration refers to a process of recording a student's basic information and purpose of visit, and symptom registration may include a function of systematically classifying and inputting the symptoms complained of by the student.

[0051] According to one embodiment, prescription registration is a function for recording medications or treatment methods prescribed by a school nurse, and local prescription can be implemented as a function for managing items that can be treated immediately in the infirmary. Checking past records can ensure continuity of care by querying a student's previous visit records, and bed management and item rental / return management can be designed as functions for efficiently operating resources within the infirmary.

[0052] According to one embodiment, the main function (510) may include additional functions such as content auto-completion, photo and memo registration, body diagram display, and vital information management. Content auto-completion can simplify repetitive input based on previous records, and photo and memo registration may include a function for visually recording symptoms or treatment processes. Body diagram display can intuitively represent the location of symptoms by graphically displaying body parts, and vital information management can be implemented as a function for systematically recording and managing vital signals such as body temperature, pulse, and blood pressure. In this case, vital information may include various vital signals measurable in the infirmary, and additional vital signals may be added as needed.

[0053] According to one embodiment, the kiosk function (520) may include functions such as patient registration, item return, and self-treatment, thereby enabling students to perform basic treatments or return items without the assistance of a school nurse. Patient registration may be implemented by having the student directly input information through the kiosk's touchscreen interface, and item return may be designed as a function that can be used when returning rented items. Self-treatment may include a function that guides the student to perform simple treatments (e.g., applying a bandage, disinfecting) on ​​their own, and this may be implemented by providing step-by-step instructions through the kiosk screen.

[0054] According to one embodiment, additional function 1 (512) may include functions such as psychological counseling, emergency situations, and log output, as well as infectious disease registration and protection survey and registration functions. Psychological counseling is a function for managing counseling records to support the mental health of students, and emergency situations may be implemented as a function for recording relevant information to enable a rapid response in the event of an emergency. Log output may include a function for printing the contents of a health log into a document for storage or sharing, and infectious disease registration may be designed as a function for systematically recording and managing relevant information in the event of an infectious disease outbreak. Protection survey and registration may include a function for separately managing information on students requiring special protection.

[0055] According to one embodiment, additional function 2 (522) may include management functions such as statistics, personnel registration, inventory management, prescription detail registration, and transfer of other health logs. The statistics function may provide a report by analyzing the status of health room usage, symptom distribution, and treatment frequency, and personnel registration may be implemented as a function to manage information on students and faculty using the health room. Inventory management is a function that allows checking and managing the quantity of items in the health room in real time, and prescription detail registration may include a function to record and manage prescription details in detail. Transfer of other health logs may be designed as a function that allows sharing or transferring information through data linkage with other systems or institutions.

[0056] According to one embodiment, the shopping mall (530) management function may include a function to systematically manage items such as medicines, supplies, medical devices, and educational aids, and this may be implemented as a function to comprehensively manage the purchase, storage, and usage history of items necessary for the operation of the health room. The medicine management function may systematically manage the expiration date, stock quantity, and usage records of medicines, and the supply and medical device management may monitor and manage the condition of various tools and equipment used within the health room. The educational aid management may include a function to systematically manage materials and tools used in health education, which may contribute to increasing the efficiency of health education.

[0057] These functions can operate independently and, if necessary, can be interconnected to implement an integrated health room management system. For example, patient registration information via a kiosk can be automatically reflected in the health log, and item management information can be linked with an inventory management system. Such linkage can contribute to increasing data consistency and work efficiency. The system (400) can improve the overall efficiency and accuracy of health room operations through these functions. All functions and linkage processes described herein are merely examples and may be modified in various ways according to the requirements and operational policies of the health room during actual implementation.

[0060] Figure 6 shows a flowchart of an optimal prescription management method utilizing artificial intelligence recommendation according to one embodiment.

[0061] Although process steps, method steps, algorithms, etc. are described in a sequential order in the flowchart of FIG. 6, such processes, methods, and algorithms may be configured to operate in any suitable order. In other words, the steps of the processes, methods, and algorithms described in various embodiments of the present invention do not need to be performed in the order described in the present invention. Furthermore, even if some steps are described as being performed asynchronously, in other embodiments, such steps may be performed simultaneously. Also, the example of a process by the depiction in the drawings does not mean that the exampled process excludes other variations and modifications thereof, nor does it mean that any of the exampled process or any of its steps is essential to one or more of the various embodiments of the present invention, nor does it mean that the exampled process is desirable.

[0062] In operation 610, the system (400) can receive symptom information and collect prescription data from a database.

[0063] According to one embodiment, the system (400) receives symptom information and can collect prescription data corresponding to the symptom from a database. At this time, the symptom information may include not only the patient's subjective symptoms but also objective data (e.g., body temperature, blood pressure, etc.), and such data is converted into a structured form through natural language processing technology and stored in the database. The database contains prescription history classified by symptom, and each prescription history includes the type of prescribed drug, dosage, number of administrations, date and time of prescription, patient age group, seasonal information, etc. Based on this data, the system (400) can analyze the correlation between the symptom and the prescription to extract an effective prescription pattern.

[0064] In operation 620, the system (400) can analyze the effectiveness of the prescription based on the collected prescription data.

[0065] According to one embodiment, the system (400) can analyze the effectiveness of a prescription based on collected prescription data. The effectiveness analysis is performed by calculating the ratio of the number of revisits to the total number of prescriptions for a specific symptom. For example, if drug A prescribed for a specific symptom resulted in 10 revisits out of a total of 100 cases, the revisit rate of the drug is calculated as 10%. In this case, the lower the revisit rate, the higher the effectiveness of the prescription is considered, and the system (400) can calculate an utility score for each prescription based on this revisit rate. In addition, effectiveness can be evaluated multidimensionally by utilizing additional indicators such as patient satisfaction and the incidence of side effects, in addition to the revisit rate. The figures mentioned are merely examples, and specific figures or types may vary depending on the settings.

[0066] In operation 630, the system (400) can select top prescriptions based on analyzed effectiveness to determine prescription priorities and generate a recommendation list.

[0067] According to one embodiment, the system (400) can determine prescription priorities and generate a recommendation list by selecting top prescriptions based on analyzed effectiveness. In this case, the prescription priority is determined by comprehensively considering various indicators such as prescription frequency, symptom similarity, and utility score. For example, a drug that is most frequently prescribed for a specific symptom and has a low revisit rate may be given a high priority. Additionally, the system (400) can synchronize recommendation rankings by reflecting prescription patterns collected in real-time from multiple user systems. Through this, customized prescription recommendations reflecting regional and seasonal characteristics may be possible.

[0068] After generating a recommendation list, the system (400) can display the generated list to the user. At this time, it can provide the reliability score and the basis for recommendation for each prescription to help the user select a prescription. For example, it can explain in detail why a specific prescription received a high utility score or why that prescription received high satisfaction from other users. In addition, the system (400) can analyze revisit data collected after the prescription is applied in real time to update the utility score for each prescription and readjust the recommendation ranking based on this. Through this dynamic adjustment function, the system (400) can continuously provide accurate prescription recommendations that reflect the latest data.

[0069] According to one embodiment, the system (400) can store and manage information on a patient's underlying disease and information on medications currently being taken in a database. Through this, it can analyze the risk of drug interactions and evaluate the association between the underlying disease and the prescribed medication to generate a comprehensive safety score. If this score is below a set threshold value, the prescription may be excluded from the recommendation list or its ranking may be adjusted. Additionally, the system (400) can analyze prescription review data collected from multiple users using natural language processing technology to extract information related to effectiveness, side effects, and satisfaction, and reflect this in the recommendation ranking in real time. Through these functions, the system (400) can provide safe and effective prescriptions to the user.

[0070] According to one embodiment, the system (400) may receive symptom information from a user. At this time, the user may directly input the symptoms they are experiencing into the system or select from a predefined list of symptoms. The symptom information entered by the user may be analyzed through natural language processing technology and converted into a structured form. The converted symptom information may be stored in a database and subsequently used for prescription recommendations.

[0071] The system (400) can collect prescription data corresponding to individual symptoms selected by the user from a database and assign priorities. To this end, the system can utilize a machine learning model that analyzes the association between symptoms and prescriptions. The model can learn from large-scale medical data to predict prescriptions effective for specific symptoms. A priority score for each prescription can be calculated based on the model's prediction results and actual prescription cases. A prescription with a higher priority score indicates a higher likelihood of being effective for the corresponding symptom.

[0072] Additionally, the system (400) can recommend a suitable prescription for the entire input symptom information. To this end, the system can analyze prescription patterns for various symptom combinations. By utilizing large-scale medical data, it can identify cases with similar symptom combinations and determine prescriptions that frequently appear in those cases. Furthermore, it can evaluate the effectiveness of each prescription by analyzing follow-up data on the prescription results. Through this, the system can select and prioritize prescriptions that are most suitable for the input symptom combination and have been verified for effectiveness.

[0073] When the prescription priority for the symptoms selected by the user and the prescription priority for all symptoms are determined, the system (400) can synthesize them to generate a final recommendation list. When generating the recommendation list, prescriptions corresponding to the symptoms selected by the user are placed at the top of the list. This is to provide customized recommendations that focus more on the symptoms selected by the user as their primary interest. After placing the prescriptions related to the symptoms selected by the user as the highest priority, the remaining recommended prescriptions can be arranged according to the prescription priority based on the results of the overall symptom analysis.

[0074] The recommendation list generated by the system (400) can be visualized and provided through a user interface. The user can view the recommendation list and access detailed information for each prescription. The prescription information may include the type of drug, usage, dosage, expected effects, precautions, etc. Additionally, the recommendation algorithm can be continuously improved by collecting user feedback.

[0075] The system (400) can provide personalized prescription recommendations based on the user's symptom information through this series of processes. By utilizing vast medical data and artificial intelligence technology, it can derive an optimal prescription combination that matches the user's characteristics and needs, and present it through a user-friendly interface, thereby enabling personalized medical services.

[0077] FIG. 7 is a flowchart illustrating an optimal prescription management method utilizing artificial intelligence recommendations according to one embodiment.

[0078] In operation 710, the system (400) can collect health log data including symptom data, prescription history, and whether there is a return visit, and calculate the prescription frequency.

[0079] According to one embodiment, symptom data includes subjective symptoms (e.g., headache, abdominal pain) and objective data (e.g., body temperature, blood pressure) entered by the patient when visiting the infirmary, and prescription history includes the type, dosage, number of administrations, and date and time of prescription of the prescribed drug. Revisit status records cases where the patient revisits with the same symptoms within 24 hours of the prescription, and this data is stored in a database in a structured form. The system (400) can calculate the prescription frequency by symptom based on the collected data. For example, if drug A prescribed for a specific symptom (e.g., headache) is used in 30 out of 100 cases, the prescription frequency of the drug is calculated as 30%. At this time, the prescription frequency can be calculated by assigning weights according to the passage of time, and higher weights can be applied to more recent data.

[0080] In operation 720, the system (400) can classify similar symptom groups by calculating the similarity between symptoms.

[0081] According to one embodiment, the system (400) can analyze the semantic similarity between symptom keywords by utilizing natural language processing technology to calculate the similarity between symptoms and classify similar symptom groups. For example, "headache" and "head pain" have similar meanings and can therefore be classified as the same symptom group. Additionally, the system (400) can quantify the similarity between symptom keywords through a text matching algorithm and group symptoms that exceed a set similarity threshold into the same group. This classification of similar symptom groups enhances the consistency of prescription patterns and supports the integrated analysis of prescription data for similar symptoms.

[0082] In operation 730, the system (400) can calculate a utility score per prescription based on the revisit rate within 24 hours of prescription and calculate a composite weight.

[0083] According to one embodiment, the revisit rate refers to the ratio of patients who revisit with the same symptoms after a specific prescription has been applied, and the lower this ratio, the higher the effectiveness of the prescription is considered. For example, if a specific drug causes revisits in 5 out of 100 cases, the revisit rate of the drug is calculated as 5%, and the utility score can be calculated as 95 points. The system (400) can calculate this utility score as a composite weight together with the prescription frequency and symptom similarity. The composite weight is calculated by applying a set ratio to each indicator (prescription frequency, symptom similarity, utility score), for example, the composite score can be calculated with a ratio of 40% for prescription frequency, 30% for symptom similarity, and 30% for utility score. At this time, the weight of each indicator can be configured to be adjustable by the system administrator.

[0084] In operation 740, the system (400) can synchronize recommendation rankings and generate a prescription list by reflecting prescription patterns collected in real time.

[0085] According to one embodiment, the system (400) can analyze prescription data collected from multiple user systems to extract prescription patterns that reflect regional and seasonal characteristics. For example, prescription patterns for symptoms that occur frequently in a specific region during a specific season can be analyzed in real time and reflected in the recommendation ranking.

[0086] According to one embodiment, the system (400) can execute a batch process every night to analyze all prescription data generated on the day and update existing prescription patterns. Through this real-time data synchronization, the system (400) can generate an accurate recommendation list that reflects the latest prescription trends. The generated recommendation list provides a reliability score and the basis for recommendation for each prescription, and can help the user select a prescription.

[0087] According to one embodiment, the system (400) can analyze revisit data collected after the application of a prescription in real time to update the utility score for each prescription and readjust the recommendation ranking based on this. Through this dynamic adjustment function, the system (400) can continuously provide the optimal prescription.

[0089] FIG. 8 is a flowchart illustrating an optimal prescription management method utilizing artificial intelligence recommendation according to one embodiment.

[0090] In operation 810, the system (400) can collect prescription data for a set period in chronological order and store the prescription history in a database.

[0091] According to one embodiment, the system (400) can perform data collection and storage in chronological order to efficiently manage and analyze prescription data. In this process, each prescription record may include information such as symptom name, treatment details, prescribed medication, time of visit, and time of discharge, and such prescription history may be indexed by treatment time period and stored in a database in a structure that enables fast searching. At this time, indexing may be implemented using a B-tree structure, and efficient searching with low time complexity may be possible.

[0092] In operation 820, the system (400) can automatically classify the collected prescription data by symptom.

[0093] According to one embodiment, the system (400) can automatically classify collected prescription data by symptom using natural language processing technology. In this process, the text matching algorithm may be implemented based on a word embedding model such as Word2Vec or FastText, and similarity may be determined by calculating cosine similarity between symptom keywords. When calculating similarity, weights may be assigned considering the semantic characteristics of the symptoms, and this may be adjusted to reflect the knowledge of medical professionals.

[0094] In operation 830, the system (400) can calculate the similarity between symptom keywords using a natural language processing-based text matching algorithm.

[0095] According to one embodiment, the system (400) can calculate similarity between symptom keywords using a text matching algorithm based on natural language processing, and for this purpose, it may utilize mathematical models such as cosine similarity or Jaccard similarity, and may apply a pre-trained language model (e.g., BERT, Word2Vec) to identify semantic similarity between symptoms. The system (400) can classify symptoms that exceed a set similarity threshold into the same group, wherein the similarity threshold can be dynamically adjusted by user settings or the system's automatic optimization algorithm, and can extract prescription patterns for each symptom group to identify prescriptions that are frequently used within that group or prescriptions effective for specific symptoms.

[0096] In operation 840, the system (400) can classify symptoms that exceed a set similarity threshold into the same group and extract a prescription pattern for each symptom group.

[0097] According to one embodiment, symptoms exceeding a set similarity threshold (e.g., 0.8) can be classified into the same group, and prescription patterns can be extracted for each symptom group. Prescription pattern extraction can be implemented using an Association Rule Mining algorithm, through which major prescription combinations and their frequency for a specific symptom group can be identified.

[0098] In operation 850, the system (400) can calculate the ratio of the number of revisits to the total number of prescriptions for a specific symptom and classify effective prescriptions.

[0099] According to one embodiment, the system (400) may calculate the ratio of the number of revisits to the total number of prescriptions for a specific symptom to evaluate the effectiveness of the prescription. At this time, a revisit rate threshold (e.g., 20%) may be set so that prescriptions showing a revisit rate lower than this threshold are classified as effective prescriptions. When calculating the revisit rate, time weighting may be applied to give higher importance to recent prescription results, thereby reflecting changes in the effectiveness of the prescription over time. This series of processing steps may be performed in parallel through a distributed processing system, and a stream processing architecture may be utilized for real-time data processing. Additionally, transaction management and data verification procedures may be included to ensure the accuracy and consistency of the data, and the system may be implemented as a microservices architecture to consider scalability.

[0102] According to one embodiment, the system (400) calculates the prescription frequency for each symptom and can calculate the time-weighted prescription frequency by applying weights set for each period to differentially reflect the importance of prescription data over time. This can more accurately reflect current prescription trends by giving higher weights to the latest prescription data.

[0103] According to one embodiment, the system (400) can calculate a composite score based on a set ratio for prescription frequency, effectiveness, and seasonality association to determine the priority of prescriptions, and the reflection ratio of each indicator can be adjusted according to the administrator's settings. Through this, it is possible to determine the priority by comprehensively considering various aspects of the prescription.

[0104] According to one embodiment, the system (400) can update existing prescription patterns by analyzing prescription data collected through a batch process that runs automatically at a set interval. At this time, the batch process can extract seasonal and time-of-day characteristics by performing a time-series analysis of prescription patterns by symptom. The system (400) can immediately evaluate the effectiveness of new prescription data input in real time and adjust the recommendation ranking of the corresponding prescription, and the adjusted ranking information can be synchronized in real time to the systems of multiple users through a distributed database.

[0105] According to one embodiment, the system (400) can calculate the prescription frequency for each symptom, and can calculate the time-weighted prescription frequency by applying weights set for each period to differentially reflect the importance of prescription data over time. This allows for adaptation to a real-time changing medical environment by more accurately reflecting current prescription trends through a method of assigning higher weights to the latest prescription data. For example, high weights can be assigned to prescription data from the last month, and relatively low weights can be applied to data older than six months to ensure the reliability of the data over time. Through this, the system (400) can provide analysis results based on the most recent prescription patterns for each symptom.

[0106] According to one embodiment, the system (400) can calculate a composite score based on a set ratio by considering various indicators such as prescription frequency, effectiveness, and seasonality association to determine the priority of prescriptions. The reflection ratio of each indicator can be flexibly adjusted according to the manager's settings, thereby allowing for a multifaceted evaluation of the priority of prescriptions for specific symptoms. For example, in the case of symptoms that occur frequently during a specific season, a high weight can be assigned to the seasonality association to prioritize the recommendation of prescriptions suitable for that period. In this way, the system (400) can determine the optimal prescription by comprehensively analyzing the correlation between symptoms and prescriptions.

[0107] According to one embodiment, the system (400) can continuously update existing prescription patterns by analyzing prescription data collected through a batch process that runs automatically at a set interval. This batch process can perform time-series analysis of prescription patterns by symptom to extract seasonal and time-of-day characteristics, thereby predicting and responding to patterns of symptoms occurring at specific times. Additionally, the system (400) can dynamically adjust the recommendation ranking of the corresponding prescription by immediately evaluating the effectiveness of new prescription data input in real time. The adjusted ranking information can be synchronized in real time to the systems of multiple users through a distributed database, ensuring that all users can share and utilize the latest prescription information.

[0108] These functions enable the system (400) to support new health teachers in real-time referencing the prescription patterns of experienced teachers in the region through an AI-based recommendation algorithm, thereby shortening the learning curve caused by lack of experience and improving prescription reliability. Additionally, by applying a multi-weighted recommendation system that combines symptom frequency and utility scores, the scientific standardization of prescriptions can be achieved by prioritizing the presentation of prescriptions that showed the highest efficiency for the same symptom over the past three months. The utility evaluation algorithm, which automatically analyzes patient revisit data after prescription, dynamically updates recommendation priorities and can continuously improve the long-term efficacy of specific drugs.

[0111] According to one embodiment, the system (400) can collect prescription data for the past three months in chronological order and store prescription history for each prescription record, including symptom name, treatment details, prescribed medication, time of arrival, and time of discharge, in a database. Through such structured data storage, systematic management of prescription history is possible. The system (400) can automatically classify the collected prescription data by symptom, and in this process, can group similar symptoms using a text matching algorithm and extract prescription patterns for each symptom group.

[0112] According to one embodiment, the system (400) can calculate the ratio of the number of revisits to the total number of prescriptions for the corresponding symptom regarding the effectiveness of the prescription, and can classify a prescription with a revisit rate of less than 20% within 24 hours of application of the prescription as an effective prescription. Through this, it is possible to evaluate the effectiveness of the prescription based on objective criteria.

[0113] According to one embodiment, the system (400) can collect prescription data for the past three months in chronological order, and each prescription record may include detailed information such as the name of the symptom, treatment details, prescribed medication, time of arrival, and time of discharge. This data is stored in a database in a structured form to enable systematic management of prescription history. The structured data storage method supports efficient search and analysis of data, thereby enabling the system (400) to quickly track and analyze the history of specific symptoms or prescriptions. For example, the prescription history for a specific symptom can be classified by time period to identify the occurrence pattern of the symptom.

[0114] According to one embodiment, the system (400) can automatically classify collected prescription data by symptom and, in this process, group similar symptoms by utilizing a text matching algorithm. The text matching algorithm is based on natural language processing technology and calculates similarity by comparing the input symptom name with the symptom name in the database. Through this, identical or similar symptoms can be grouped into one, and prescription patterns can be extracted for each symptom group. For example, similar symptom names such as "headache" and "head pain" can be integrated into one group to analyze the prescription pattern for that symptom. In this way, the system (400) can derive a more accurate prescription pattern based on the integrated data by symptom.

[0115] According to one embodiment, the system (400) can calculate the ratio of the number of revisits to the total number of prescriptions for the corresponding symptom to evaluate the effectiveness of the prescription. In this case, the revisit rate can be defined as the ratio of patients who revisit with the same symptom within 24 hours of the prescription being applied, and the system (400) can classify a prescription with a revisit rate of less than 20% as an effective prescription. Through these objective criteria, the system (400) can evaluate whether a specific prescription has actually contributed to the improvement of symptoms. For example, if the revisit rate is low after prescribing a specific drug, it can be determined that the drug is effective in alleviating symptoms, thereby scientifically verifying the efficacy of the prescription. These evaluation results are subsequently reflected in a prescription recommendation algorithm to support the priority recommendation of more effective prescriptions.

[0116] These functions enable the system (400) to perform systematic prescription history management through structured data storage and analysis, and to accurately extract prescription patterns for similar symptoms through symptom grouping using a text matching algorithm. In addition, the utility of a specific prescription can be verified through an objective evaluation of prescription effectiveness based on the revisit rate, thereby establishing a reliable prescription recommendation system.

[0119] According to one embodiment, the system (400) calculates the prescription frequency for each symptom and can calculate the time-weighted prescription frequency by assigning a weight of 0.5 to the prescription frequency within the last 1 month and a weight of 0.3 to the prescription frequency within 2-3 months. Through the application of such differential weights, higher importance can be given to the latest prescription information.

[0120] According to one embodiment, the system (400) can calculate a composite score with a ratio of 40% for prescription frequency score, 40% for effectiveness score, and 20% for seasonal association to determine the priority of prescriptions. Additionally, the system (400) can update existing prescription patterns by analyzing all prescription data generated on the day through a batch process that runs automatically at 23:00 every day, and can immediately evaluate the effectiveness of new prescription data entered in real time to adjust the recommendation ranking of the corresponding prescription, and reflect the adjusted ranking in real time on the screens of other users.

[0121] According to one embodiment, the system (400) can calculate the prescription frequency for each symptom, and to differentially reflect the importance of prescription data over time, it can calculate the prescription frequency with applied time weights by assigning a weight of 0.5 to the prescription frequency within the last month and a weight of 0.3 to the prescription frequency within 2-3 months. The weights are merely examples and may vary depending on the settings. This differential weighting method can more accurately reflect current medical trends or changes in symptoms by giving higher importance to the latest data. For example, if the prescription frequency for a specific symptom has surged over the last month, this is likely to reflect external factors such as seasonal factors or the epidemic of a specific disease, and the system (400) can derive a more realistic and appropriate prescription pattern by reflecting this latest data at a higher rate. This application of time weights ensures the freshness and reliability of the data, while simultaneously maintaining long-term patterns by considering past data to a certain extent.

[0122] According to one embodiment, the system (400) can calculate a composite score with a ratio of 40% for prescription frequency, 40% for effectiveness, and 20% for seasonal association to determine the priority of prescriptions. The prescription frequency score indicates how frequently a prescription for a specific symptom is made, and the effectiveness score evaluates how much the prescription contributed to symptom improvement through indicators such as the revisit rate. The seasonal association assigns a score by considering whether a specific symptom occurs more frequently in a specific season. This method of calculating a composite score by combining multiple indicators allows for a comprehensive evaluation of various aspects of a prescription that are difficult to grasp with a single indicator alone. For example, if a specific prescription has high frequency but low effectiveness, the system (400) can adjust it to a lower priority, and conversely, if a prescription has low frequency but high effectiveness, it can be raised to a higher priority. In this way, the system (400) can determine the priority of prescriptions according to scientific and objective criteria.

[0123] According to one embodiment, the system (400) can update existing prescription patterns by analyzing all prescription data generated on the day through a batch process that runs automatically at 23:00 every day. This batch process can derive the latest prescription patterns by comprehensively analyzing data such as prescription frequency by symptom, effectiveness, and seasonality association, thereby enabling the system (400) to adapt to a medical environment that changes in real time. Additionally, the system (400) can immediately evaluate the effectiveness of new prescription data entered in real time. For example, when a specific prescription is entered, the system (400) can calculate the revisit rate of the prescription, evaluate its effectiveness by comparing it with existing data, and then dynamically adjust the recommendation ranking of the prescription. The adjusted ranking information can be reflected in real time on other users' screens through a distributed database, thereby allowing all users to share and utilize the latest prescription information. These real-time data processing and synchronization functions can play a key role in enabling the system (400) to provide rapid and accurate prescription recommendations.

[0124] Through this, the system (400) can provide reliable prescription recommendations to users and, in particular, support consistent and scientific prescription decisions, even for new health teachers or users with little experience.

[0127] According to one embodiment, the system (400) may construct a multidimensional symptom database, which includes patient-specific symptom keywords, age group classification information, time of visit, and seasonal information as data fields, and each field may be configured to be indexed and searchable. Additionally, the system (400) may construct a prescription history database, which includes information on the type and dosage of prescribed drugs and administration cycle information, and the drug information may be managed by mapping it to a drug code.

[0128] According to one embodiment, the system (400) can collect revisit data due to the same symptoms within a set monitoring period, and the identity of symptoms for determining revisit can be determined through text similarity analysis based on natural language processing. The system (400) can analyze semantic similarity based on word embedding for input symptom keywords and classify symptoms whose similarity score exceeds a set threshold into the same symptom group. Additionally, the system (400) can determine drug suitability by patient age group by distinguishing between elementary school students and other age groups, and can filter out unsuitable drugs by comparing with administration standard information by age group registered in the drug database.

[0129] According to one embodiment, the system (400) may construct a multidimensional symptom database, which may include various data fields such as patient-specific symptom keywords, age group classification information, time of visit, and seasonal information. Each data field may be indexed to facilitate searching, thereby enabling the rapid extraction and analysis of data regarding specific symptoms or conditions. For example, symptoms that occur frequently in a specific age group during a specific season may be searched, or symptoms that primarily occur during a specific time period may be analyzed. This multidimensional database can contribute to deriving more accurate prescription patterns by comprehensively considering various factors related to symptoms. Additionally, the system (400) may construct a prescription history database, which may include details such as the type of prescribed drug, dosage information, and administration cycle information. Drug information may be managed by mapping it to drug codes, which allows tracking the usage history of specific drugs or analyzing drug interactions. This prescription history database can play an important role in maintaining the consistency and reliability of prescriptions.

[0130] According to one embodiment, the system (400) can collect revisit data due to the same symptom within a set monitoring period, and the identity of the symptom for determining revisit can be determined through text similarity analysis based on natural language processing. Natural language processing technology can group similar symptoms by analyzing the meaning of input symptom keywords, thereby defining a group of the same symptom. For example, similar symptom keywords such as "headache" and "head pain" can be classified into a group of the same symptom. The system (400) can perform semantic similarity analysis based on word embedding, which is a method of calculating similarity between words by expressing the meaning of words in a vector space. Symptoms whose similarity score exceeds a set threshold can be classified into a group of the same symptom, thereby increasing the accuracy of the revisit data. Such text similarity analysis enables a more accurate determination of revisit by considering the diversity of symptoms.

[0131] According to one embodiment, the system (400) can determine drug suitability by distinguishing between elementary school students and other age groups, and can filter out unsuitable drugs by comparing them with age-specific administration standard information registered in the drug database. For example, if a specific drug is prohibited or not recommended for use in a specific age group, the system (400) can automatically identify this and exclude the drug from the prescription list. This determination of drug suitability by age group can play an important role in ensuring patient safety and preventing side effects. Additionally, the system (400) can analyze drug interactions based on information registered in the drug database, thereby preventing risks that may occur when prescribing combination drugs in advance.

[0132] The system (400) utilizes these functions to build a multidimensional symptom database and a prescription history database to support systematic data management, and can accurately determine the identity of symptoms through text similarity analysis based on natural language processing. In addition, it can distinguish between elementary school students and other age groups to determine drug suitability for each patient age group and provide safe prescriptions.

[0135] According to one embodiment, the system (400) can determine whether a disease is seasonal by analyzing the correlation between symptoms and the timing of occurrence, and can extract a pattern of frequency of occurrence in a specific season or period through time series analysis. This enables a preemptive response to seasonal diseases. The system (400) can calculate an effectiveness score per prescription based on revisit data and can quantify effectiveness based on the ratio of revisit cases to the total number of prescriptions.

[0136] According to one embodiment, the system (400) can calculate a composite score by applying weights set for prescription frequency, symptom similarity, and utility score, and the weights of each indicator can be configured to be adjustable by a system administrator. The system (400) can correct recommendation rankings by analyzing real-time prescription data collected at the regional level, and can reflect prescription preferences reflecting regional characteristics in the ranking determination.

[0137] According to one embodiment, the system (400) can generate a recommendation list by selecting prescriptions with the highest composite scores among prescriptions excluding unsuitable drugs by age group, and can provide a reliability score and the basis for recommendation for each prescription. The system (400) can update the efficacy score for each prescription and readjust the recommendation ranking in real time whenever new prescription result data is collected, and the updated information can be synchronized to all user systems through a distributed database.

[0138] According to one embodiment, the system (400) can determine whether a condition is a seasonal disease by analyzing the correlation between symptoms and the timing of occurrence, and may utilize a time series analysis technique for this purpose. Time series analysis is a method of identifying patterns or trends by analyzing data over a specific period in chronological order, and the system (400) can extract a pattern of symptom occurrence frequency in a specific season or period through this. For example, if a specific symptom occurs frequently during the winter, it can be determined as a seasonal disease, and a preemptive response strategy for that period can be established. This analysis supports preventive prescriptions or resource allocation for seasonal diseases, thereby enabling the provision of more efficient medical services.

[0139] According to one embodiment, the system (400) can calculate a utility score for each prescription based on revisit data, and can quantify the utility by calculating the ratio of the number of revisits to the total number of prescriptions. For example, if 100 specific prescriptions are made and 10 of them lead to revisits, the revisit rate can be calculated as 10%. The lower this figure, the higher the utility of the prescription can be judged, and the system (400) can evaluate the effectiveness of the prescription based on objective criteria through this. This utility score is subsequently reflected in a prescription recommendation algorithm to support the priority recommendation of more effective prescriptions.

[0140] According to one embodiment, the system (400) can calculate a composite score by applying weights set to various indicators such as prescription frequency, symptom similarity, and utility score. The weights of each indicator can be configured to be adjustable by the system administrator, thereby allowing the prescription recommendation algorithm to be flexibly adjusted to suit specific situations or purposes. For example, if prescriptions with high symptom similarity need to be given more importance in a specific region, the weight for symptom similarity can be increased. This method of calculating a composite score enables the comprehensive evaluation of various aspects of a prescription that are difficult to grasp with a single indicator alone.

[0141] According to one embodiment, the system (400) can correct recommendation rankings by analyzing real-time prescription data collected at the regional level and can reflect prescription preferences reflecting regional characteristics in the ranking determination. For example, if there is a high preference for a specific drug in a specific region, this can be reflected in the recommendation ranking to provide prescriptions more suitable for users in that region. Such regional data analysis has the effect of enabling prescription recommendations that more accurately reflect the user's needs and environment.

[0142] According to one embodiment, the system (400) can generate a recommendation list by selecting prescriptions with the highest composite scores among prescriptions excluding unsuitable drugs by age group, and can provide a reliability score and the basis for recommendation for each prescription. The reliability score can be calculated by comprehensively evaluating the efficacy, frequency, and symptom similarity of the prescription, and the basis for recommendation can clearly explain the reason why the prescription was recommended. For example, by informing the user that a specific prescription was recommended based on a high efficacy score and symptom similarity, it supports the user in trusting and accepting the prescription.

[0143] According to one embodiment, the system (400) can update the utility score for each prescription and readjust the recommendation ranking in real time whenever new prescription result data is collected. This enables the system (400) to continuously improve the prescription recommendation algorithm based on the latest data. The updated information can be synchronized to all user systems through a distributed database, allowing all users to share and utilize the latest prescription information in real time. This real-time data synchronization function can play a key role in enabling the system (400) to provide rapid and accurate prescription recommendations. Through these functions, the system (400) can provide users with reliable and scientific prescription information and, in particular, support consistent and effective prescription decisions, even for new health teachers or users with limited experience.

[0146] According to one embodiment, the system (400) may receive multidimensional symptom data including symptom keywords, patient age group, time of visit, and seasonal information, and may collect prescription history including the type, dosage, and number of administrations of prescribed drugs, and revisit data including whether the patient revisits due to the same symptoms within 24 hours of the prescription.

[0147] According to one embodiment, the system (400) can collect prescription history for the same or similar symptoms within a specific period and can analyze semantic similarity between symptom keywords using natural language processing technology. Additionally, the system (400) can distinguish between elementary school students and other age groups to determine drug suitability by patient age group and determine whether it is a seasonal disease. The system (400) can calculate an efficacy score per prescription using the formula [(1 - (number of revisiting patients / total number of patients to whom the corresponding prescription was applied)) × 100] and can calculate a composite score by applying a weight of 40% to the prescription frequency, 30% to the symptom similarity, and 30% to the efficacy score.

[0148] According to one embodiment, the system (400) can correct the ranking by reflecting the results of real-time prescription pattern analysis by region, and can select and display the top 5 prescriptions in order of highest composite score, excluding contraindicated drugs by age group. The system (400) can provide a reliability score and prescription basis for each prescription together, and can update the efficacy score for each prescription by reflecting revisit data collected after the prescription is applied in real time, and readjust the recommendation ranking in real time based on this.

[0149] According to one embodiment, the system (400) may receive multidimensional symptom data such as symptom keywords, patient age group, time of visit, and seasonal information, and this data may be used to analyze the pattern of symptom occurrence and recommend a prescription suitable for specific conditions. For example, it may identify symptoms that occur frequently in a specific age group during a specific season, or analyze symptoms that mainly occur at a specific time. In addition, the system (400) may collect prescription history data such as the type, dosage, and frequency of administration of the prescribed drug, and may collect revisit data including whether the patient revisits due to the same symptom within 24 hours of the prescription. Revisit data may be used as an important indicator to evaluate how effective a specific prescription is in improving symptoms, thereby allowing for an objective analysis of the efficacy of the prescription.

[0150] According to one embodiment, the system (400) can collect prescription history for the same or similar symptoms within a specific period and analyze prescription patterns for specific symptoms. At this time, natural language processing technology can be utilized to analyze semantic similarity between symptom keywords, thereby integrating similar symptoms such as "headache" and "head pain" into a single group. Natural language processing technology calculates similarity between words by representing the meaning of words in a vector space, thereby enabling accurate analysis that considers the diversity of symptom keywords. Additionally, the system (400) can distinguish between elementary school students and other age groups to determine drug suitability by patient age group and can filter out unsuitable drugs by comparing them with administration standard information by age group registered in the drug database. For example, if a specific drug is prohibited or not recommended for use in a specific age group, the system (400) can automatically identify this and exclude the drug from the prescription list. Along with this, the system (400) can determine whether a disease is seasonal and classify it as a seasonal disease by analyzing whether a specific symptom occurs frequently during a specific season. This can enable a preemptive response to seasonal diseases.

[0151] According to one embodiment, the system (400) can correct the recommendation ranking by reflecting the results of real-time prescription pattern analysis by region, thereby allowing prescription preferences reflecting regional characteristics to be reflected in the ranking determination. For example, if there is a high preference for a specific drug in a specific region, this can be reflected in the recommendation ranking to provide a prescription that is more suitable for users in that region. Additionally, the system (400) can select and display the top 5 prescriptions in order of highest composite score, excluding drugs contraindicated by age group, and can provide a reliability score and the basis for the prescription for each prescription. The reliability score can be calculated by comprehensively evaluating the efficacy, frequency, symptom similarity, etc. of the prescription, and the basis for the prescription can clearly explain the reason why the prescription was recommended. Through this, users can trust and accept the prescription.

[0152] According to one embodiment, the system (400) can update the utility score for each prescription by reflecting revisit data collected after the prescription is applied in real time, and can readjust the recommendation ranking in real time based on this. This enables the system (400) to continuously improve the prescription recommendation algorithm based on the latest data. The updated information can be synchronized to all user systems through a distributed database, thereby allowing all users to share and utilize the latest prescription information in real time. This real-time data synchronization function can play a key role in enabling the system (400) to provide rapid and accurate prescription recommendations.

[0155] FIG. 9 illustrates a user interface (UI) that provides a prescription method to enable optimal prescription in an optimal prescription management system utilizing artificial intelligence recommendation according to one embodiment.

[0156] According to one embodiment, the system (400) can provide prescription management functions through an intuitive and efficient user interface, and this interface is designed to be easily understood and utilized by the user. On the left, a list of visiting students may be displayed in chronological order, which is sorted in the order students visited the infirmary to support the school nurse in quickly identifying students currently waiting. In the central area, the symptom input and prescription records of the selected student may be displayed, and this area serves as a space to record the student's symptoms in detail and check the prescription history accordingly. The system (400) may recommend an AI-optimized prescription based on the entered symptoms, which may be displayed in the prescription list area on the right. The right area displays a list of prescriptions recommended according to the symptoms, and each prescription may be provided along with information such as effectiveness, safety, and frequency of use.

[0157] According to one embodiment, the system (400) can implement a user interface for optimal prescription management utilizing artificial intelligence recommendations. The user interface can be configured by dividing the main screen into three main areas. In the left area, a list of visiting students sorted in chronological order may be displayed, and for each student, the time of visit and major symptoms may be displayed together to enable rapid identification of the patient.

[0158] According to one embodiment, the central area can be utilized as a main workspace for symptom input and prescription management. In this area, an intuitive interface for symptom input may be provided, and frequently occurring symptoms may be provided in the form of quick buttons to enable rapid input. For example, general symptoms such as 'headache', 'abdominal pain', and 'fever' may be entered immediately with just a button click, and detailed symptoms may be added through text input.

[0159] According to one embodiment, an AI-based prescription recommendation list may be displayed in the right area. The system (400) can analyze the input symptoms and recommend the optimal prescription in real time, and a reliability score and the basis for the recommendation for each prescription may be presented together. For example, in the case of a student complaining of a headache, the system may display prescriptions such as "Take 1 tablet of Tylenol 500mg (95% reliability)" and "Observe after resting for 15 minutes (88% reliability)" according to priority.

[0160] According to one embodiment, the system (400) can provide treatment guidelines for each symptom to support the work of new school nurses. For example, in the case of fever symptoms, a standard protocol such as "measuring body temperature - checking criteria for administering antipyretics - selecting medication - observing progress after administration" may be presented step by step. In addition, the system can analyze the prescription patterns of school nurses at other schools and provide statistical insights such as "prescriptions selected by 80% of school nurses for the same symptom."

[0161] According to one embodiment, the system (400) may provide a visualized timeline view for managing prescription history. This allows for a quick overview of a specific student's past visit history, trends in symptom changes, and the effectiveness of prescriptions. Additionally, precautions such as allergy information or underlying diseases may be highlighted with warning icons to support safe prescribing.

[0162] According to one embodiment, the user interface is implemented with a responsive design based on HTML5 and CSS3, so that it can provide an optimized screen not only on PCs but also on tablets or mobile devices. Important information can be intuitively distinguished through color coding and can be displayed by distinguishing it according to urgency, such as red (urgent), yellow (caution), and green (general).

[0163] According to one embodiment, the system (400) can continuously update the latest prescription trends and effectiveness information through real-time data synchronization. For example, if a new prescription pattern for a specific symptom is discovered or the effectiveness of an existing prescription is verified, this is immediately reflected in the recommendation system, which may enable more reliable prescription recommendations.

[0164] According to one embodiment, the system (400) can provide support functions for new school nurses. The system (400) can recommend the optimal prescription by analyzing the prescription patterns of other school nurses for each symptom, which can significantly shorten the learning curve of new school nurses. For example, regarding a student complaining of a headache, the system (400) can recommend the most effective treatment method and medication by analyzing prescription data from school nurses at other schools. In this case, the system (400) can prioritize recommending prescriptions for headache symptoms that have a low return-visit rate and high user satisfaction. Additionally, the system (400) can provide detailed information, such as the frequency of use and effectiveness evaluations by other school nurses, along with the basis for the prescription, so that new school nurses can refer to it when deciding on a prescription. Through this, new school nurses can decide on prescriptions based on systematic and scientific evidence, even if they lack experience.

[0165] According to one embodiment, the system (400) can provide visualized prescription history information and can also display detailed information and precautions for each prescription. For example, by visualizing a specific student's past prescription history in the form of a graph or chart, it supports the school nurse in grasping changes in the student's health status at a glance. Additionally, the system (400) can support safe prescriptions by considering the student's past prescription history and allergy information. For example, if a specific student has a record of an allergic reaction to penicillin-based drugs, the system (400) can exclude drugs containing that ingredient from the recommendation list or display a warning message. These functions help the school nurse make safe and appropriate prescriptions by considering the student's individual health status and history.

[0166] According to one embodiment, the user interface of the system (400) is designed to be responsive so that it can be used seamlessly on various devices and can effectively convey key information through intuitive icons and color coding. For example, if an urgent prescription is required, the relevant information may be highlighted in red, and general prescriptions may be displayed in blue to help the school nurse quickly understand the situation and respond. In addition, the latest prescription trends and effectiveness information can be continuously reflected through real-time updates. For example, if a specific drug has recently been evaluated as highly effective, the system (400) may display that drug at the top of the recommendation list. This allows the school nurse to always make prescription decisions based on the latest medical information.

[0167] According to one embodiment, the system (400) supports the provision of optimal prescriptions to school nurses through an intuitive and efficient user interface, and can particularly contribute to shortening the learning curve of new school nurses and increasing the consistency and reliability of prescriptions. It also supports safe prescriptions that take into account the individual health status and history of students, and enables prescription recommendations that reflect the latest medical information through real-time updates. Through this, the system (400) can optimize the prescription management process of the school health room and support the health management of students in a more scientific and systematic way.

[0170] According to one embodiment, the system (400) can store and manage information on the user's underlying disease and information on the medication currently being taken in a database, and the medication information can be stored in a structured form mapped to an International Drug Identification Code. The system (400) can analyze the risk of interaction with prescription data collected from the database and can identify combinations of ingredients that may interact by comparing the ingredient information of the medication being taken with the medication ingredient database.

[0171] According to one embodiment, the system (400) can input information on the ingredients, dosage, and dosing cycle of a prescribed drug into a convolutional neural network-based drug interaction prediction model to calculate a risk grade according to established criteria and analyze the association between the underlying disease and the prescribed drug. At this time, the risk can be evaluated based on information on the interaction between the disease and the drug extracted from a medical literature database.

[0172] According to one embodiment, the system (400) can generate a comprehensive safety score for each prescription by combining the drug interaction risk grade and the results of the underlying disease association analysis according to set weights, and can adjust the ranking within the recommendation list for prescriptions with a comprehensive safety score below a set threshold value. At this time, the ranking can be recalculated by reflecting the correlation between the safety score and the recommendation ranking.

[0173] According to one embodiment, the system (400) can generate a warning message by analyzing risk factors for prescriptions that exceed a set risk standard, and the message can be visually highlighted and displayed on a user interface along with specific details of the risk factors. The system (400) can extract information related to effectiveness, side effects, and satisfaction by analyzing prescription review data collected from multiple users through natural language processing technology, calculate a comprehensive evaluation index by applying set weights to the extracted information, and reflect this in the recommendation ranking in real time.

[0174] According to one embodiment, the system (400) can store and manage information on the user's underlying disease and information on drugs currently being taken in a database. In this case, the drug information can be stored in a structured form mapped to the International Drug Identification Code (IDIC). The structured data storage method supports the efficient search and analysis of drug information, thereby enabling the system (400) to quickly extract detailed information such as the ingredients, dosage, and dosing cycle of a specific drug. Additionally, the system (400) can analyze the risk of interaction with prescription data collected from the database. To this end, it can identify combinations of ingredients with potential interactions by comparing the ingredient information of the drugs being taken with the drug ingredient database. For example, it can predict in advance side effects or drug interactions that may occur when specific drug A and drug B are taken simultaneously. Such analysis of interaction risks can play an important role in ensuring patient safety and preventing side effects.

[0175] According to one embodiment, the system (400) can input information on the ingredients, dosage, and dosing cycle of the medication into a drug interaction prediction model based on a Convolutional Neural Network (CNN) to calculate a risk grade according to established criteria. A Convolutional Neural Network is a type of deep learning technology that is effective in learning and predicting complex data patterns. Through this, the system (400) can quantitatively evaluate the risk of drug interactions, and the risk grade can be classified into stages such as low, medium, and high. Additionally, the system (400) can analyze the association between an underlying disease and a prescribed medication, and in this case, can evaluate the risk based on information on the interaction between the disease and the medication extracted from a medical literature database. For example, the risk that may occur when prescribing a specific medication to a patient with a specific underlying disease can be evaluated by referring to medical literature data. This analysis can be used as an important criterion for determining the suitability between the patient's underlying disease and the prescribed medication.

[0176] According to one embodiment, the system (400) can generate a comprehensive safety score for each prescription by combining the drug interaction risk grade and the results of the underlying disease association analysis according to set weights, and this score can be used as an indicator to comprehensively evaluate the safety of a specific prescription. For example, if the drug interaction risk grade is high and the risk of association with an underlying disease is high, the comprehensive safety score of the prescription may be calculated as low. The system (400) can adjust the ranking within the recommendation list for prescriptions with a comprehensive safety score below a set threshold value, and in this case, the ranking may be recalculated by reflecting the correlation between the safety score and the recommendation ranking. For example, prescriptions with a low safety score may be adjusted to a lower position in the recommendation ranking, thereby allowing safer prescriptions to be provided to the user preferentially.

[0177] According to one embodiment, the system (400) can analyze risk factors for prescriptions that exceed set risk criteria and generate a warning message, and the message can be visually highlighted and displayed on a user interface along with specific details of the risk factors. For example, if a specific prescription has a high risk of drug interaction or a high risk of association with an underlying disease, the system (400) can generate a warning message for the prescription and clearly notify the user of this. This warning message helps the user consider risk factors when selecting a prescription, thereby ensuring patient safety.

[0178] According to one embodiment, the system (400) can extract information related to effectiveness, side effects, and satisfaction by analyzing prescription review data collected from multiple users through natural language processing technology, and thereby evaluate the utility of the prescription based on the users' actual experiences. Natural language processing technology can identify specific keywords or contexts by analyzing text data of user reviews, for example, by identifying keywords such as "it was effective" or "there were side effects." The extracted information can be calculated as a comprehensive evaluation indicator by applying set weights, and this indicator can be reflected in the recommendation ranking in real time. For example, if user satisfaction with a specific prescription is high, the recommendation ranking of that prescription can be adjusted upward. Such user review-based evaluation can support the establishment of a more reliable recommendation system by reflecting the actual effectiveness of the prescription and user experience.

Claims

Claim 1 An optimal prescription management system utilizing artificial intelligence recommendations includes a memory for storing instructions and a processor. When the instructions are executed by the processor, the system receives symptom information from a user, collects prescriptions corresponding to individual symptom information selected by the user from a database and assigns priorities, collects prescription data corresponding to all symptom information to analyze the effectiveness of prescriptions and selects top prescriptions to determine prescription priorities, combines the prescription priority for user-selected symptoms with the prescription priority for all symptoms to generate a final recommendation list, places prescriptions corresponding to user-selected symptoms as the highest priority and then arranges the remaining prescriptions according to the overall priority to generate a recommendation list containing a set number of prescriptions, controls the display of the generated recommendation list through a user interface, collects health log data including symptom data, prescription history, and whether there was a return visit after the prescription, calculates the prescription frequency by symptom based on the collected health log data, calculates similarity between symptoms to classify similar symptom groups, calculates a utility score by prescription based on the return visit rate within a specified time after the prescription, and the prescription frequency, symptom similarity, and utility A system that calculates composite weights for scores, synchronizes recommendation rankings by reflecting prescription patterns collected in real-time from multiple user systems, compares and analyzes the prescription frequency and utility scores of multiple users for the same symptom within a preset period, generates a list of the top 5 prescriptions based on the composite weights and prescription patterns, displays reliability scores for each prescription in the generated list, continuously collects and analyzes revisit data for the corresponding symptom after the prescription is applied to dynamically adjust utility weights per prescription, and controls the system to reflect the adjusted weights in the recommendation rankings in real-time. Claim 2 delete Claim 3 In claim 1, the above instructions, when executed by the processor, the system collects prescription data for a preset period in chronological order, stores a prescription history for each prescription record including a symptom name, treatment details, prescribed medication, time of visit, and time of discharge in a database, wherein the prescription history is structured to be indexed and searchable by treatment time period, automatically classifies the collected prescription data by symptom, calculates similarity between symptom keywords using a natural language processing-based text matching algorithm, classifies symptoms exceeding a preset similarity threshold into the same group, extracts prescription patterns for each symptom group, calculates the ratio of the number of revisits to the total number of prescriptions for a specific symptom to evaluate the effectiveness of the prescription, and controls the system to classify prescriptions below a preset revisit rate threshold as effective prescriptions. Claim 4 In claim 1, the above instructions, when executed by the processor, the system calculates the prescription frequency for each symptom, calculates the time-weighted prescription frequency by applying weights set for each period to differentially reflect the importance of prescription data over time, calculates a composite score based on ratios set for prescription frequency, effectiveness, and seasonality correlation to determine the priority of prescriptions, wherein the reflection ratio of each indicator can be adjusted according to the administrator settings, and updates existing prescription patterns by analyzing collected prescription data through a batch process that is automatically executed at a set interval, wherein the batch process performs time-series analysis of the prescription pattern by symptom to extract seasonal and time-of-day characteristics, immediately evaluates the effectiveness of new prescription data input in real time to adjust the recommendation ranking of the corresponding prescription, and controls the adjusted ranking information to be synchronized with the systems of multiple users through a distributed database. Claim 5 In claim 1, the above instructions, when executed by the processor, control the system to collect prescription data for a preset period in chronological order, store prescription history including symptom name, treatment details, prescribed medication, time of visit, and time of discharge in a database, automatically classify the collected prescription data by symptom, calculate the effectiveness of the prescription as the ratio of the number of revisits to the total number of prescriptions for the corresponding symptom, and classify prescriptions with a revisit rate below a preset threshold within a preset monitoring period as effective prescriptions. Claim 6 In claim 1, the above instructions, when executed by the processor, are a system that calculates the prescription frequency for each symptom over the past three months to generate a prescription ranking, calculates a composite score based on the ratio set for the prescription frequency score, effectiveness score, and seasonality correlation over the past three months to determine the priority of prescriptions, analyzes prescription data generated on the day through a batch process that is automatically executed at a set time to update existing prescription patterns, immediately evaluates the effectiveness of new prescription data input in real time to adjust the recommendation ranking of prescriptions, and controls the system to reflect the said ranking on another user's screen in real time. Claim 7 In claim 1, the above instructions, when executed by the processor, construct a multidimensional symptom database of the system, wherein the database includes patient-specific symptom keywords, age group classification information including whether the patient is an elementary school student, time of visit, and seasonal information as data fields, and each field is configured to be indexed and searchable; construct a prescription history database, wherein the database includes information on the type and dosage of prescribed drugs and administration cycle information, and drug information is managed by mapping to drug codes; collect data on revisits due to the same symptoms within a set monitoring period, wherein symptom identity for determining revisits is determined through text similarity analysis based on natural language processing, analyze semantic similarity based on word embedding for input symptom keywords, classify symptoms whose similarity score exceeds a set threshold as the same symptom group, determine drug suitability by patient age group by distinguishing between elementary school students and other age groups, and control the system to filter out unsuitable drugs by comparing with administration standard information by age group registered in the drug database. Claim 8 In claim 1, the above instructions, when executed by the processor, are configured such that the system analyzes the correlation between symptoms and the timing of occurrence to determine whether a disease is seasonal, extracts a pattern of frequency of occurrence in a specific season or period through time series analysis, calculates a utility score per prescription based on revisit data, quantifies utility based on the ratio of revisit cases to total prescription cases, calculates a composite score by applying weights set for prescription frequency, symptom similarity, and utility score, wherein the weights of each indicator are configured to be adjustable by the system administrator, generates a recommendation list by selecting prescriptions with the highest composite scores among prescriptions excluding unsuitable drugs by age group, provides a reliability score and recommendation basis for each prescription, updates the utility score per prescription and readjusts the recommendation ranking in real time whenever new prescription result data is collected, and controls the updated information to be synchronized with all user systems through a distributed database. Claim 9 In claim 1, the above instructions are a system that, when executed by the processor, controls the system to receive multidimensional symptom data including symptom keywords, patient age group, time of visit, and seasonal information, collect prescription history including the type, dosage, and number of administrations of prescribed drugs, and revisit data including whether there was a revisit due to the same symptom within a set monitoring period, and to update the efficacy score per prescription and readjust the recommendation ranking in real time by reflecting the revisit data collected after the prescription is applied. Claim 10 In claim 1, when the instructions are executed by the processor, the system stores and manages the user's underlying disease information and information on medications currently being taken in a database; the medication information is stored in a structured form mapped to the International Drug Identification Code; the system analyzes the risk of interaction with prescription data collected from the database; it identifies combinations of ingredients with potential interactions by comparing the ingredient information of the medications being taken with a medication ingredient database; it inputs the ingredients, dosage, and dosing cycle information of the medications being taken into a convolutional neural network-based drug interaction prediction model to calculate a risk grade according to established criteria; it analyzes the association between the underlying disease and the prescribed medications, while evaluating the risk based on disease-drug interaction information extracted from a medical literature database; it generates a comprehensive safety score for each prescription by combining the drug interaction risk grade and the results of the underlying disease association analysis according to established weights; for prescriptions where the comprehensive safety score is below a established threshold value, it adjusts the ranking within the recommendation list, but recalculates the ranking by reflecting the correlation between the safety score and the recommendation ranking; and for prescriptions exceeding the established risk threshold, it analyzes risk factors and sends a warning message A system that generates the message, which is visually highlighted on the user interface along with specific details of the risk factors, analyzes prescription review data collected from multiple users using natural language processing technology to extract information related to effectiveness, side effects, and satisfaction, calculates a comprehensive evaluation index by applying set weights to the extracted information, and controls the system to reflect this in the recommendation ranking in real time.

Images