Enterostomy postoperative infection monitoring and early warning method

Abstract

The invention discloses an enterostomy postoperative infection monitoring and early warning method, and belongs to the technical field of monitoring and early warning. The method comprises the following steps: acquiring multi-modal monitoring data which comprises visual data and non-visual data; processing the visual data, and extracting and quantifying visual features of the enterostomy and the surrounding area of the enterostomy; and according to the quantized visual features and the non-visual data, calculating an evaluation value indicating an infection risk, and determining a risk level. Judging and sending out an early warning signal according to whether the risk level or the quantitative visual feature exceeds a preset key early warning threshold value; when the early warning signal is received, a manual medical examination and verification process is started; and feedback and suggestions are provided for the user. According to the method, by carrying out objective quantitative analysis on multi-modal data, early-stage and accurate evaluation and graded early warning on the infection risk after the enterostomy operation are realized, so that the timeliness and accuracy of monitoring and intervention are effectively improved, and the prognosis of a patient is improved.

Description

Technical Field

[0001] The invention belongs to the technical field of monitoring and early warning, and particularly relates to a method for monitoring and early warning of infection after enterostomy surgery. Background Art

[0002] Enterostomy surgery is an important surgical method used clinically to treat or manage a variety of gastrointestinal diseases. The surgery creates an artificial opening in the abdominal wall to allow intestinal contents to be discharged from the body through this opening, thereby changing or bypassing the normal excretion pathway. Although ostomy surgery has saved the lives of many patients or improved their quality of life, the care and complication management of the postoperative stoma and peristomal skin pose a long-term challenge to patients and their caregivers. Among them, infection of the stoma or peristomal skin is one of the common complications that may lead to serious consequences. The infection will not only cause local pain, redness, swelling, and ulceration, affect the wearing and use of the ostomy bag, and reduce the patient's quality of life, but in severe cases it may also lead to local tissue necrosis, fistula formation, systemic infection, and even life-threatening. Therefore, timely and accurate monitoring and early warning of postoperative enterostomy infection are crucial.

[0003] However, the current monitoring of post-stomy infection mainly relies on the patient or his caregiver to observe the appearance of the stoma and surrounding skin with the naked eye and the patient's subjective feeling. Medical professionals usually conduct inspections and guidance during the patient's hospitalization or regular follow-up. This monitoring method has many limitations and lacks an active early warning mechanism. The existing monitoring method mainly waits passively for patients to report symptoms or find problems during examinations, and lacks a mechanism that can actively identify potential risks and issue early warnings based on objective data changes. These limitations result in post-stomy infection being discovered only when symptoms are more obvious or even have developed to moderate or severe conditions. At this time, the difficulty of treatment increases, patients suffer more pain, and medical costs also increase accordingly. Therefore, there is an urgent need for a more objective, accurate, continuous and early warning method for post-stomy infection monitoring and early warning to improve the quality of care and prognosis of patients. Summary of the invention

[0004] In view of the above technical problems, the present invention provides a method for monitoring and early warning of post-enterostomy infection, comprising the following steps:

[0005] Step S1, data collection, obtaining multimodal monitoring data for evaluating the user's post-stomosis status.

[0006] Step S2, visual feature extraction and quantification, processes the visual data in the multimodal monitoring data, extracts and quantifies the visual features of the intestinal stoma and its surrounding area.

[0007] Step S3: Risk assessment calculation. Based on the quantified visual features and the non-visual data in the multimodal monitoring data, and based on preset rules and weights, calculate an assessment value indicating the current infection risk of the stoma area.

[0008] Step S4: Risk level judgment. Compare the assessment value with multiple preset risk level thresholds to determine the risk level to which the current situation belongs, and the risk level reflects the current infection risk degree of the stoma area.

[0009] Furthermore, it includes an early warning method, which includes the following steps:

[0010] Step S101, Judgment of early warning trigger conditions. Judge whether the conditions for triggering an early warning signal are met.

[0011] Step S102, Sending of early warning signal. When the early warning trigger conditions are met, send an early warning signal.

[0012] Step S103, Initiation of the manual medical review process. When the early warning signal is received, initiate the manual medical review process according to the determined risk level and the preset manual review priority weight associated with this risk level.

[0013] Step S104, Generation of feedback and suggestions. Provide the corresponding feedback and medical suggestions to the user according to the determined risk level and the results of the manual medical review.

[0014] Furthermore, the multimodal monitoring data includes the collected visual data of the stoma and its surrounding skin area, and the obtained non-visual data associated with the stoma. The non-visual data includes the symptom information input by the user associated with the stoma.

[0015] Furthermore, the visual feature extraction and quantification in Step S2 includes the following steps:

[0016] Step S201, Determination of the monitoring area. Identify and locate a specific range area of the stoma and its surroundings in the visual data.

[0017] Step S202, Identification of pathological features. Detect whether there are pathological visual features associated with postoperative infection of the stoma in the specific range area.

[0018] Step S203, Quantification of pathological features. Perform quantitative analysis on the identified pathological visual features to obtain their quantification parameters.

[0019] Step S204, Scoring of quantification results. Assign corresponding sub-item scores and / or risk levels to the identified pathological visual features according to the quantification parameters.

[0020] Furthermore, the pathological visual features include one or more of the following: abnormal color areas, erythema areas, ulcerated or damaged areas, abnormal secretions, and separation at the connection between the stoma and the surrounding skin.

[0021] Furthermore, the determination of the warning trigger condition in step S101 is based on the comparison result between the risk level and the preset risk level threshold for triggering a warning, and the determination result of whether the quantization parameter of any one of the quantified visual features exceeds its corresponding preset key warning threshold, so as to determine whether the warning trigger condition is satisfied.

[0022] Furthermore, the calculation of the risk assessment value can be expressed as:

[0023] , where represents the assessment value, represents the number of quantified visual features, represents the sub - score derived from the th quantified visual feature, corresponds to the preset weight of the sub - score of the th quantified visual feature, represents the number of non - visual data, represents the sub - score derived from the th non - visual data, represents the sub - score based on the th non - visual data sub - score.

[0024] Furthermore, the preset rules and weights, including the weights for the risk assessment calculation and the rules for setting the risk level threshold, are determined according to clinical medical knowledge and / or through machine learning methods training to reflect the clinical importance of different features and symptoms in assessing the risk of postoperative infection of intestinal stomas.

[0025] Furthermore, the warning signals include device interface prompts sent to the user, mobile application notifications, and remote communication information as device reminders, as well as system message pushes, emails, and text message reminders sent to designated medical staff as manual reminders.

[0026] The beneficial effects of the present invention compared with the prior art are as follows: (1) The present invention can objectively and finely quantify the assessment of the intestinal stoma and its surrounding state, thus significantly improving the early detection ability and diagnostic accuracy of infections; (2) The present invention establishes an intelligent warning and manual review mechanism based on quantified data and graded risks, realizing active monitoring, precise prompting, and remote professional support for potential infection risks. Description of the Drawings

[0027] Figure 1 This is an exemplary flowchart of the monitoring method of the present invention.

[0028] Figure 2 This is an exemplary flowchart of the early warning method of the present invention.

[0029] Figure 3 This is an exemplary flowchart of the visual feature extraction and quantification of the present invention. Detailed implementation manners

[0030] This application provides a method for monitoring and early warning of postoperative infection of intestinal stoma, aiming to objectively quantify the assessment of infection risk and early warning by collecting and analyzing multimodal data. The method includes: obtaining multimodal monitoring data; extracting and quantifying features of visual data; combining the quantified visual features with non-visual data to calculate and judge the infection risk level. Further, the method triggers an early warning signal according to the risk level or the key feature exceeding the threshold; starts an artificial medical review process; and provides feedback and suggestions to the user. The method of this application significantly improves the objectivity, timeliness and accuracy of infection monitoring.

[0031] As Figure 1 shown, this is an exemplary flowchart of the monitoring method of this embodiment, including the following steps:

[0032] Step S1: Data collection, obtaining multimodal monitoring data for evaluating the postoperative state of the user's intestinal stoma.

[0033] In one embodiment, the technical solution of step S1 is elaborated: The core of this step is to collect information that comprehensively and objectively reflects the postoperative state of the user's intestinal stoma. The multimodal monitoring data mainly includes two categories: visual data and non-visual data. Visual data is the image or video sequence of the intestinal stoma and its surrounding skin area obtained by an image acquisition device. Specific lighting conditions and acquisition distances can be set during data collection to ensure stable and clear image quality, which is beneficial to subsequent image analysis. For example, the user can be guided to use the mobile phone App to automatically take close-up photos or short videos of the intestinal stoma area at a fixed time and in a fixed posture every day. Non-visual data is obtained through user initiative input, sensors or other means. User input is an important source of non-visual data, including symptom information associated with the intestinal stoma, such as pain level, stoma bleeding condition (slight / moderate / severe), diarrhea / constipation condition, body temperature, and the user's description of their own feelings (such as itching, burning sensation, etc.). In addition, non-visual data can also include physiological parameters or environmental information obtained from other wearable devices or smart home devices. The system pools these data from different sources and different modalities as the input for subsequent risk assessment.

[0034] Step S2: Visual Feature Extraction and Quantification. Process the visual data in the multimodal monitoring data, and extract and quantify the visual features of the intestinal stoma and its surrounding area.

[0035] As Figure 3 shown in the following is an exemplary step flowchart of the visual feature extraction and quantification in step S2 of this embodiment, including the following steps:

[0036] Step S201, Determine the monitoring area. Identify and locate a specific range area of the intestinal stoma and its surrounding area in the visual data.

[0037] Step S202, Identify pathological features. Detect whether there are pathological visual features associated with postoperative infection of the intestinal stoma within the specific range area.

[0038] Step S203, Pathological feature quantification. Perform quantitative analysis on the identified pathological visual features to obtain their quantification parameters.

[0039] Step S204, Quantification result scoring. Assign corresponding sub-item scores and / or risk levels to the identified pathological visual features according to the quantification parameters.

[0040] In one embodiment, the technical solution of step S2 is elaborated: For the visual data collected in step S1, the system uses advanced image processing and computer vision technologies for analysis. First, the system accurately identifies and locates the intestinal stoma itself and the preset skin area with clinical significance around it, such as a 2-cm circular area around the stoma, through object detection and image segmentation algorithms in the image. Determining the monitoring area helps to focus the analysis on the key parts and exclude interference from irrelevant backgrounds. Then, within the determined monitoring area, the system performs pathological feature recognition. This involves analyzing the pixels in the area to detect whether there are visual signs related to infection. The pathological visual features include one or more of the following: abnormal color areas, erythema areas, ulcerated or damaged areas, abnormal secretions, and separation at the connection between the stoma and the surrounding skin. Multiple image analysis techniques can be combined, such as texture analysis, color features, shape features, or more complex image classification and segmentation models based on convolutional neural networks, to identify these features.

[0041] Then, the identified pathological visual features are quantified. For example, the area of the erythema area, the average chroma or saturation of the abnormal color area, the perimeter and area of the ulcerated area, the coverage area or color intensity of the secretions, the maximum distance or average distance of the stoma separation, etc. can be calculated. These quantification parameters provide an objective numerical description of the severity of the pathological features.

[0042] Finally, based on the quantization parameters obtained in step S203, the system generates a sub - score for each pathological visual feature by referring to a preset scoring standard or through a trained model. For example, if the erythema area is less than X square centimeters, it gets 1 point; between X and Y, it gets 3 points; greater than Y, it gets 5 points; detecting a small amount of turbid secretions gets 4 points, and a large amount of purulent secretions gets 8 points. These sub - scores are designed to convert the quantization parameters into a numerical representation of the risk contribution, preparing for the subsequent comprehensive risk assessment. These scoring standards or models are based on clinical medical knowledge and expert experience, and are obtained through machine - learning training on a large number of labeled image data.

[0043] Step S3: Risk assessment calculation. Based on the quantized visual features and non - visual data in the multimodal monitoring data, according to preset rules and weights, calculate an assessment value indicating the current infection risk of the stoma area.

[0044] In one embodiment, the technical solution of step S3 is elaborated: In this step, information from different modalities is fused to obtain a comprehensive risk assessment value. The system combines the sub - scores of the quantized visual features obtained in step S2 with the non - visual data collected in step S1. The method of calculating the comprehensive risk assessment value R can adopt a weighted sum model or other more complex fusion models.

[0045] Step S4: Risk level judgment. Compare the assessment value with multiple preset risk level thresholds to determine the risk level to which the current situation belongs. The risk level reflects the current infection risk degree of the stoma area.

[0046] In one embodiment, the technical solution of step S4 is elaborated: To convert the continuous risk assessment value R into an indicator that is easy to understand and act on, the system maps it to a preset, discrete risk level in this step. The system sets multiple risk level thresholds, which divide the range of the risk assessment value R into different intervals, and each interval corresponds to a risk level, such as "low risk", "medium risk", "high risk", "extremely high risk", etc. For example, the rule can be: If , it is determined to be low risk; if , it is determined to be medium risk; if , it is determined to be high risk; if , it is determined to be extremely high risk. These thresholds are determined based on clinical practice experience and considerations of risk tolerance. The determined risk level intuitively informs users and medical staff of the severity of the current situation and is directly related to subsequent warning trigger conditions and recommended actions. For example, low risk may only require routine care, medium risk may recommend that users strengthen observation or perform simple home treatments, while high risk or extremely high risk may immediately trigger an emergency warning and recommend seeking professional medical help.

[0047] As Figure 2 shown in the following is an exemplary flowchart of the warning method of this embodiment, including the following steps:

[0048] Step S101, judge the warning trigger condition, and judge whether the condition for triggering the warning signal is satisfied.

[0049] In one embodiment, the technical solution of step S101 is elaborated: The activation of the warning system does not simply depend on a certain fixed risk level. The warning trigger condition can be more flexible and refined. Judging whether the condition for triggering the warning signal is satisfied is mainly based on two aspects: First, whether the comprehensive risk level determined in step S4 reaches the preset level threshold that requires triggering a warning. Second, even if the overall risk level is not high, if the quantification parameters of one or several specific pathological visual features detected in step S2 exceed their corresponding preset key warning thresholds, a warning will also be triggered independently. For example, even if the overall score is below the medium risk level, but if the image analysis detects obvious signs of stoma necrosis or severe stoma separation in the intestinal stoma, once the quantification parameters of these key features exceed the preset emergency threshold, a warning signal will be immediately triggered. This dual judgment mechanism ensures that the system can not only identify medium and high risks caused by the accumulation of multiple factors, but also capture a few key and serious signs, so as to achieve early and accurate warnings.

[0050] Step S102, send the warning signal. When the warning trigger condition is satisfied, send the warning signal.

[0051] In one embodiment, the technical solution of step S102 is elaborated: Once the warning trigger condition in step S101 is determined to be satisfied, the system will immediately generate and send a warning signal. The sending methods of the warning signal are diverse to adapt to different usage scenarios and notification objects. The warning signal sent to the user may include: displaying a prominent visual prompt on the monitoring App interface or device screen; pushing a notification to the user's bound smartphone or tablet through the mobile application; directly contacting the user through remote communication methods such as text messages, emails or phone calls as a reminder for the device. The warning signal sent to the designated medical staff may include: generating a high-priority system message push in the medical management platform or electronic medical record system; sending an email or text message containing detailed warning information; or triggering an automatic phone call system to notify the on-duty staff. Different notification channels and priorities can be selected according to the urgency of the warning.

[0052] Step S103, start the manual medical review process. When receiving the warning signal, start the manual medical review process according to the determined risk level and the preset manual review priority weight associated with this risk level.

[0053] In one embodiment, the technical solution of step S103 is elaborated: The warning signal issued by the system is not just a notification, but an instruction to trigger subsequent actions. When the warning signal is issued, the system will automatically initiate or recommend initiating a manual medical review process according to the nature of the warning and the preset manual review priority weights associated with the risk level or trigger cause. For example, warnings with high risks or triggered by specific urgent key features will be given the highest review priority, and the system will automatically send the relevant case information to the designated healthcare worker workstation or review queue, and may notify the healthcare workers to handle it immediately through instant messaging tools or phone calls. Medium-risk warnings may have a lower priority and enter the regular review queue for healthcare workers to handle within a certain period of time. The manual review process allows healthcare workers to access all the original data and system analysis results for professional judgment and diagnosis, so as to verify or correct the system's automatic assessment.

[0054] Step S104, feedback and advice generation, provide corresponding feedback and medical advice to the user according to the determined risk level and the results of the manual medical review.

[0055] In one embodiment, the technical solution of step S104 is elaborated: After completing the automated risk assessment and possible manual medical review, the system will synthesize these results and provide clear feedback and personalized medical advice to the user. The feedback content usually includes the current risk level and the main problems identified. The medical advice is guidance based on the current risk situation and the review opinions of healthcare workers. The advice may include but is not limited to: detailed home care guidance; treatment advice for specific problems; when to strengthen self-monitoring or recording; advice on when to contact healthcare workers or go to the hospital for treatment; if there is a manual review, the advice will also include the professional diagnosis, treatment plan or further examination advice of healthcare workers. These feedback and advice can be presented to the user in the form of App messages, emails or system-generated reports to help the user understand their own situation and take appropriate actions.

[0056] The multimodal monitoring data includes the visual data of the stoma and its surrounding skin area collected, and the non-visual data associated with the stoma obtained, and the non-visual data includes the symptom information related to the stoma input by the user.

[0057] The pathological visual features include one or more of the following: abnormal color area, erythema area, ulcerated or damaged area, abnormal secretion, separation at the connection between the stoma and the surrounding skin.

[0058] The determination of the warning trigger condition in step S101 is based on the comparison result between the risk level and the preset risk level threshold for triggering a warning, as well as the judgment result on whether any of the quantization parameters in the quantified visual features exceed their corresponding preset key warning thresholds, to determine whether the warning trigger condition is met.

[0059] The calculation of the risk assessment value can be expressed as:

[0060] , where represents the assessment value, represents the number of quantified visual features, represents the sub - score derived from the th quantified visual feature, corresponds to the preset weight of the sub - score of the th quantified visual feature, represents the number of non - visual data, represents the sub - score derived from the th non - visual data, represents the sub - score of the th non - visual data.

[0061] In specific implementation, the calculation process is as follows: The system obtains the sub - scores S vi of each quantified and scored pathological visual feature in step S2. At the same time, the system processes the non - visual data collected in step S1. For non - visual data, it may be necessary to perform quantization or classification first and convert it into corresponding sub - scores according to preset rules . For example, map the pain levels of "mild", "moderate", and "severe" to scores 1, 3, and 5 respectively; record a high - risk score for a body temperature above 38°C. Then, the system looks up the preset weights W vi and corresponding to each visual feature and non - visual data type. These weights and the rules for converting raw data into sub - scores are determined in advance and stored in the system. The determination method is as described above, and can be based on clinical medical expert knowledge and / or optimized through machine learning training on a large amount of historical case data to maximize the prediction accuracy of the model for the infection risk. Finally, the system multiplies the sub - score of each indicator by its corresponding weight and sums up all the products to obtain the final risk assessment value R. This value R is a quantitative indicator that comprehensively considers current visible signs, subjective feelings, and potential physiological changes, providing a basis for the next risk level judgment.

[0062] The preset rules and weights, including the weights for risk assessment calculation and the rules for setting risk level thresholds, are determined based on clinical medical knowledge and / or through machine learning method training to reflect the clinical importance of different features and symptoms in assessing the risk of postoperative infection of intestinal stoma.

[0063] The warning signals include device interface prompts sent to users, mobile application notifications, and remote communication information as device reminders, as well as system message pushes, emails, and SMS reminders sent to designated medical staff as manual reminders.

[0064] The above content is only an example and illustration of the present invention. Those skilled in the art of this technology can make various modifications or supplements to the described specific embodiments or use similar methods for substitution. As long as they do not deviate from the scope defined by the invention, they should fall within the protection scope of the present invention.

Claims

1. A method for monitoring and warning of infections after enterostomy, characterized in that: It includes the following steps: Step S1, data collection, to obtain multimodal monitoring data for evaluating the state of the user after enterostomy. Step S2, visual feature extraction and quantification, to process the visual data in the multimodal monitoring data, and extract and quantify the visual features of the enterostomy and its surrounding area. Step S3, risk assessment calculation, based on the quantified visual features and the non-visual data in the multimodal monitoring data, and based on preset rules and weights, calculate an evaluation value indicating the current infection risk of the enterostomy area. Step S4, risk level judgment, compare the evaluation value with multiple preset risk level thresholds to determine the risk level to which the current situation belongs, and the risk level reflects the current infection risk degree of the enterostomy area.

2. The method for monitoring and warning of infection after enterostomy according to claim 1, characterized in that, It includes an early warning method, and the early warning method includes the following steps: Step S101, early warning trigger condition judgment, to judge whether the condition for triggering an early warning signal is satisfied. Step S102, early warning signal issuance, when the early warning trigger condition is satisfied, issue an early warning signal. Step S103, start of the manual medical review process, when the early warning signal is received, start the manual medical review process according to the determined risk level and the preset manual review priority weight associated with this risk level. Step S104, feedback and suggestion generation, according to the determined risk level and the result of the manual medical review, provide the corresponding feedback and medical suggestions to the user.

3. The method for monitoring and warning of infection after enterostomy according to claim 1, characterized in that: The multimodal monitoring data includes the visual data of the collected enterostomy and its surrounding skin area, and the non-visual data obtained related to the enterostomy, and the non-visual data includes the symptom information input by the user related to the enterostomy.

4. The method for monitoring and warning of infection after enterostomy according to claim 1, characterized in that: The visual feature extraction and quantification in step S2 includes the following steps: Step S201, determine the monitoring area, identify and locate a specific range area of the enterostomy and its surrounding area in the visual data. Step S202, identify pathological features, detect whether there are pathological visual features associated with enterostomy infection in the specific range area. Step S203, pathological feature quantification, perform quantitative analysis on the identified pathological visual features to obtain their quantification parameters. Step S204, quantification result scoring, according to the quantification parameters, assign corresponding sub-item scores and / or risk levels to the identified pathological visual features.

5. A method for monitoring and warning of infections after enterostomy according to claim 4, characterized in that The pathological visual features include one or more of the following: abnormal color area, erythema area, ulcerated or damaged area, abnormal secretion, separation at the connection between the stoma and the surrounding skin.

6. A method for monitoring and warning of infection after intestinal stoma surgery according to claim 2, characterized in that: The early warning trigger condition judgment in step S101 is based on the comparison result of the risk level with the preset risk level threshold for triggering an early warning, and the judgment result based on whether the quantification parameter of any item in the quantified visual features exceeds its corresponding preset key early warning threshold, to determine whether the early warning trigger condition is satisfied.

7. A method for monitoring and warning of infections after enterostomy according to claim 1, characterized in that The calculation of the risk assessment value can be expressed as: , wherein, represents the evaluation value, represents the quantity of quantified visual features, represents the sub - score derived based on the th kind of quantified visual feature, corresponds to the preset weight of the sub - score of the th kind of quantified visual feature sub - score, represents the quantity of non - visual data, represents the sub - score derived based on the th kind of non - visual data, represents the sub - score based on the th kind of non - visual data sub - score.

8. A method for monitoring and warning of infections after enterostomy according to claim 1, characterized in that: The preset rules and weights, including the weights for the risk assessment calculation and the rules for setting the risk level thresholds, are determined based on clinical medical knowledge and / or through machine learning method training to reflect the clinical importance of different features and symptoms in assessing the risk of postoperative infection of intestinal stoma.

9. A method for monitoring and warning of infections after intestinal stoma surgery according to claim 2, characterized in that: The warning signals include device interface prompts sent to the user, mobile application notifications, and remote communication information as device reminders, as well as system message pushes, emails, and text message reminders sent to designated medical staff as manual reminders.

Images