Bayesian network-based medical OLED screen burn-in risk prediction and protection system and method

Abstract

The invention discloses a Bayesian network-based medical OLED screen burn-in risk prediction and protection system and method, and aims to improve the burn-in risk prediction precision and reliability of an OLED screen in a medical environment. The use data, environmental parameters and display content characteristics of the OLED screen are collected through the data acquisition module, and a Bayesian network model is constructed for screen burn-in risk assessment. The parameter learning module updates model parameters in combination with variational inference and a Markov chain Monte Carlo (MCMC) method, and the prediction module calculates the burn-in risk probability based on observation data. And the protection strategy execution module implements dynamic protection measures, including pixel compensation, image transformation and the like, according to the data. According to the method, real-time prediction can be carried out on a specific medical scene, a protection scheme is optimized, the burn-in risk is effectively reduced, and meanwhile the image quality is guaranteed.

Description

Technical Field

[0001] The present invention relates to the field of medical display technology, and in particular to a medical OLED screen burn-in risk prediction and protection system and method based on a Bayesian network. Background Art

[0002] OLED (organic light-emitting diode) display technology has gradually become the first choice for display devices in the medical field due to its high contrast, wide viewing angle, color accuracy and fast response. Compared with traditional LCD displays, OLED displays can present medical images more accurately, which is crucial for medical scenarios such as diagnosis, surgery and monitoring. However, the self-luminous characteristics of OLED displays make them more susceptible to burn-in problems, especially in medical environments, where display devices often need to present static interfaces or monitoring data for a long time, which significantly increases the risk of burn-in on OLED screens. In medical applications, burn-in may cause distortion of diagnostic images or inaccurate monitoring information, which in turn affects the judgment of medical staff. Therefore, effectively preventing the burn-in problem of OLED displays is crucial to ensure their reliability and service life in medical environments.

[0003] Current pixel compensation methods usually balance pixel aging by adjusting brightness and usage time, but these methods are mainly applicable to general display scenarios, relying on fixed rules or simple statistical models, and cannot effectively meet the special needs of medical images. Medical displays not only need to display static images for a long time, but also must accurately present medical images such as X-rays, CT scans, MRIs, etc. The details of these images are very important for diagnosis. Although existing technologies have attempted to predict the aging risk of different pixel areas through data-driven methods, traditional compensation technologies have failed to fully cope with the complex characteristics of medical images and cannot effectively prevent burn-in for complex medical images. Summary of the invention

[0004] The purpose of this application is to provide a medical OLED screen burn-in risk prediction and protection system and method based on a Bayesian network to improve the accuracy and reliability of medical OLED display burn-in risk prediction.

[0005] In order to achieve the above objectives, this application adopts the following technical solutions.

[0006] A medical OLED screen burn-in risk prediction and protection system based on Bayesian network, comprising: Data collection module, used to collect preset usage data, environmental parameters and display content characteristics of the OLED screen; Bayesian network model module, used to build and maintain OLED screen burn-in risk prediction model; Parameter learning module, used to update the parameters of the Bayesian network model based on historical data and real-time update data, and adjust the conditional probability distribution of each node in the model; A risk prediction module, used to calculate and output the burn-in risk in the current state according to the data corresponding to the observed variable node and the hidden variable node; A protection strategy execution module, used to implement preset protection measures according to the output results of the risk prediction module; Wherein, the data acquisition module includes: Display content type parameter, used to indicate the type of content displayed on the screen, including static images and dynamic videos; A pixel usage intensity parameter, used to identify pixel areas that exceed a preset usage rate; The cumulative usage time parameter is used to record the cumulative usage time of the OLED screen; An ambient temperature parameter, used to indicate the ambient temperature of the medical display device; Brightness setting parameter, used to indicate the current screen brightness level; The structure of the Bayesian network model includes: An observation variable node, used to use the acquisition parameters as model input to predict the burn-in risk; Hidden variable nodes are used to determine the aging status and material properties of the OLED screen. The hidden variable nodes include pixel aging degree, material stability, and material anti-burn-in capability. An output node, used to output a burn-in risk probability, wherein the burn-in risk probability is obtained by calculating the observed variable node and the hidden variable node through a joint distribution; The parameter learning module is used to update the parameters of the Bayesian network model according to historical data and real-time update data, and adjust the conditional probability distribution of each node in the model; The risk prediction module is used to calculate and output the screen burn-in risk in the current state according to the data corresponding to the observed variable node and the hidden variable node; The protection strategy execution module is used to implement preset protection measures according to the output results of the risk prediction module, and the dynamic protection measures include at least one of the following: adjusting brightness, changing display content, and triggering a screen protection mechanism.

[0007] A method for predicting and preventing burn-in risk of medical OLED screens based on a Bayesian network is applied to the above system and includes the following steps: Step S1: collecting acquisition parameters of the OLED screen, wherein the acquisition parameters include display content type, brightness setting, ambient temperature, pixel usage intensity, and cumulative usage time; Step S2: Construct an initial risk prediction framework based on the Bayesian network model, which includes the following nodes: An observation variable node uses the acquisition parameters as a model input to predict the burn-in risk; Hidden variable nodes are used to determine the aging status and material properties of the OLED screen. The hidden variable nodes include pixel aging degree, material stability, and material anti-burn-in capability. An output node, used to output a burn-in risk probability, wherein the burn-in risk probability is obtained by calculating the observed variable node and the hidden variable node through a joint distribution; Parameter learning module, used to update the parameters of the Bayesian network model based on historical data and real-time update data, and adjust the conditional probability distribution of each node in the model; A risk prediction module, used to calculate and output the burn-in risk in the current state according to the data corresponding to the observed variable node and the hidden variable node; A protection strategy execution module, used to implement preset protection measures according to the output result of the risk prediction module, wherein the dynamic protection measures include at least one of the following: adjusting brightness, changing display content, and triggering a screen protection mechanism; Step S3: estimating the state of the hidden variable node by updating the parameters of the Bayesian network model according to the historical data and the real-time update data; Step S4: Based on the updated Bayesian network model, calculate and output the burn-in risk probability in the current state; Step S5: According to the screen burn-in risk probability, take preset protection measures, wherein the preset protection measures include at least one of the following: adjusting brightness, changing display content, and triggering a screen protection mechanism.

[0008] Beneficial effect: The present application can predict the risk of OLED screen burn-in and perform protection treatment through steps S1-S5, thereby achieving accurate prediction of the risk of OLED screen burn-in, so that medical images such as X-rays, CT scans, MRI, etc. can be displayed with anti-burn-in, thereby extending the service life of the screen and improving equipment reliability. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 The overall structural diagram of the present application method is as follows; Figure 2 It is a schematic diagram of the structure of the Bayesian network model; Figure 3 A flowchart of burn-in risk prediction and protection methods; Figure 4 This is a comparison chart of the prediction accuracy of the present application method and the traditional method. DETAILED DESCRIPTION

[0010] The present application is further described in detail with reference to the accompanying drawings and specific embodiments.

[0011] like Figure 1 As shown, the medical OLED screen burn-in risk prediction and protection system based on Bayesian network of the present application includes a data acquisition module, a Bayesian network model module, a parameter learning module, a risk prediction module and a protection strategy execution module.

[0012] The data acquisition module is responsible for collecting data related to the use of the OLED screen, including cumulative usage time, brightness setting history; environmental parameters such as temperature and humidity; and display content characteristics. The image analysis unit uses advanced image processing technology to analyze display content in real time and identify potential high-risk areas.

[0013] The Bayesian network model module implements Figure 2 The network structure shown in Figure 1 contains observation variable nodes (such as display content type, pixel usage intensity, etc.), latent variable nodes (such as pixel aging degree, material stability, material anti-burn-in ability) and output nodes (burn-in risk probability). The design of the network structure integrates the domain knowledge of OLED material science and medical image display.

[0014] The parameter learning module uses a variational inference method for fast approximate posterior calculations to adapt to real-time updates of most model parameters. For key parameters, such as the transition probability of material stability, a Markov chain Monte Carlo (MCMC) method is used for more accurate posterior sampling. This hybrid approach achieves a good balance between computational efficiency and accuracy.

[0015] The risk prediction module uses the forward probability propagation algorithm to calculate the probability distribution of the burn-in risk based on the value of the current observed variable. This module not only outputs point estimates, but also provides confidence intervals, providing more comprehensive information for subsequent protection decisions.

[0016] The protection strategy execution module implements targeted anti-burn-in measures based on the predicted risk probability distribution. The pixel compensation unit dynamically adjusts the pixel drive voltage in high-risk areas; the content adaptation unit performs minor image transformations on high-risk areas without affecting medical diagnosis; and the usage suggestion generation unit provides operators with suggestions for optimizing usage, such as adjusting screen brightness or changing the display position of static elements.

[0017] like Figure 1 As shown, the medical OLED screen burn-in risk prediction and protection system based on Bayesian network provided in the embodiment of the present application includes the following steps: Step S1: Collect key usage data of acquisition parameters of OLED screen The collection parameters include display content characteristics (such as display content type, brightness setting), environmental parameters (such as ambient temperature), and usage parameters (such as pixel usage intensity and cumulative usage time); First, the data acquisition module is used to collect the usage data and environmental parameters of the OLED screen, including: Display content characteristics: such as display content type (static image or dynamic video), brightness setting, etc. These data reflect the current working status of the screen; Environmental parameters: such as ambient temperature, which is monitored in real time through sensors. Higher temperatures will accelerate the aging of OLED materials and increase the risk of screen burn-in; Usage parameters: such as pixel usage intensity (the working condition of each pixel area) and cumulative usage time (the total time the screen is running). These data are important indicators for assessing the risk of screen burn-in; By collecting these data, detailed basic data can be provided for model building and prediction in subsequent steps; Step S2: Construct and maintain the initial risk prediction framework of OLED screens based on the Bayesian network model Observation variable nodes: such as display content type, pixel usage intensity, cumulative usage time, ambient temperature and brightness setting, which are directly obtained from the data acquisition module as input parameters of the model; Hidden variable nodes: such as pixel aging, material stability, and material anti-burn-in ability. These variables cannot be measured directly, but are inferred through the Bayesian network model to reflect the state of OLED materials and internal structures; Formula: P(hidden variable | observed variable) = P(pixel aging degree, material stability, anti-burn-in ability | display content type, pixel usage intensity, cumulative usage time, ambient temperature, brightness setting), where P represents conditional probability; Step S3: Update the parameters of the Bayesian network model based on historical data and real-time update data to estimate the state of the hidden variable nodes This step is to update the parameters of the Bayesian network model through dynamically collected data. The specific implementation method is as follows: Variational inference method: Variational inference method: Fast approximate posterior calculation of the posterior distribution parameters of latent variables (such as pixel aging, material stability, and anti-burn-in ability) can improve computational efficiency in complex Bayesian networks. This method is suitable for scenarios where a large number of parameters need to be estimated quickly; Markov Chain Monte Carlo (MCMC) method: accurate posterior sampling of key latent variables (such as pixel aging and material stability) that directly affect screen risk and lifespan. The MCMC method is highly accurate and suitable for multi-modal distribution and complex dependency structures. The sensitivity analysis of latent variables can effectively handle the problem of accurate sampling in high-dimensional parameter space. The calculation formula of the variational inference method is: P(parameter|data)≈q(parameter), where q(parameter) is the approximate distribution of the parameter; MCMC method calculation formula: P(hidden variable|data)∼MCMC sampling; Step S4: Calculate the probability distribution of screen burn-in risk under the current state based on the updated Bayesian network model The Bayesian network model is used to calculate the burn-in risk probability of the OLED screen. The specific implementation method is as follows: Input the observed variable data (display content features, environmental parameters, and usage parameters collected in step S1) and the initial estimated values ​​of the latent variables, and calculate the probability distribution of the burn-in risk through the probability propagation algorithm; risk point estimates and confidence intervals; Formula: P(burn-in risk | current state) = P(observed variable, hidden variable); Risk point estimation: P^(burn-in risk) = argmax(P(burn-in risk | current state)); Confidence interval: CI=[P^(burn-in risk)−δ,P^(burn-in risk)+δ], where CI represents the confidence interval, P^ represents the point estimate of the burn-in risk, and δ represents the margin of error of the confidence interval; Step S5: Implementing preset anti-screen burn-in measures; According to the burn-in risk probability calculated in step S4, the following protective measures will be automatically taken: If the static display exceeds 1 hour, the pixel drive voltage is adjusted; If the high brightness exceeds 90%, the image will be slightly offset or the brightness will be adjusted; If the ambient temperature exceeds 70°C, protective measures such as slight changes in RGB values ​​are implemented; Formula: ΔVdrive=f(burn-in risk), where ΔVdrive represents the adjustment amplitude of the pixel drive voltage; ΔPixel=Pixel shift, ΔBrightness=brightness adjustment, ΔRGB=color adjustment, where ΔPixel represents a slight shift of pixels, ΔBrightness represents a slight adjustment of brightness, and ΔRGB represents a slight adjustment of RGB color values.

[0018] Automated dynamic adjustments and protection measures can effectively reduce the risk of screen burn-in, extend the service life of OLED screens, and improve the stability and usability of the device.

[0019] Example 1.

[0020] This embodiment provides a medical OLED screen burn-in risk prediction and protection system based on a Bayesian network. The system can collect the usage data of the OLED screen in real time, and use the Bayesian network model to calculate the burn-in risk probability, and then take dynamic protection measures. The system mainly includes the following modules: A data acquisition module is used to collect preset usage data, environmental parameters and display content characteristics of the OLED screen, including display content type parameters (such as static or dynamic images), pixel usage intensity parameters, cumulative usage time parameters, environmental temperature parameters, and brightness setting parameters; Bayesian network model module, used to build and maintain OLED screen burn-in risk prediction model; Parameter learning module, used to update the parameters of the Bayesian network model based on historical data and real-time update data, and adjust the conditional probability distribution of each node in the model; A risk prediction module, used to calculate and output the burn-in risk in the current state according to the data corresponding to the observed variable node and the hidden variable node; The protection strategy execution module is used to implement preset protection measures according to the output results of the risk prediction module.

[0021] Procedure Step S1: Collecting the acquisition parameters of the OLED screen Through the data acquisition module, the usage data of the OLED screen will be automatically collected, including: Display content characteristics: type of static image or dynamic video, brightness setting. These data can be directly extracted from the screen control method; Environmental parameters: Use temperature sensors to monitor ambient temperature in real time and record the impact of temperature changes on OLED materials; Usage parameters: pixel usage intensity (brightness level of each pixel), cumulative usage time (total screen operation time), these parameters are obtained through the screen usage history; Step S2: Constructing an initial risk prediction framework based on the Bayesian network model In this step, the Bayesian network model module inputs the above collected observation variable data into the model and establishes the following Bayesian network structure: Observation variable nodes: display content type, pixel usage intensity, cumulative usage time, ambient temperature, and brightness setting; Hidden variable nodes: pixel aging degree, material stability, material anti-burn-in ability. These hidden variables cannot be observed directly, but are inferred through the Bayesian network model; The conditional probability formula of the model is: P(hidden variable | observed variable) = P(pixel aging degree, material stability, material anti-burn-in ability | display content type, pixel usage intensity, cumulative usage time, ambient temperature, brightness setting); Step S3: Estimate the state of the hidden variable node by updating the parameters of the Bayesian network model based on historical data and real-time update data In this step, the parameters of the Bayesian network model are updated and learned in real time through the parameter learning module. For most parameters, the variational inference method is used to quickly approximate the posterior distribution to improve the model calculation efficiency: Formula: P(parameter|data)≈q(parameter), where q(parameter) is the approximate distribution of the parameter; For key latent variables (such as pixel aging and material stability), the Markov Chain Monte Carlo (MCMC) method is used for precise posterior sampling to ensure accurate prediction of high-risk areas: Formula: P(hidden variable | data) ∼ MCMC sampling; Step S4: Based on the updated Bayesian network model, calculate and output the burn-in risk probability in the current state Through the risk prediction module, the observed variables collected in step 1 and the initial estimated values ​​of the latent variables in step 3 are input into the model, and the probability distribution of the burn-in risk is calculated using the probability propagation algorithm, and the following data is output: Burn-in risk point estimation value: P^(burn-in risk) = argmax(P(burn-in risk | current state)); Confidence interval: CI = [P^(burn-in risk) − δ, P^(burn-in risk) + δ]; Step S5: taking preset protective measures according to the risk probability of screen burn-in, wherein the preset protective measures include at least one of the following: adjusting brightness, changing display content, triggering screen protection mechanism Based on the burn-in risk probability calculated in step 4, the protection strategy execution module will automatically take the following protection measures: Pixel drive voltage adjustment: If the static display time is detected to be more than 1 hour, the pixel drive voltage in the high-risk area will be automatically reduced; Image fine-tuning: If the high brightness is detected to be over 90%, a slight image shift is performed on the high-risk area to avoid screen burn-in caused by displaying the same static image for a long time; Temperature monitoring and adjustment: If the ambient temperature exceeds 70°C, the RGB value will be fine-tuned, specifically: the RGB value change does not exceed ±10 units; The protection formula is as follows: Pixel voltage adjustment: ΔVdrive=f (risk of screen burn-in); Pixel shift and brightness adjustment: ΔPixel = Pixel shift, ΔBrightness = brightness adjustment, ΔRGB = color adjustment; Operation optimization suggestions: If the displayed content is static for a long time, the operator will be prompted to adjust the screen content or brightness settings to optimize usage.

[0022] Output Data Burn-in risk estimation: Output the current burn-in risk probability value and its confidence interval of the OLED screen as a basis for decision-making on protective measures; Protection strategy execution: Dynamically adjust pixel drive voltage, brightness, etc. to prevent permanent screen burn-in caused by long-term use or high temperature conditions on OLED screens; Optimization suggestions: Provide operators with optimization suggestions, such as adjusting brightness, avoiding static display, etc.

[0023] It should be noted that the above embodiments are only preferred implementation modes of the present application and are not intended to limit the present application. Any modifications, equivalent substitutions and improvements made by technicians in this field within the spirit and scope of the present application should be included in the protection scope of the present application.

Claims

1. A medical OLED screen burn-in risk prediction and protection system based on Bayesian network, applied to medical display equipment, characterized in that: include: The data collection module is used to collect preset usage data, environmental parameters and display content characteristics of the OLED screen, including the following collection parameters: Display content type parameter, used to indicate the type of content displayed on the screen, including static images and dynamic videos; A pixel usage intensity parameter, used to identify pixel areas that exceed a preset usage rate; The cumulative usage time parameter is used to record the cumulative usage time of the OLED screen; An ambient temperature parameter, used to indicate the ambient temperature of the medical display device; Brightness setting parameter, used to indicate the current screen brightness level; The Bayesian network model module is used to build and maintain the OLED screen burn-in risk prediction model, including the following nodes: An observation variable node, used to use the acquisition parameters as model input to predict the burn-in risk; Hidden variable nodes are used to determine the aging status and material properties of the OLED screen. The hidden variable nodes include pixel aging degree, material stability, and material anti-burn-in capability. An output node, used to output a burn-in risk probability, wherein the burn-in risk probability is obtained by calculating the observed variable node and the hidden variable node through a joint distribution; Parameter learning module, used to update the parameters of the Bayesian network model based on historical data and real-time update data, and adjust the conditional probability distribution of each node in the model; A risk prediction module, used to calculate and output the burn-in risk in the current state according to the data corresponding to the observed variable node and the hidden variable node; A protection strategy execution module is used to implement preset protection measures according to the output results of the risk prediction module, and the dynamic protection measures include at least one of the following: adjusting brightness, changing display content, and triggering a screen protection mechanism.

2. The system according to claim 1, characterized in that The data acquisition module comprises: An image analysis unit, used for analyzing the display content of the OLED screen in real time; Usage mode recording unit to track the usage of the OLED screen; The environmental monitoring unit is used to collect parameters such as temperature, brightness, pixel usage intensity, and cumulative usage time of the OLED screen operating environment.

3. The system according to claim 1, characterized in that The parameter learning module is also used to use the variational inference method to calculate the approximate posterior distribution, and use the Markov chain Monte Carlo method to perform precise posterior sampling for preset key parameters to obtain parameter learning results, wherein the preset key parameters include at least one of the following: model parameters, hyperparameters, and hidden variables, and the parameter learning results include at least one of the following: parameter estimates of the posterior distribution, posterior distribution samples, confidence intervals, model prediction results, uncertainty estimates, and model evaluation indicators.

4. The system according to claim 1, characterized in that The risk prediction module is also used to obtain the probability distribution of the burn-in risk by calculating the observed variable nodes using a forward probability propagation algorithm.

5. The system according to claim 1, characterized in that The protection strategy execution module includes: A pixel compensation unit, used to dynamically adjust the pixel driving voltage of a target risk area, wherein the target risk area is an area where the static display time exceeds a preset time length; Content adaptive unit, used to perform small image transformation with pixel shift of no more than 3 pixels, brightness adjustment of no more than 5%, and RGB value change of no more than ±10 units in areas with static display of more than 1 hour, high brightness of more than 90%, or temperature of more than 70°C; Use the suggestion generation unit to provide preset optimization suggestions.

6. A method for predicting and protecting medical OLED screen burn-in risk based on Bayesian network, characterized in that: The system as claimed in any one of claims 1 to 5 comprises the following steps: Step S1: Collecting acquisition parameters of the OLED screen, wherein the acquisition parameters include display content type, brightness setting, ambient temperature, pixel usage intensity, and cumulative usage time; Step S2: Construct an initial risk prediction framework based on the Bayesian network model, which includes the following nodes: An observation variable node uses the acquisition parameters as a model input to predict the burn-in risk; Hidden variable nodes are used to determine the aging status and material properties of the OLED screen. The hidden variable nodes include pixel aging degree, material stability, and material anti-burn-in capability. An output node, used to output a burn-in risk probability, wherein the burn-in risk probability is obtained by calculating the observed variable node and the hidden variable node through a joint distribution; Parameter learning module, used to update the parameters of the Bayesian network model based on historical data and real-time update data, and adjust the conditional probability distribution of each node in the model; A risk prediction module, used to calculate and output the burn-in risk in the current state according to the data corresponding to the observed variable node and the hidden variable node; A protection strategy execution module, used to implement preset protection measures according to the output result of the risk prediction module, wherein the dynamic protection measures include at least one of the following: adjusting brightness, changing display content, and triggering a screen protection mechanism; Step S3: Update the parameters of the Bayesian network model according to the historical data and the real-time update data, and estimate the state of the hidden variable nodes; Step S4: Calculate and output the screen burn-in risk probability in the current state based on the updated Bayesian network model; Step S5: taking preset protective measures according to the risk probability of screen burn-in, wherein the preset protective measures include at least one of the following: adjusting brightness, changing display content, and triggering a screen protection mechanism.

7. The method according to claim 6, characterized in that The method of updating the parameters of the Bayesian network model according to historical data and real-time update data includes: Use variational inference methods to quickly approximate the posterior distribution parameters of latent variable nodes; The Markov Chain Monte Carlo method is used to perform accurate a posteriori sampling on key latent variable nodes, which are pixel aging degree and material stability.

8. The method according to claim 6, characterized in that The method of calculating and outputting the screen burn-in risk probability in the current state based on the updated Bayesian network model includes: By inputting the data of the observed variable nodes and the initial estimated values ​​of the hidden variable nodes, the probability distribution of the burn-in risk is calculated using the probability propagation algorithm; Outputs burn-in risk point estimate and confidence interval.

9. The method according to claim 6, characterized in that According to the risk probability of screen burn-in, the preset protection measures are taken, including: Dynamically adjust the pixel drive voltage in high-risk areas where static display exceeds 1 hour, high brightness exceeds 90%, or temperature exceeds 70°C; Alternatively, for high-risk areas with static display exceeding 1 hour, high brightness exceeding 90%, or temperature exceeding 70°C, perform slight image transformation with pixel shift not exceeding 3 pixels, brightness adjustment not exceeding 5%, and RGB value change not exceeding ±10 units; or provide preset optimization suggestions.

10. The method according to any one of claims 6 to 9, characterized in that: The method can be integrated with hospital information systems for: Update screen usage data in real time; Providing visual monitoring services through historical data analysis; Generate an alarm signal based on the analysis results; Automatically adjust OLED screen parameters.

Images