Individualized infection control decision system and data processing method

Abstract

The invention discloses an individualized infection control decision-making system based on wound surface microecology and a data processing method, the system collects a local sample of a wound surface, obtains a 16SrDNA or metagenome sequencing result, conventional etiology, inflammation indexes and patient conditions, and records plant defensin BKX-1 and antibacterial drug administration information; the micro-ecological and clinical features are subjected to fusion modeling based on machine learning, a wound infection severity score and a drug resistance risk score are output, an individualized infection control scheme is generated in combination with a preset antibacterial drug use rule, BKX-1 local drug use is preferentially recommended when the risk is low, broad-spectrum antibacterial drugs are reduced, and the risk is reduced. When the risk is high, etiology inspection and medication upgrading are prompted; and data of multiple patients can be counted to form quality indexes such as etiology inspection rate and the like. The method is beneficial to improving the etiology inspection rate, optimizing the use of antibacterial drugs and promoting wound infection control.

Description

Technical Field

[0001] This invention belongs to the field of medical testing and clinical infection control technology, and relates to a personalized infection control decision system and data processing method. Background Technology

[0002] Infected or suspected infected wounds are very common in clinical practice, including but not limited to diabetic foot ulcers, venous ulcers, pressure injuries, postoperative wound infections, and malignant tumor-related ulcers. These wounds are often accompanied by persistent inflammation, bacterial biofilm formation, and local microcirculatory disturbances, leading to prolonged wound healing, significantly extending hospital stays, and increasing medical costs.

[0003] Currently, the assessment and treatment of wound infections mainly rely on the following methods: Routine pathogen and inflammatory marker testing is performed. This typically involves bacterial / fungal culture and drug sensitivity testing using wound exudate or wound edge swabs. Simultaneously, markers such as procalcitonin (PCT), interleukin-6 (IL-6), and β-D-glucan are monitored to assess the degree of infection and identify suspected fungal infections. While these tests provide information on pathogen type and drug sensitivity, they suffer from drawbacks such as long processing times, limited detection rates, and difficulty in reflecting the overall microecological state of the wound.

[0004] Empirical antimicrobial therapy and local treatment. Clinically, treatment often relies on experience and limited test results, choosing local debridement, ordinary dressings, or antimicrobial dressings, combined with systemic antibiotics. In recent years, novel dressings such as topical biologics (e.g., phytodefensin BKX-1 class antimicrobial peptides) and recombinant collagen have provided more options for controlling infection and promoting healing. However, there is currently a lack of quantifiable and repeatable decision support tools regarding how topical biologics and systemic antibiotics can be used synergistically at the individual level, and under what conditions systemic antibiotic exposure can be reduced.

[0005] Therefore, it is necessary to propose a new individualized infection control decision system and method. By integrating wound microecological testing data, routine infection indicators, local and systemic antimicrobial treatment information, and antimicrobial drug management rules, this system can achieve integrated infection risk assessment, antimicrobial strategy generation, and quality control analysis, thereby improving the level of wound infection control and reducing the inappropriate use of antimicrobial drugs. Summary of the Invention

[0006] The purpose of this invention is to overcome the shortcomings of the prior art and provide a personalized infection control decision system and data processing method.

[0007] To achieve the above objectives, the present invention employs the following technical solutions: A personalized infection control decision-making system, comprising: The detection data acquisition module is used to collect local samples from infected or suspected infected wounds and integrate the results generated by different detection platforms. The local samples include wound exudate and / or wound edge swabs. The detection data acquisition module is configured to acquire wound microbiota abundance distribution, at least one α-diversity index, at least one β-diversity index, and functional pathway prediction information related to virulence factors and antimicrobial drug resistance based on 16S rDNA sequencing and / or metagenomic sequencing of the local samples, forming microecological characteristic data; and to acquire inflammation or infection index data, wound area and its rate of change, quantitative indicators of granulation tissue, and patient's overall condition information, as well as information on the patient's current and previous local treatment regimens and systemic antimicrobial drug use. The quantitative indicators of granulation tissue include at least the percentage of granulation tissue area, and the patient's overall condition information includes at least one of body temperature, complete blood count, and organ function score. The microecological infection risk assessment module includes a feature fusion unit and an assessment model unit. The feature fusion unit is used to combine the microecological feature data, the inflammation or infection indicators, the wound area and its rate of change, the quantitative indicators of granulation tissue, and the patient's overall condition information to obtain a multidimensional feature vector. The assessment model unit is an infection risk assessment model trained based on machine learning and / or deep learning, used to infer the multidimensional feature vector and output a wound infection severity score and a drug resistance risk score. The antimicrobial strategy decision-making module, connected to the microecological infection risk assessment module and the detection data acquisition module, is used to generate an individualized infection control plan by calling a pre-set antimicrobial drug use rule base based on the infection severity score and the drug resistance risk score, combined with the local treatment plan information and the systemic antimicrobial drug use information.

[0008] As a preferred embodiment, the inflammatory or infection indicator data includes at least one of bacterial culture and / or fungal culture and drug susceptibility test results, procalcitonin, interleukin-6, and β-D-glucan.

[0009] As a preferred approach, a quality control and monitoring module is also included, which is connected to the detection data acquisition module and the antimicrobial strategy decision-making module. This module is used to perform statistical analysis on the detection data and individualized infection control plans of multiple patients within a preset time window, and to calculate quality control indicators including at least the pathogen submission rate before the application of therapeutic systemic antimicrobial drugs, the incidence of multidrug-resistant bacteria, and the proportion of systemic antibiotics replaced or reduced by topical application of phytodefensin BKX-1.

[0010] As a preferred embodiment, the antimicrobial drug use rule base includes at least the rule that whether etiological testing has been completed before the application of therapeutic systemic antimicrobial drugs is a prerequisite for restricting the use of restricted and / or special-use antimicrobial drugs.

[0011] As a preferred approach, when the infection severity score is below a first threshold and the drug resistance risk score is below a second threshold, and the preconditions for using restricted and / or special-use antimicrobial drugs are not met, the antimicrobial strategy decision module is configured to generate an infection control plan that primarily uses topical application of phytodefensin BKX-1 and reduces or does not recommend systemic broad-spectrum antimicrobial drugs.

[0012] As a preferred embodiment, the α diversity index includes the Shannon index and / or the Simpson index, and the β diversity index includes a diversity measure calculated based on the Bray-Curtis distance and / or the UniFrac distance.

[0013] As a preferred embodiment, the evaluation model unit includes at least one of the gradient boosting tree model, random forest model, deep neural network model, or an ensemble model thereof.

[0014] This invention also discloses a method for processing individualized infection control decision data, comprising the following steps: S1 collects local samples from infected or suspected infected wounds and obtains wound microecological data, routine pathogen detection data and inflammatory marker data through high-throughput sequencing platform and clinical testing platform. At the same time, it obtains wound area, wound area change rate, quantitative indicators of granulation tissue and patient's overall condition information, as well as information on the patient's current and previous local treatment plans and systemic antimicrobial drug use. S2, the wound microecological data is converted into microecological feature data including microbial abundance distribution, at least one α diversity index, at least one β diversity index, and functional pathway prediction information related to virulence factors and antimicrobial drug resistance. The microecological feature data is combined with the conventional pathogen detection data, inflammatory index data, wound and systemic condition information to obtain a multidimensional feature vector. S3, input the multidimensional feature vector into the micro-ecology-infection risk assessment model trained based on machine learning and / or deep learning to obtain the wound infection severity score and drug resistance risk score; S4, The infection severity score and the drug resistance risk score are combined with a pre-set antimicrobial drug use rule base, the local treatment plan information, and the systemic antimicrobial drug use information to generate an individualized infection control plan for the patient. When the infection severity score is lower than a first threshold and the drug resistance risk score is lower than a second threshold and the preconditions for the use of restricted and / or special use antimicrobial drugs are not met, the individualized infection control plan includes a strategy of mainly using topical application of phytodefensin BKX-1 and reducing or not recommending systemic broad-spectrum antimicrobial drugs. S5. Statistical analysis is performed on the test data, infection severity scores, and individualized infection control plans of multiple patients within a preset time window to calculate quality control indicators such as the pathogen submission rate before the application of therapeutic systemic antimicrobial drugs, the incidence of multidrug-resistant bacteria, and the proportion of systemic antibiotics replaced or reduced by topical application of phytodefensin BKX-1.

[0015] As a preferred approach, in step S3, the microecology-infection risk assessment model is obtained through supervised learning training on labeled data of historical wound cases. The labeled data includes at least one or more of the following: infection severity level, whether systemic infection has occurred, whether the antimicrobial regimen needs to be upgraded, and infection control time.

[0016] The present invention has the following advantages: I. By combining the abundance distribution of microbial communities, α / β diversity indicators, and functional pathway information related to virulence factors and antimicrobial resistance, and integrating them with conventional etiology and inflammatory markers for modeling, this invention can more accurately identify the severity of wound infection and potential drug resistance risk compared to schemes that rely solely on conventional culture or simple microecological descriptions.

[0017] Second, this invention incorporates the administration method, dosage, and frequency of topical biological agents such as phytodefensin BKX-1, along with information on the use of systemic antimicrobial drugs, as input decision variables. Based on an assessment of infection risk, it prioritizes a low-risk strategy that primarily uses topical BKX-1 and reduces or eliminates the use of systemic broad-spectrum antimicrobial drugs, thereby reducing unnecessary exposure to systemic antibiotics while ensuring infection control.

[0018] III. Simulation analysis based on historical case data shows that, under reasonable assumptions, using the system of this invention for infection control decision-making is expected to increase the rate of pathogen testing before the application of therapeutic systemic antibiotics from about 55% to about 90%, shorten the average course of systemic antibiotic treatment for patients from about 12 days to about 8 days, reduce the incidence of multidrug-resistant bacteria or fungal infections from about 18% to about 11%, and shorten the average time for effective control of wound infections from about 21 days to about 16 days. At the same time, it significantly increases the proportion of cases that mainly use BKX-1 topical medication and reduce or eliminate the use of systemic antibiotics. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0020] Example 1: The individualized infection control decision-making system in this embodiment is deployed in a hospital local area network environment and includes one or more servers and several clinical terminal devices. The servers run a data acquisition module, a microecological infection risk assessment module, an antimicrobial strategy decision-making module, and a quality control and monitoring module. The clinical terminals interact with the servers through the hospital information system interface.

[0021] The data acquisition module interfaces with the Clinical Laboratory Information System (LIS), gene sequencing analysis system, and Electronic Medical Record (EMR) system to periodically or in real-time retrieve test reports and clinical records from wound patients. For patients with infected or suspected infected wounds, clinicians collect wound exudate and / or wound edge swabs according to hospital standard operating procedures, and record the sample number and collection time. Some samples are sent to a high-throughput sequencing platform for 16S rDNA sequencing or metagenomic sequencing to generate OTU / ASV abundance tables, diversity indicators, and functional pathway prediction data. Simultaneously, bacterial / fungal cultures and drug susceptibility tests are performed on the samples to generate etiological and drug susceptibility reports.

[0022] The data acquisition module converts the aforementioned microecological detection results, pathogen detection results, and inflammatory marker results into structured data, and retrieves information from the electronic medical record system at corresponding time points, including wound area, wound area change rate, quantitative indicators of granulation tissue, exudate volume, patient body temperature, blood routine, and vital organ function scores. The module also records local treatment plan information, including whether phytodefensin BKX-1 type local antibacterial agents were used, the administration method of BKX-1 (e.g., spray, irrigation, local wet compress), single dose, and frequency of administration; as well as the type, dosage, route of administration, and course of treatment of systemic antibacterial drugs.

[0023] Example 2: The microecological infection risk assessment module in this embodiment includes a feature fusion unit and an assessment model unit.

[0024] The feature fusion unit first preprocesses the 16S rDNA sequencing results, converting the sequencing data into a microbial abundance table through a standard analysis pipeline. It calculates α-diversity indices, including the Shannon index, and β-diversity indices based on Bray-Curtis distance or other distance metrics, screening for the relative abundance of dominant and potentially pathogenic bacterial genera in the wound. Based on the output of the functional prediction tool, it further extracts the abundance information of functional pathways related to virulence factors, adhesion factors, biofilm formation, and antimicrobial resistance. Simultaneously, the feature fusion unit selects features reflecting the degree of infection and the body's response status from bacterial / fungal culture and drug sensitivity results, PCT, IL-6, β-D-glucan, wound area change rate, granulation tissue ratio, exudate volume, and patient's overall condition information obtained from the detection data acquisition module. These features are then merged with the aforementioned microecological features to form a multidimensional feature vector.

[0025] The evaluation model unit employs machine learning and / or deep learning-based models, such as gradient boosting tree models, random forest models, deep neural network models, or combinations thereof. It is trained on labeled data from a large number of wound cases in a historical case database. The multidimensional feature vectors of the cases are used as input, and outcome indicators such as infection severity grading, whether systemic infection has occurred, whether an upgrade to the antimicrobial regimen is needed, and infection control time are used as supervisory signals to learn the mapping relationship between features and infection risk. After model training, during the runtime phase, the evaluation model unit infers from the multidimensional feature vectors of new patients, outputting a wound infection severity score and a drug resistance risk score. The infection severity score can be a continuous numerical value from 0 to 100 or a multi-level classification result, while the drug resistance risk score reflects the likelihood of multidrug-resistant bacterial or fungal infections.

[0026] Example 3: The antimicrobial strategy decision-making module stores a rule base for the use of antimicrobial drugs. This rule base is set according to national and hospital regulations for the clinical application management of antimicrobial drugs and includes at least the following: assessment requirements for the rate of pathogen testing before the use of therapeutic systemic antimicrobial drugs; usage conditions and approval processes for restricted and special-use antimicrobial drugs; and selection principles for systemic antimicrobial drugs based on the severity of different infections.

[0027] In practical applications, when the microecological infection risk assessment module provides a low infection severity score and a low drug resistance risk score, the antimicrobial strategy decision module determines the patient's etiological testing status based on the rule base: if the patient has completed etiological testing, the system can recommend a control strategy primarily based on topical application of phytodefensin BKX-1, with short-term or no use of systemic broad-spectrum antibiotics; if testing has not been completed, narrow-spectrum or non-restricted systemic antibiotics are recommended only if other conditions are met and the rule base is not violated, and priority is given to prompting for supplementary etiological testing. In cases of high infection severity or high drug resistance risk, the module recommends combined or upgraded systemic antibiotic regimens based on the rule base, and suggests additional or repeated etiological and / or microecological testing.

[0028] The quality control and monitoring module continuously aggregates and statistically analyzes test data, evaluation results, and antimicrobial strategy decision-making results from multiple patients in the background. It calculates quality control indicators including the rate of pathogen testing before the application of therapeutic systemic antibiotics, the incidence of multidrug-resistant bacteria, the proportion of cases where systemic antibiotics are reduced or eliminated primarily through topical BKX-1 treatment, and the average time to wound infection control. These indicators are provided to the hospital infection control committee and pharmacy department in the form of charts or dashboards for evaluating and optimizing antimicrobial application strategies and wound infection management pathways. The quality control and monitoring module can also use these quality control indicators to dynamically adjust the risk thresholds for infection severity scores and drug resistance risk scores, enabling the system to adapt to different hospitals or different time periods.

[0029] Example 4: This embodiment provides a theoretical simulation analysis based on historical case data to evaluate the potential effectiveness of the individualized infection control decision-making system of this invention in optimizing wound infection control and the rational use of antimicrobial drugs. This embodiment is a retrospective simulation and theoretical derivation based on existing case data and is not a prospective clinical trial result. The relevant data and values ​​are only used to illustrate the theoretically achievable effects of the technical solution of this invention and as examples of technical effects, and do not constitute a limitation on specific therapeutic effects or the scope of protection of this invention.

[0030] I. Source of Cases and Inclusion Criteria This study selected 200 patients with chronic or complex wounds treated at a tertiary general hospital over a specified period. Case types included, but were not limited to, diabetic foot ulcers, venous ulcers, pressure injuries, and postoperative wound infections. Inclusion criteria included, for example: Age ≥ 18 years; with an infected or suspected infected wound that has not healed for more than 2 weeks upon admission; completion of at least one routine etiological test (bacterial culture and / or fungal culture) and / or at least one inflammatory or infection marker test (such as procalcitonin, interleukin-6, β-D-glucan, etc.); complete electronic medical record of key information such as wound area, wound treatment method, type of systemic antimicrobial drug and total course of treatment; exclusion of cases with severe immunodeficiency and other systemic severe infections (such as definite septic shock) where the outcome of wound-related infection cannot be reasonably distinguished.

[0031] Without altering the existing actual treatment process, two decision-making scenarios were constructed for the aforementioned 200 cases: Control group (conventional strategy group) The control group corresponds to the actual historical treatment strategy implemented in the case, namely wound management and systemic antimicrobial therapy carried out according to the clinician's judgment and the hospital's existing antimicrobial management procedures at the time, including wound debridement, selection of local dressings or topical medications, and the type and duration of systemic antimicrobial therapy. All indicators in this group were obtained through statistical analysis of past medical records and laboratory records.

[0032] This invention simulates the application group (system strategy group). Under identical baseline conditions and test data, the system strategy group assumes that the individualized infection control decision system of this invention is used throughout the entire diagnosis and treatment process. For each case, at key assessment points, based on the wound infection severity score and drug resistance risk score of this invention, combined with the antimicrobial drug use rule base, a theoretically individualized infection control plan is regenerated, and the activation and course of systemic antimicrobial drugs are adjusted accordingly, thereby deducing the possible levels of various indicators under the system strategy of this invention.

[0033] Based on the feature construction method and training process defined in the foregoing embodiments of the specification, a microecology-infection risk assessment model is trained using existing outcome data from historical cases (including infection severity grading, whether systemic infection occurred, whether an upgrade to the antimicrobial regimen was needed, and infection control time, etc.). For each case, at each assessment time point, the model outputs: Wound infection severity score: The score ranges from 0 to 100. The higher the score, the more severe the local infection or the higher the possibility of developing into a systemic infection. Drug resistance risk score: The score ranges from 0 to 100. The higher the score, the higher the risk of developing multidrug-resistant bacteria or fungal infections.

[0034] To facilitate strategy division, the following risk thresholds are set in this embodiment: First threshold T1: Infection severity score = 50 points; Second threshold T2: Drug resistance risk score = 40 points.

[0035] When the infection severity score < T1 and the drug resistance risk score < T2, this status is regarded as the situation of "low to moderate infection and low drug resistance risk"; when any score is higher than the corresponding threshold, it is regarded as the situation of "moderate to severe infection and / or high drug resistance risk". The above thresholds are only exemplary settings in this embodiment and can be adjusted according to the case structure and model performance of different hospitals in specific applications.

[0036] In the system policy group, under the following reasonable assumptions in this embodiment, the historical cases are re - decision - reconstructed based on the antibacterial policy decision - making module of the present invention: It is assumed that for all cases, 16SrDNA sequencing and / or metagenomic detection of local wound samples are completed at the time of admission or key time points, and the corresponding micro - ecological characteristic data are obtained, including the distribution of flora abundance, diversity indexes, and prediction information of functional pathways related to virulence factors and antibacterial drug resistance.

[0037] The antibacterial drug usage rule library at least includes: 1) Completing pathogen detection before the application of therapeutic systemic antibacterial drugs is a prerequisite for enabling restricted - class and / or special - use - class antibacterial drugs; 2) When both the infection severity score and the drug resistance risk score are in the low - risk range, a control strategy mainly based on local application of plant defensin BKX - 1 is preferred, and the initiation or appropriate shortening of the course of systemic broad - spectrum antibacterial drugs is restricted; 3) When the infection severity score and / or the drug resistance risk score are in the high - risk range, on the premise of completing or supplementing pathogen detection, it is allowed to upgrade the systemic antibacterial drug treatment in advance or in a timely manner, and cooperate with local application of BKX - 1.

[0038] For the cases in the control group that originally used systemic broad - spectrum antibacterial drugs and were evaluated as "low infection severity score and low drug resistance risk score" according to the model of the present invention, the following theoretical adjustments are made in the system policy group: If the pathogen detection of this case has been completed, the system of the present invention gives priority to recommending local application of BKX - 1. When the wound infection score drops to the safe range over time, the systemic broad - spectrum antibacterial drugs are stopped 3 - 5 days in advance; if the pathogen detection of this case has not been completed, in the system policy group, it is no longer recommended to initiate or continue to use restricted - class and / or special - use - class systemic antibacterial drugs. Instead, it is recommended to supplement the detection and give priority to local application of BKX - 1, and when necessary, cooperate with relatively narrow - spectrum systemic antibacterial drugs with a course not exceeding the average level of the control group.

[0039] For the cases in the control group that have developed systemic infection or need to upgrade the systemic antibacterial regimen, the following principles are deduced in the system policy group: If the model of this invention assesses the case as having a "high infection severity score and / or high drug resistance risk score" in the early stages of the disease, the system will provide a prompt at that time point, suggesting that etiological testing be performed or repeated testing be performed as soon as possible, and that the appropriate systemic antimicrobial regimen be upgraded in advance according to the rule base, in conjunction with topical application of BKX-1. In the simulation, it is assumed that this early intervention can shorten the infection control time for some cases, thereby bringing them into a state where "further upgrade of antimicrobial drugs is not required", thus having a favorable impact on the overall infection control time and the incidence of multidrug-resistant bacteria.

[0040] Throughout all simulations, this embodiment maintains the original clinical outcomes of the control group unchanged, and only makes theoretical adjustments to the prescription regimen for the system strategy group based on the decision-making rules of this invention. For cases where patients do not fully comply with the system's recommendations due to individual differences, insufficient adherence, or other objective conditions, these are handled conservatively in the simulation, meaning they are not considered as representing all potential benefits from the system.

[0041] This embodiment mainly examines the theoretical differences of the following indicators between the two groups: The rate of etiological testing before the use of therapeutic systemic antibiotics (%) is defined as the proportion of all cases receiving therapeutic systemic antibiotics where etiological testing (bacterial culture and / or fungal culture and / or drug resistance-related testing) was completed at least once.

[0042] The average duration of systemic antibiotic treatment (d) for each patient is calculated by averaging the total number of days of systemic antibiotic use during each patient's hospitalization, thus providing an overall average duration of treatment.

[0043] The incidence rate (%) of multidrug-resistant bacteria or fungal infections is calculated by defining an "event" as a case meeting the hospital's microbiology laboratory's criteria for identifying multidrug-resistant bacteria or a clear diagnostic standard for fungal infection, and then calculating the proportion of patients experiencing that event. In the simulation, it is assumed that early risk identification and rational drug use can reduce this proportion within a certain range.

[0044] The average time (d) for effective control of wound infection was determined according to the hospital's existing clinical standards. The first date on which wound pain, redness, swelling, exudation, odor, and related inflammatory indicators significantly improved and reached a stable state was taken as the time point for "effective control of infection". The number of days from the start of treatment to this time point was calculated and averaged.

[0045] The percentage of cases where topical application of BKX-1 was the primary treatment, with reduced or no systemic antibiotics, was defined as: the percentage of cases in which topical application of the phytodefensin BKX-1 was the primary means of infection control throughout the entire treatment process, and where systemic antibiotics were not used or the total course of systemic antibiotic treatment was significantly shorter than the average course of treatment in the control group, while infection control was good.

[0046] Under the aforementioned scoring model, threshold settings, and decision rules, theoretical simulation analysis was conducted on 200 cases, yielding the following set of results. The control group directly used previously recorded statistical values; the system strategy group, without altering the patients' baseline conditions or the ultimate control of infection, theoretically adjusted the systemic antimicrobial regimen and local medication strategy based on the system's recommendations, obtaining simulated estimated values.

[0047] Table 1. Comparison of theoretical effects between conventional strategy groups and systematic strategy groups. As can be seen from the theoretical derivation results shown in Table 1: By introducing the prerequisite of "completing etiological testing before the application of therapeutic systemic antibiotics" into the decision-making process, and discouraging the direct use of restricted and / or special-use antibiotics in low-risk cases, the system of this invention is expected to increase the etiological testing rate from about 50% to 60% to about 85% to 92%, which is conducive to improving etiological testing compliance. In cases where both the infection severity score and drug resistance risk score are in the low-risk range, the system prioritizes topical application of BKX-1 and discontinues systemic broad-spectrum antibiotics early after the infection is effectively controlled, thus shortening the average course of systemic antibiotic treatment by about 3 to 5 days in the simulation. Thanks to the relatively earlier timing of infection control and the reduction of unnecessary exposure to broad-spectrum antibiotics, the theoretical incidence of multidrug-resistant bacteria or fungal infections is estimated to decrease from about 15% to 20% to about 10% to 12% in this embodiment. By identifying high-risk cases in advance and upgrading treatment plans in a timely manner, while implementing local microecological friendly strategies for low-risk cases, the average time for effective control of wound infection was shortened from about 20-22 days to about 15-17 days in the simulation, which is conducive to accelerating the control process of wound infection. In the system strategy group, the proportion of cases where topical application of phytodefensin BKX-1 was the main anti-infection method and the use of systemic antibiotics was significantly reduced increased from about 8% to 12% to about 30% to 40%, suggesting that the system of the present invention is beneficial to significantly increase the application rate of topical biological agents and reduce the overall amount of systemic antibiotics used while ensuring infection control.

[0048] This invention is not limited to the specific embodiments described above. The invention extends to any new feature or combination disclosed in this specification, as well as any new method or process step or combination disclosed herein.

Claims

1. A personalized infection control decision-making system, characterized in that, include: The detection data acquisition module is used to collect local samples from infected or suspected infected wounds and integrate the results generated by different detection platforms. The local samples include wound exudate and / or wound edge swabs. The detection data acquisition module is configured to acquire wound microbiota abundance distribution, at least one α-diversity index, at least one β-diversity index, and functional pathway prediction information related to virulence factors and antimicrobial drug resistance based on 16S rDNA sequencing and / or metagenomic sequencing of the local samples, forming microecological characteristic data; and to acquire inflammation or infection index data, wound area and its rate of change, quantitative indicators of granulation tissue, and patient's overall condition information, as well as information on the patient's current and previous local treatment regimens and systemic antimicrobial drug use. The quantitative indicators of granulation tissue include at least the percentage of granulation tissue area, and the patient's overall condition information includes at least one of body temperature, complete blood count, and organ function score. The microecological infection risk assessment module includes a feature fusion unit and an assessment model unit. The feature fusion unit is used to combine the microecological feature data, the inflammation or infection indicators, the wound area and its rate of change, the quantitative indicators of granulation tissue, and the patient's overall condition information to obtain a multidimensional feature vector. The assessment model unit is an infection risk assessment model trained based on machine learning and / or deep learning, used to infer the multidimensional feature vector and output a wound infection severity score and a drug resistance risk score. The antimicrobial strategy decision-making module, connected to the microecological infection risk assessment module and the detection data acquisition module, is used to generate an individualized infection control plan by calling a pre-set antimicrobial drug use rule base based on the infection severity score and the drug resistance risk score, combined with the local treatment plan information and the systemic antimicrobial drug use information.

2. The personalized infection control decision-making system according to claim 1, characterized in that, The inflammatory or infection marker data include bacterial and / or fungal cultures and their drug susceptibility test results, procalcitonin, interleukin-6, and at least one of β-D-glucan.

3. The personalized infection control decision-making system according to claim 1, characterized in that: It also includes a quality control and monitoring module, which is connected to the detection data acquisition module and the antimicrobial strategy decision-making module. It is used to perform statistical analysis on the detection data and individualized infection control plans of multiple patients within a preset time window, and to calculate quality control indicators including at least the pathogen submission rate before the application of therapeutic systemic antimicrobial drugs, the incidence of multidrug-resistant bacteria, and the proportion of the use of systemic antibiotics replaced or reduced by topical application of phytodefensin BKX-1.

4. The personalized infection control decision-making system according to claim 1, characterized in that, The rule base for the use of antimicrobial drugs includes at least the rule that whether etiological testing has been completed before the application of therapeutic systemic antimicrobial drugs is a prerequisite for restricting the use of restricted and / or special-use antimicrobial drugs.

5. The personalized infection control decision-making system according to claim 1, characterized in that, When the infection severity score is below a first threshold and the drug resistance risk score is below a second threshold, and the preconditions for using restricted and / or special-use antimicrobial drugs are not met, the antimicrobial strategy decision module is configured to generate an infection control plan that primarily uses topical application of phytodefensin BKX-1 and reduces or does not recommend systemic broad-spectrum antimicrobial drugs.

6. The personalized infection control decision-making system according to claim 1, characterized in that, The α diversity index includes the Shannon index and / or the Simpson index, and the β diversity index includes a diversity measure calculated based on the Bray-Curtis distance and / or the UniFrac distance.

7. The personalized infection control decision-making system according to claim 1, characterized in that, The evaluation model unit includes at least one of the following: gradient boosting tree model, random forest model, deep neural network model, or an ensemble model thereof.

8. A method for processing individualized infection control decision data, executed in the system according to any one of claims 1 to 7, characterized in that, Includes the following steps: S1 collects local samples from infected or suspected infected wounds and obtains wound microecological data, routine pathogen detection data and inflammatory marker data through high-throughput sequencing platform and clinical testing platform. At the same time, it obtains wound area, wound area change rate, quantitative indicators of granulation tissue and patient's overall condition information, as well as information on the patient's current and previous local treatment plans and systemic antimicrobial drug use. S2, the wound microecological data is converted into microecological feature data including microbial abundance distribution, at least one α diversity index, at least one β diversity index, and functional pathway prediction information related to virulence factors and antimicrobial drug resistance. The microecological feature data is combined with the conventional pathogen detection data, inflammatory index data, wound and systemic condition information to obtain a multidimensional feature vector. S3, input the multidimensional feature vector into the micro-ecology-infection risk assessment model trained based on machine learning and / or deep learning to obtain the wound infection severity score and drug resistance risk score; S4, The infection severity score and the drug resistance risk score are combined with a pre-set antimicrobial drug use rule base, the local treatment plan information, and the systemic antimicrobial drug use information to generate an individualized infection control plan for the patient. When the infection severity score is lower than a first threshold and the drug resistance risk score is lower than a second threshold and the preconditions for the use of restricted and / or special use antimicrobial drugs are not met, the individualized infection control plan includes a strategy of mainly using topical application of phytodefensin BKX-1 and reducing or not recommending systemic broad-spectrum antimicrobial drugs. S5, perform statistical analysis on the test data, infection severity score and individualized infection control plan of multiple patients within the preset time window, and calculate one or more of the following quality control indicators: pathogen submission rate before the application of therapeutic systemic antimicrobial drugs, incidence of multidrug-resistant bacteria, and the proportion of systemic antibiotics replaced or reduced by topical application of phytodefensin BKX-1.

9. The data processing method according to claim 8, characterized in that, In step S3, the microecology-infection risk assessment model is obtained by supervised learning training on labeled data of historical wound cases. The labeled data includes at least one or more of the following: infection severity level, whether systemic infection has occurred, whether the antimicrobial regimen needs to be upgraded, and infection control time.

Images