Radiation machine room management method and system

Abstract

The invention discloses a radiation machine room management method and system. The method comprises the following steps: firstly, collecting historical examination data of a hospital, performing cleaning and feature extraction, and establishing an examination demand prediction model by using ARIMA and multiple linear regression algorithms; and meanwhile, equipment operation data are acquired through a sensor, and an LSTM model is utilized to predict the equipment fault probability, so that a personalized maintenance plan is formulated according to the equipment fault probability. And when a new patient needs to be examined, equipment and time are distributed through an optimization algorithm by integrating the equipment operation state, the appointment queuing condition and the predicted examination time. The system comprises a data acquisition module, an analysis and prediction module, an equipment scheduling module and a maintenance management module. The system solves the problems of low service efficiency and poor pertinence of maintenance of the equipment in the radiation machine room, improves the utilization rate of the equipment, reduces the waiting time of a patient, reduces the operation cost, and promotes the intelligent and scientific management of the radiation machine room.

Description

Technical Field

[0001] The present invention relates to the technical field of medical equipment management, and in particular to a radiology room management method and system. Background Art

[0002] In the medical industry, radiology rooms are crucial venues for medical imaging diagnostics. The efficient operation and proper management of their equipment are crucial for improving the quality and efficiency of medical services. Currently, radiology rooms suffer from a widespread problem of inefficient equipment utilization. On the one hand, a lack of accurate forecasting of patient examination needs leads to excessive idleness of some equipment during certain periods, while creating long wait times during other periods. On the other hand, equipment maintenance plans are often inadequately targeted. Either overly frequent maintenance impacts normal equipment use, or untimely maintenance leads to increased equipment failure rates, further reducing equipment availability. Summary of the Invention

[0003] The purpose of the present invention is to provide a radiology room management method and system to solve the problems raised in the above background technology.

[0004] To achieve the above object, the present invention provides the following technical solution: a radiology room management method, comprising the following steps: Step 1: Predicting Examination Demand: By collecting historical hospital examination data, including the time distribution of different types of examinations and patient age and gender information, we use big data analysis and machine learning algorithms to establish an examination demand prediction model; Step 2: Develop an equipment maintenance plan: Install sensors on the equipment in the radiology room to collect real-time operating data, including but not limited to equipment temperature, operating hours, and exposure times. Based on the collected data, a fault prediction model is used to predict when equipment failures are likely to occur. Based on the prediction results, a personalized equipment maintenance plan is developed to schedule maintenance before equipment failures occur, thereby avoiding sudden equipment failures and reducing unnecessary maintenance. Step 3. Real-time scheduling: When a new patient examination request comes in, the system uses an optimization algorithm to allocate the most appropriate equipment and examination time to the patient based on the current operating status of each device, the queue of scheduled examinations, and the predicted examination time.

[0005] Preferably, the specific implementation steps of using big data analysis and machine learning algorithms to establish an inspection demand forecasting model are as follows: Step 11. Data Collection: Comprehensively collect patient examination data from the hospital information system over the past 3-5 years. The examination data includes detailed records of different types of examinations, including but not limited to examination time, patient age, gender, department of origin, and diagnosis information; Step 12: Data cleaning: The data collected in step 11 are cleaned and preprocessed to obtain preprocessed data. First, erroneous values ​​in the data are identified and corrected. Second, missing values ​​are processed. Data records with a small amount of missing key information are directly deleted. For data with some missing information, they are filled in based on the statistical characteristics of similar patients. Duplicate data records are removed to ensure data uniqueness. Step 13: Feature Engineering: Extract features from the preprocessed data for model training. Specifically, for time features, decompose the examination time into different dimensions such as year, month, day, week, and hour to obtain the changing patterns of examination demand at different time scales. For patient characteristics, divide the age into different age groups and analyze the differences in demand for different examination types in each age group. At the same time, gender and department are used as classification features. For disease characteristics, classify and summarize the disease diagnosis and establish the correlation features between disease and examination type. Step 14. Algorithm selection: Based on the characteristics of the radiology laboratory inspection demand data, the ARIMA algorithm is selected to capture the periodic and trend changes in inspection demand over time. Combined with the multiple linear regression algorithm, the various features extracted in step 13 are used as independent variables and the inspection demand as the dependent variable to establish a linear regression model, namely the inspection demand forecasting model; Step 15: Model training: Divide the cleaned and feature-engineered data into a training set and a test set at a ratio of 70% to 30%. Use the training set data to train the linear regression model. By continuously adjusting the p, d, and q parameters in the ARIMA model and checking the coefficients in the demand forecast model, the demand forecast model is tested to achieve the best fit on the training set. Step 16: Model evaluation: Use the test set data to evaluate the model trained in step 15. By calculating and checking the prediction accuracy, mean square error, and mean absolute error indicators of the demand forecast model on the test set, the performance of the linear regression model is judged. Step 17: Model update: Based on the hospital's patient flow, disease spectrum, and medical service policy factors, the examination demand prediction model is updated regularly to ensure that the model can continue to accurately predict the examination needs of the radiology room.

[0006] Preferably, the specific working logic of using the fault prediction model to predict the time point when the device may fail is as follows: Data preprocessing: The device temperature, operating time, and exposure count data collected by the sensors are first cleaned to remove outliers and missing values. Preset features are then extracted from the cleaned data. These include the original data of device temperature, operating time, and exposure count as features, and new features generated through data transformation. These new features calculate the rate of change of device temperature, the deviation between the cumulative value of operating time and the average value, and the distribution characteristics of exposure counts over different time periods. Model selection and training: The LSTM model was selected to build a fault prediction model. Specifically, a neural network structure was constructed consisting of an input layer, multiple LSTM hidden layers, and an output layer. The input layer receives preset features, the hidden layer learns the time series features and long-term dependencies in the data through LSTM units, and the output layer outputs the probability of equipment failure within a certain time period in the future. The fault prediction model was trained using historical equipment operation data and corresponding fault labels. During the training process, the backpropagation algorithm was used to adjust the model parameters. At the same time, regularization technology was used to prevent model overfitting. By setting appropriate learning rate and number of training rounds hyperparameters, the model's training effect and generalization ability were improved, resulting in a trained fault prediction model. Prediction: Input the pre-set features collected and processed in real time into the trained fault prediction model. Based on the learned feature patterns and rules, the model predicts the probability of equipment failure within a certain period of time. Application: Based on the output results of the fault prediction model, the radiology room management system takes corresponding measures; specifically, for equipment with a higher predicted probability of failure, the maintenance management module promptly issues early warning information and notifies maintenance personnel to conduct further inspection and maintenance of the equipment; maintenance personnel formulate targeted maintenance plans based on the prediction results and the actual operating status of the equipment, such as preparing parts that need to be replaced in advance and arranging professional technicians to conduct comprehensive inspections and preventive maintenance of the equipment; at the same time, the equipment scheduling module reasonably adjusts the equipment usage plan according to the equipment's failure risk, tries to avoid arranging important inspection tasks during periods of high failure rates, and gives priority to arranging patients for inspection on equipment with a lower failure risk, to ensure the normal operation of the radiology room and the continuity of medical services; the fault prediction results and actual equipment operation status are continuously tracked and recorded, the fault prediction model is regularly evaluated and updated, and the prediction accuracy and adaptability of the model are continuously improved.

[0007] Preferably, the specific implementation logic of formulating a personalized equipment maintenance plan based on the prediction results is as follows: Quantifying failure probability: Using the probability of equipment failure at different time periods in the future output by the failure prediction model, the failure risk of the equipment is quantified. Based on the quantified failure probability, the equipment failure risk is divided into three levels: low risk, medium risk, and high risk. Specifically, a failure probability below 0.2 is considered low risk, indicating that the equipment is unlikely to fail in the short term; a failure probability between 0.2 and 0.5 is considered medium risk, meaning that the equipment has a certain failure risk and requires close attention; a failure probability above 0.5 is considered high risk, indicating that the equipment is facing a high probability of failure and that maintenance measures are urgently needed. Maintenance strategy development: For low-risk equipment, a maintenance strategy of regular inspections and basic maintenance is adopted. For medium-risk equipment, in addition to regular inspections and basic maintenance, key component testing and performance evaluation are added. For high-risk equipment, an emergency maintenance plan is initiated. First, experienced professional maintenance personnel are assigned to conduct a comprehensive and in-depth inspection of the equipment, carefully evaluating the overall performance of the equipment and the working status of each component. Targeted inspection and repair are then carried out based on the fault points indicated by the fault prediction model. Maintenance scheduling: When developing a maintenance schedule, consider the equipment's daily usage and patient examination needs, and schedule maintenance during periods of low equipment usage to avoid significant disruption to normal medical services. Dynamic adjustment of maintenance plans: The system continuously collects equipment operating data in real time and feeds this data back to the maintenance management module. If, during the execution of a maintenance plan, the equipment operating data changes abnormally, or the fault prediction model updates the equipment's fault prediction results based on new data, the maintenance management module will dynamically adjust the maintenance plan in a timely manner. Maintenance effect evaluation: After each maintenance task is completed, the maintenance effect is evaluated by comparing the operating data of the equipment before and after maintenance, that is, whether the equipment temperature has returned to normal, whether the operating time is stable, and whether the exposure times are accurate, to determine whether the maintenance measures have effectively solved the problems with the equipment.

[0008] Preferably, the specific working steps of the real-time scheduling in step 3 are as follows: Step 31: Collect the operating status information of each device in use, including whether the device is currently performing an examination, the estimated remaining time to complete the examination, and the device's real-time failure rate. Simultaneously, obtain the queue status data for scheduled examinations from the hospital information system interface to identify each scheduled patient's queue position and the type of examination scheduled. Then, invoke the examination demand prediction model trained by the data analysis and prediction module to obtain the predicted examination time data for the examination type of this new patient, i.e., the typical time required to complete this type of examination. Integrate the operating status information, queue status data, and predicted examination time data for each device to form a comprehensive data set. Step 32: Using the comprehensive data set obtained in step 31 as input, the optimization algorithm within the system calculates the input data. Specifically, the optimization algorithm first selects a list of devices that are currently available, can complete the current task within a reasonable time, and can be used for new examinations based on the operating status of the equipment. Then, based on the queue status of scheduled examinations, the priority of assigning new patients to equipment with a large queue and a long waiting time is lowered to avoid further exacerbating queue congestion. At the same time, based on the predicted examination time, simple, short-term examinations are preferentially matched to equipment that is expected to be idle within a short time. Complex, long-term examinations are assigned to equipment that is expected to be idle for a longer time and has more stable performance. Step 33: After the optimization algorithm completes the calculation, it generates the most appropriate equipment and examination time allocation plan for the new patient. The equipment scheduling module converts the plan into specific scheduling instructions, which are sent to the relevant equipment to prepare for the new patient examination. The examination time and equipment information are fed back to the staff and the patient end, so that the patient can know the examination arrangement and go to the designated equipment for examination on time. At the same time, the system will update the equipment operation status and appointment queue information in real time.

[0009] Preferably, a radiology room management system includes a data acquisition module, a data analysis and prediction module, an equipment scheduling module, and a maintenance management module; Data acquisition module: This module consists of sensors installed on the device and a hospital information system interface. The sensors are responsible for collecting the device's operating data, while the hospital information system interface is used to obtain the patient's examination appointment information and historical examination data. Data Analysis and Prediction Module: This module runs big data analysis and machine learning algorithms to process and analyze collected data; it trains and inspects demand prediction models and fault prediction models to provide data support for equipment scheduling and maintenance planning; Equipment Scheduling Module: Based on the results of the data analysis and prediction module, combined with real-time equipment status and patient needs, it uses optimization algorithms to generate equipment scheduling plans and sends scheduling instructions to relevant equipment and staff; Maintenance Management Module: Develops equipment maintenance plans based on the results of the fault prediction model and tracks maintenance execution. It also records maintenance history to provide a basis for subsequent maintenance optimization.

[0010] Compared with existing technologies, this invention has the following benefits: It improves equipment utilization efficiency: By leveraging big data analysis and machine learning algorithms to establish a test demand forecasting model, it can accurately identify demand trends for various types of tests at different times. This allows hospitals to rationally allocate equipment resources in advance, reduce equipment idle time, and avoid situations where equipment is excessively idle during certain periods while waiting in line for appointments during other periods, thereby significantly improving the overall utilization efficiency of radiology room equipment.

[0011] Ensuring stable equipment operation: Sensors collect equipment operating data, using fault prediction models to predict equipment failures in advance and develop personalized maintenance plans. This preventative maintenance strategy schedules maintenance before equipment failures occur, effectively avoiding sudden equipment failures. Furthermore, it reduces unnecessary maintenance, minimizing disruption to normal equipment use and extending its lifespan, ensuring stable equipment operation and providing reliable support for medical imaging diagnosis.

[0012] Optimizing the patient experience: Real-time scheduling assigns the most appropriate device and examination time to new patients based on device operating status, appointment queues, and predicted examination times. This significantly shortens patient wait times, reduces anxiety, and improves patient satisfaction. Furthermore, optimal device allocation ensures efficient examinations and enhances the quality of medical services.

[0013] Reduced operating costs: Accurate inspection demand forecasts and equipment maintenance plans avoid excessive equipment maintenance and increased repair costs due to unexpected failures. This reduces equipment idle time and downtime, improves equipment utilization, and indirectly reduces the frequency of new equipment purchases, effectively reducing the operating costs of the radiology room. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 It is a schematic diagram of the method flow structure of the present invention; Figure 2 Schematic diagram of the system structure of the present invention. DETAILED DESCRIPTION

[0015] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0016] Example 1

[0017] See also Figure 1 The present invention provides a technical solution: a method for managing a radiology room, comprising the following steps: Step 1: Forecasting Examination Demand: By collecting historical hospital examination data, including the time distribution of different types of examinations and patient age and gender information, and using big data analysis and machine learning algorithms, a model for predicting examination demand is established. The specific implementation steps are as follows: Step 11. Data Collection: Comprehensively collect patient examination data from the hospital information system over the past 3-5 years. The examination data includes detailed records of different types of examinations, including but not limited to examination time, patient age, gender, department of origin, and diagnosis information; Step 12: Data cleaning: The data collected in step 11 is cleaned and preprocessed to obtain preprocessed data. First, the erroneous values ​​in the data are identified and corrected. Second, missing values ​​are processed. Data records with a small amount of missing key information are directly deleted. For data with some missing information, the data is filled in based on the statistical characteristics of similar patients. Duplicate data records are removed to ensure data uniqueness. For example, the missing gender can be inferred by calculating the gender ratio of patients in the same age group and department. In addition, duplicate data records are removed to ensure data uniqueness. Step 13, Feature Engineering: Extract features from the preprocessed data for model training, including: In terms of time features, decompose the examination time into different dimensions such as year, month, day, week, and hour to obtain the changing patterns of examination demand at different time scales; in terms of patient characteristics, divide the age into different age groups and analyze the differences in demand for different examination types in each age group. At the same time, use gender and department as classification features; in terms of disease characteristics, classify and summarize the diagnosis of the disease and establish the correlation characteristics between the disease and the examination type. For example, count the proportion of patients with respiratory diseases who undergo chest X-ray and CT examinations. In addition, external data such as season and local demographic data are also converted into corresponding feature variables; Step 14. Algorithm selection: Based on the characteristics of the radiology laboratory inspection demand data, the ARIMA algorithm is selected to capture the periodic and trend changes in inspection demand over time. Combined with the multiple linear regression algorithm, the various features extracted in step 13 are used as independent variables and the inspection demand as the dependent variable to establish a linear regression model, namely the inspection demand forecasting model; Step 15: Model training: Divide the cleaned and feature-engineered data into a training set and a test set at a ratio of 70% to 30%. Use the training set data to train the linear regression model. By continuously adjusting the p, d, and q parameters in the ARIMA model and checking the coefficients in the demand forecast model, the demand forecast model is tested to achieve the best fit on the training set. Step 16: Model evaluation: Use the test set data to evaluate the model trained in step 15. By calculating and checking the prediction accuracy, mean square error, and mean absolute error indicators of the demand forecast model on the test set, the performance of the linear regression model is judged. Step 17: Model update: Based on the hospital's patient flow, disease spectrum, and medical service policy factors, the examination demand prediction model is updated regularly to ensure that the model can continue to accurately predict the examination needs of the radiology room.

[0018] Step 2: Develop an equipment maintenance plan: Install sensors on the equipment in the radiology room to collect real-time operating data, including but not limited to equipment temperature, operating hours, and exposure times. Based on the collected data, a fault prediction model is used to predict when equipment failures are likely to occur. Based on the prediction results, a personalized equipment maintenance plan is developed to schedule maintenance before equipment failures occur, thereby avoiding sudden equipment failures and reducing unnecessary maintenance. The specific working logic of using the fault prediction model to predict the time when the equipment may fail is as follows: Data preprocessing: The device temperature, operating time, and exposure count data collected by the sensors are first cleaned to remove outliers and missing values. Preset features are then extracted from the cleaned data. These include the original data of device temperature, operating time, and exposure count as features, and new features generated through data transformation. These new features calculate the rate of change of device temperature, the deviation between the cumulative value of operating time and the average value, and the distribution characteristics of exposure counts over different time periods. Model selection and training: The LSTM model was selected to build a fault prediction model. Specifically, a neural network structure was constructed consisting of an input layer, multiple LSTM hidden layers, and an output layer. The input layer receives preset features, the hidden layer learns the time series features and long-term dependencies in the data through LSTM units, and the output layer outputs the probability of equipment failure within a certain time period in the future. The fault prediction model was trained using historical equipment operation data and corresponding fault labels. During the training process, the backpropagation algorithm was used to adjust the model parameters. At the same time, regularization technology was used to prevent model overfitting. By setting appropriate learning rate and number of training rounds hyperparameters, the model's training effect and generalization ability were improved, resulting in a trained fault prediction model. Prediction: Input the pre-set features collected and processed in real time into the trained fault prediction model. Based on the learned feature patterns and rules, the model predicts the probability of equipment failure within a certain period of time. Application: Based on the output results of the fault prediction model, the radiology room management system takes corresponding measures; specifically, for equipment with a higher predicted probability of failure, the maintenance management module promptly issues early warning information and notifies maintenance personnel to conduct further inspection and maintenance of the equipment; maintenance personnel formulate targeted maintenance plans based on the prediction results and the actual operating status of the equipment, such as preparing parts that need to be replaced in advance and arranging professional technicians to conduct comprehensive inspections and preventive maintenance of the equipment; at the same time, the equipment scheduling module reasonably adjusts the equipment usage plan according to the equipment's failure risk, tries to avoid arranging important inspection tasks during periods of high failure rates, and gives priority to arranging patients for inspection on equipment with a lower failure risk, to ensure the normal operation of the radiology room and the continuity of medical services; the fault prediction results and actual equipment operation status are continuously tracked and recorded, the fault prediction model is regularly evaluated and updated, and the prediction accuracy and adaptability of the model are continuously improved.

[0019] The specific implementation logic for formulating a personalized equipment maintenance plan based on the prediction results is as follows: Quantifying failure probability: Using the probability of equipment failure at different time periods in the future output by the failure prediction model, the failure risk of the equipment is quantified. Based on the quantified failure probability, the equipment failure risk is divided into three levels: low risk, medium risk, and high risk. Specifically, a failure probability below 0.2 is considered low risk, indicating that the equipment is unlikely to fail in the short term; a failure probability between 0.2 and 0.5 is considered medium risk, meaning that the equipment has a certain failure risk and requires close attention; a failure probability above 0.5 is considered high risk, indicating that the equipment is facing a high probability of failure and that maintenance measures are urgently needed. Maintenance strategy formulation: For equipment at a low-risk level, a maintenance strategy of regular inspections and basic maintenance is adopted. The period of regular inspections can be relatively long, such as a comprehensive inspection every two weeks to check whether the appearance of the equipment is damaged, whether the connection lines are loose, and whether there is any abnormal noise during equipment operation. Basic maintenance includes routine maintenance operations such as cleaning and lubricating the equipment to ensure that the equipment is in a good operating environment and basic condition, and to prevent potential minor faults from turning into major faults; for equipment at a medium-risk level, in addition to regular inspections and basic maintenance, increase the inspection and performance evaluation of key components, such as the tube of X-ray equipment and the detector of CT equipment. These key components are inspected using professional testing equipment and technology, such as using a tube tester to detect parameters such as the filament current and tube voltage of the X-ray tube to evaluate its performance status. At the same time, the inspection cycle is shortened to once a week, and changes in the equipment's operating data, such as equipment temperature, operating time, number of exposures, etc., are closely monitored. Once abnormal fluctuations in the data are found, in-depth investigation and analysis are carried out in a timely manner; for equipment with high risk levels, emergency maintenance plans are initiated. First, experienced professional maintenance personnel are arranged to conduct a comprehensive and in-depth inspection of the equipment, and to evaluate the overall performance of the equipment and the working status of each component in detail. According to the fault points suggested by the fault prediction model, targeted inspection and repair are carried out. For example, if the fault prediction model indicates that the detector of a certain CT device may fail after running a certain number of times, the maintenance personnel will focus on a comprehensive inspection of the detector, including testing of key indicators such as the detector's pixel response uniformity and noise level. For potential hidden faults discovered, damaged or aging components are replaced in a timely manner to ensure that the equipment is restored to normal operating conditions before the failure occurs. At the same time, after the maintenance is completed, the equipment is rigorously performance tested and calibrated to ensure that the various performance indicators of the equipment meet the requirements of clinical use; Maintenance plan scheduling: When formulating a maintenance plan schedule, schedule maintenance during relatively low equipment usage periods based on the daily usage of the equipment and patient examination needs to avoid a significant impact on normal medical services. For example, for radiology room equipment that is busy with daytime examinations, schedule maintenance during nighttime or weekends when there are fewer patient examinations. At the same time, combined with the results of the examination demand forecasting model, predict the peak and low periods of equipment usage in the future and adjust the maintenance plan schedule reasonably. If it is predicted that the demand for a certain device will increase significantly in the next week, and the device is currently at a medium-risk level, maintenance needs to be scheduled as soon as possible during a relatively idle period this week to ensure that the equipment can operate normally during peak usage periods; Dynamic adjustment of maintenance plans: The system continuously collects the operating data of the equipment in real time and feeds this data back to the maintenance management module; if during the execution of the maintenance plan, the operating data of the equipment changes abnormally, or the fault prediction model updates the fault prediction results of the equipment based on the new data, the maintenance management module will dynamically adjust the maintenance plan in a timely manner. For the maintenance of some large and complex equipment, senior maintenance engineers with relevant professional knowledge and rich experience are assigned to be responsible; for some routine maintenance tasks, they can be completed by junior maintenance personnel under the guidance of senior engineers. At the same time, ensure the timely supply of resources such as parts, tools and testing equipment required for maintenance. When formulating a maintenance plan, check the inventory of parts in advance. For parts that need to be replaced, if the inventory is insufficient, make a purchase application in time to ensure that they can be replaced smoothly during maintenance. For example, before arranging a maintenance plan for a CT device, check the inventory of required detector modules, high-voltage generators and other parts in advance. If the detector module is found to be insufficient in inventory, immediately start the emergency procurement process to ensure that the maintenance plan is executed on time; Maintenance Effectiveness Evaluation: After each maintenance task is completed, the effectiveness of the maintenance is evaluated. By comparing the equipment's operating data before and after maintenance, such as whether the equipment temperature has returned to normal, whether the operating time is stable, and whether the exposure times are accurate, we determine whether the maintenance measures have effectively resolved the equipment's problems. At the same time, we collect feedback on the equipment's actual operating status for a period of time after maintenance to understand whether the equipment experiences any abnormalities or failures again. If the maintenance results are unsatisfactory, we analyze the causes and optimize and adjust the subsequent maintenance plan and strategy. For example, if a certain device continues to frequently experience the same type of failure after maintenance, analysis may reveal that the quality of the parts replaced during maintenance was problematic, or that the maintenance operations were not performed in full accordance with specifications. To address these issues, high-quality parts are promptly replaced, and maintenance personnel training and operational standard management are strengthened. Subsequent maintenance plans for similar equipment are adjusted accordingly to improve maintenance quality and effectiveness.

[0020] Step 3: Real-time scheduling: When a new patient requires an examination, the system uses an optimization algorithm to allocate the most appropriate equipment and examination time to the patient based on the current operating status of each device, the queue of scheduled examinations, and the predicted examination time. The specific working steps are as follows: Step 31: Collect the operating status information of each device in use, including whether the device is currently performing an examination, the estimated remaining time to complete the examination, and the device's real-time failure rate. Simultaneously, obtain the queue status data for scheduled examinations from the hospital information system interface to identify each scheduled patient's queue position and the type of examination scheduled. Then, invoke the examination demand prediction model trained by the data analysis and prediction module to obtain the predicted examination time data for the examination type of this new patient, i.e., the typical time required to complete this type of examination. Integrate the operating status information, queue status data, and predicted examination time data for each device to form a comprehensive data set. Step 32: Using the comprehensive data set obtained in step 31 as input, the optimization algorithm within the system calculates the input data. Specifically, the optimization algorithm first selects a list of devices that are currently available, can complete the current task within a reasonable time, and can be used for new examinations based on the operating status of the equipment. Then, based on the queue status of scheduled examinations, the priority of assigning new patients to equipment with a large queue and a long waiting time is lowered to avoid further exacerbating queue congestion. At the same time, based on the predicted examination time, simple, short-term examinations are preferentially matched to equipment that is expected to be idle within a short time. Complex, long-term examinations are assigned to equipment that is expected to be idle for a longer time and has more stable performance. Step 33: After the optimization algorithm completes the calculation, it generates the most appropriate equipment and examination time allocation plan for the new patient. The equipment scheduling module converts the plan into specific scheduling instructions, which are sent to the relevant equipment to prepare for the new patient examination. The examination time and equipment information are fed back to the staff and the patient end, so that the patient can know the examination arrangement and go to the designated equipment for examination on time. At the same time, the system will update the equipment operation status and appointment queue information in real time.

[0021] Example 2

[0022] like Figure 2 ,A radiology room management system includes a data acquisition module, a data analysis and prediction module, an equipment scheduling module, and a maintenance management module; Data acquisition module: This module consists of sensors installed on the device and a hospital information system interface. The sensors are responsible for collecting the device's operating data, while the hospital information system interface is used to obtain the patient's examination appointment information and historical examination data. Data Analysis and Prediction Module: This module runs big data analysis and machine learning algorithms to process and analyze collected data; it trains and inspects demand prediction models and fault prediction models to provide data support for equipment scheduling and maintenance planning; Equipment Scheduling Module: Based on the results of the data analysis and prediction module, combined with real-time equipment status and patient needs, it uses optimization algorithms to generate equipment scheduling plans and sends scheduling instructions to relevant equipment and staff; Maintenance Management Module: Develops equipment maintenance plans based on the results of the fault prediction model and tracks maintenance execution. It also records maintenance history to provide a basis for subsequent maintenance optimization.

[0023] This invention focuses on the field of medical equipment management technology and proposes an innovative radiology room management method and system. At the management level, it deeply mines the hospital's patient examination data from the past 3-5 years, covering multiple dimensions such as examination time, patient basic information, and diagnosis. A rigorous data cleaning process ensures data quality, then extracts key features. Using the ARIMA algorithm, the temporal patterns of examination requests are analyzed, and combined with a multivariate linear regression algorithm, a high-precision examination demand prediction model is constructed. Furthermore, regarding equipment management, sensors collect real-time data such as equipment temperature, operating time, and exposure times. An LSTM model is used to accurately predict equipment failure probability, and differentiated maintenance plans are developed based on risk levels to achieve preventive maintenance. When a new patient requests an examination, the system integrates equipment operating status, appointment queue information, and predicted examination time, and intelligently allocates equipment and examination time using an optimization algorithm. The radiology room management system is composed of collaborative modules for data collection, analysis and prediction, equipment scheduling, and maintenance management. This invention effectively addresses the challenges of low equipment utilization efficiency and irrational maintenance planning in traditional radiology room management, significantly improving equipment utilization, shortening patient wait times, and reducing operating costs, providing strong support for intelligent and scientific management of radiology rooms.

[0024] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A method for managing a radiology room, characterized in that: The following steps are involved: Step 1: Predicting Examination Demand: By collecting historical hospital examination data, including the time distribution of different types of examinations and patient age and gender information, we use big data analysis and machine learning algorithms to establish an examination demand prediction model; Step 2: Develop an equipment maintenance plan: Install sensors on the equipment in the radiology room to collect real-time operating data, including but not limited to equipment temperature, operating hours, and exposure times. Based on the collected data, a fault prediction model is used to predict when equipment failures are likely to occur. Based on the prediction results, a personalized equipment maintenance plan is developed to schedule maintenance before equipment failures occur, thereby avoiding sudden equipment failures and reducing unnecessary maintenance. Step 3. Real-time scheduling: When a new patient examination request comes in, the system uses an optimization algorithm to allocate the most appropriate equipment and examination time to the patient based on the current operating status of each device, the queue of scheduled examinations, and the predicted examination time.

2. A radiology room management method according to claim 1, characterized in that: The specific implementation steps for establishing an inspection demand forecasting model using big data analysis and machine learning algorithms are as follows: Step 11. Data Collection: Comprehensively collect patient examination data from the hospital information system over the past 3-5 years. The examination data includes detailed records of different types of examinations, including but not limited to examination time, patient age, gender, department of origin, and diagnosis information; Step 12: Data cleaning: The data collected in step 11 are cleaned and preprocessed to obtain preprocessed data. First, erroneous values ​​in the data are identified and corrected. Second, missing values ​​are processed. Data records with a small amount of missing key information are directly deleted. For data with some missing information, they are filled in based on the statistical characteristics of similar patients. Duplicate data records are removed to ensure data uniqueness. Step 13: Feature Engineering: Extract features from the preprocessed data for model training. Specifically, for time features, decompose the examination time into different dimensions such as year, month, day, week, and hour to obtain the changing patterns of examination demand at different time scales. For patient characteristics, divide the age into different age groups and analyze the differences in demand for different examination types in each age group. At the same time, gender and department are used as classification features. For disease characteristics, classify and summarize the disease diagnosis and establish the correlation features between disease and examination type. Step 14. Algorithm selection: Based on the characteristics of the radiology laboratory inspection demand data, the ARIMA algorithm is selected to capture the periodic and trend changes in inspection demand over time. Combined with the multiple linear regression algorithm, the various features extracted in step 13 are used as independent variables and the inspection demand as the dependent variable to establish a linear regression model, namely the inspection demand forecasting model; Step 15: Model training: Divide the cleaned and feature-engineered data into a training set and a test set at a ratio of 70% to 30%. Use the training set data to train the linear regression model. By continuously adjusting the p, d, and q parameters in the ARIMA model and checking the coefficients in the demand forecast model, the demand forecast model is tested to achieve the best fit on the training set. Step 16: Model evaluation: Use the test set data to evaluate the model trained in step 15. By calculating and checking the prediction accuracy, mean square error, and mean absolute error indicators of the demand forecast model on the test set, the performance of the linear regression model is judged. Step 17: Model update: Based on the hospital's patient flow, disease spectrum, and medical service policy factors, the examination demand prediction model is updated regularly to ensure that the model can continue to accurately predict the examination needs of the radiology room.

3. A radiology room management method according to claim 1, characterized in that: The specific working logic of using the fault prediction model to predict the time point when the equipment may fail is as follows: Data preprocessing: The device temperature, operating time, and exposure count data collected by the sensors are first cleaned to remove outliers and missing values. Preset features are then extracted from the cleaned data. These include the original data of device temperature, operating time, and exposure count as features, and new features generated through data transformation. These new features calculate the rate of change of device temperature, the deviation between the cumulative value of operating time and the average value, and the distribution characteristics of exposure counts over different time periods. Model selection and training: The LSTM model was selected to build a fault prediction model. Specifically, a neural network structure was constructed consisting of an input layer, multiple LSTM hidden layers, and an output layer. The input layer receives preset features, the hidden layer learns the time series features and long-term dependencies in the data through LSTM units, and the output layer outputs the probability of equipment failure within a certain time period in the future. The fault prediction model was trained using historical equipment operation data and corresponding fault labels. During the training process, the backpropagation algorithm was used to adjust the model parameters. At the same time, regularization technology was used to prevent model overfitting. By setting appropriate learning rate and number of training rounds hyperparameters, the model's training effect and generalization ability were improved, resulting in a trained fault prediction model. Prediction: Input the pre-set features collected and processed in real time into the trained fault prediction model. Based on the learned feature patterns and rules, the model predicts the probability of equipment failure within a certain period of time. Application: Based on the output results of the fault prediction model, the radiology room management system takes corresponding measures; specifically, for equipment with a higher predicted probability of failure, the maintenance management module promptly issues early warning information and notifies maintenance personnel to conduct further inspection and maintenance of the equipment; maintenance personnel formulate targeted maintenance plans based on the prediction results and the actual operating status of the equipment, such as preparing parts that need to be replaced in advance and arranging professional technicians to conduct comprehensive inspections and preventive maintenance of the equipment; at the same time, the equipment scheduling module reasonably adjusts the equipment usage plan according to the equipment's failure risk, tries to avoid arranging important inspection tasks during periods of high failure rates, and gives priority to arranging patients for inspection on equipment with a lower failure risk, to ensure the normal operation of the radiology room and the continuity of medical services; the fault prediction results and actual equipment operation status are continuously tracked and recorded, the fault prediction model is regularly evaluated and updated, and the prediction accuracy and adaptability of the model are continuously improved.

4. A radiology room management method according to claim 1, characterized in that: The specific implementation logic of formulating a personalized equipment maintenance plan based on the prediction results is as follows: Quantifying failure probability: Quantifying the failure risk of equipment using the probability of equipment failure at different time periods in the future output by the failure prediction model; Based on the quantified failure probability, the equipment failure risk is divided into three levels: low risk, medium risk, and high risk. Specifically, a failure probability below 0.2 is considered a low risk level, indicating that the equipment is unlikely to fail in the short term. A failure probability between 0.2 and 0.5 is considered a medium risk level, meaning that the equipment has a certain failure risk and requires close attention. A failure probability higher than 0.5 is considered a high-risk level, indicating that the equipment is likely to fail and requires urgent maintenance. Maintenance strategy formulation: For equipment at a low risk level, a maintenance strategy of regular inspections and basic maintenance is adopted; For equipment at a medium-risk level, in addition to regular inspections and basic maintenance, we will also conduct additional testing and performance evaluation of key components. For equipment at a high-risk level, we will initiate an emergency maintenance plan. First, we will arrange for experienced professional maintenance personnel to conduct a comprehensive and in-depth inspection of the equipment, carefully evaluating the overall performance of the equipment and the working status of each component. Targeted testing and repairs will be carried out based on the fault points indicated by the fault prediction model. Maintenance scheduling: When developing a maintenance schedule, consider the equipment's daily usage and patient examination needs, and schedule maintenance during periods of low equipment usage to avoid significant disruption to normal medical services. Dynamic adjustment of maintenance plan: The system continuously collects equipment operation data in real time and feeds this data back to the maintenance management module; If, during the execution of the maintenance plan, the equipment's operating data experiences abnormal changes, or the fault prediction model updates the equipment's fault prediction results based on new data, the maintenance management module will dynamically adjust the maintenance plan in a timely manner. Maintenance effect evaluation: After each maintenance task is completed, the maintenance effect is evaluated by comparing the operating data of the equipment before and after maintenance, that is, whether the equipment temperature has returned to normal, whether the operating time is stable, and whether the exposure times are accurate, to determine whether the maintenance measures have effectively solved the problems with the equipment.

5. A radiology room management method according to claim 1, characterized in that: The specific working steps of the real-time scheduling of step 3 are as follows: Step 31: Collect the operating status information of each device in use, including whether the device is currently performing an examination, the estimated remaining time to complete the examination, and the device's real-time failure rate. Simultaneously, obtain the queue status data for scheduled examinations from the hospital information system interface to identify each scheduled patient's queue position and the type of examination scheduled. Then, invoke the examination demand prediction model trained by the data analysis and prediction module to obtain the predicted examination time data for the examination type of this new patient, i.e., the typical time required to complete this type of examination. Integrate the operating status information, queue status data, and predicted examination time data for each device to form a comprehensive data set. Step 32: Using the comprehensive data set obtained in step 31 as input, the optimization algorithm within the system calculates the input data. Specifically, the optimization algorithm first selects a list of devices that are currently available, can complete the current task within a reasonable time, and can be used for new examinations based on the operating status of the equipment. Then, based on the queue status of scheduled examinations, the priority of assigning new patients to equipment with a large queue and a long waiting time is lowered to avoid further exacerbating queue congestion. At the same time, based on the predicted examination time, simple, short-term examinations are preferentially matched to equipment that is expected to be idle within a short time. Complex, long-term examinations are assigned to equipment that is expected to be idle for a longer time and has more stable performance. Step 33: After the optimization algorithm completes the calculation, it generates the most appropriate equipment and examination time allocation plan for the new patient. The equipment scheduling module converts the plan into specific scheduling instructions, which are sent to the relevant equipment to prepare for the new patient examination. The examination time and equipment information are fed back to the staff and the patient end, so that the patient can know the examination arrangement and go to the designated equipment for examination on time. At the same time, the system will update the equipment operation status and appointment queue information in real time.

6. A radiology room management system according to any one of claims 1 to 5, comprising a data acquisition module, a data analysis and prediction module, an equipment scheduling module, and a maintenance management module, characterized in that: Data acquisition module: This module consists of sensors installed on the device and a hospital information system interface. The sensors are responsible for collecting the device's operating data, while the hospital information system interface is used to obtain the patient's examination appointment information and historical examination data. Data analysis and prediction module: runs big data analysis and machine learning algorithms to process and analyze the collected data; Train inspection demand prediction models and fault prediction models to provide data support for equipment scheduling and maintenance plan formulation; Equipment Scheduling Module: Based on the results of the data analysis and prediction module, combined with real-time equipment status and patient needs, it uses optimization algorithms to generate equipment scheduling plans and sends scheduling instructions to relevant equipment and staff; Maintenance Management Module: Develops equipment maintenance plans based on the results of the fault prediction model and tracks maintenance execution. It also records maintenance history to provide a basis for subsequent maintenance optimization.

Images