A method and system for tracking infections of multi-drug resistant strains

Abstract

The present application relates to a kind of multiple drug-resistant strain infection tracking method and system, belong to microbiological detection and environmental monitoring technical field.The tracking system includes data acquisition layer module, analysis layer module, decision layer module, using dynamic information point closed-loop analysis technique, by multiple drug-resistant strain infection is marked point assignment, realize person-person, person-environment dynamic correlation analysis, and the positioning error of pollution source is reduced to <5 meter range;Closed-loop feedback mechanism makes intervention response time shorten to 2 hours;Supporting clinical, environmental, parallel analysis of space-time data, improve public health response efficiency.

Description

Technical Field

[0001] This invention belongs to the field of microbial detection and environmental monitoring technology, and specifically relates to a method and system for tracking infections caused by multidrug-resistant strains. Background Technology

[0002] Existing microbial detection technologies have the following limitations:

[0003] Single-point static analysis: Traditional tests (such as PCR and culture methods) only target individual samples and do not relate to the distribution of environmental microorganisms. At the same time, there is often a lack of information in the links and intersections between patients.

[0004] Data fragmentation: Clinical testing and environmental monitoring data are independent, making it difficult to trace the source of infection and transmission routes. Currently, clinical registration is done manually, and the data is limited to a single ward.

[0005] False positive interference: Multiplex detection (such as tNGS) is susceptible to batch contamination and requires efficient filtering of false positives.

[0006] (I) In-depth analysis of the limitations of existing technologies

[0007] The practical impact of single-point static analysis: Traditional PCR and culture methods only focus on individual samples, often leading to misdiagnosis of "isolated infection cases" in clinical practice. For example, a hospital's ICU detected carbapenem-resistant Klebsiella pneumoniae infections for three consecutive weeks. Because environmental data was not linked, the cases were initially classified as sporadic. Subsequent environmental sampling revealed homologous strains in ward sinks and floor drains, confirming an environmentally mediated cluster infection. During this period, two new infected patients were added, highlighting the limitation of single-point analysis in timely identification of transmission chains.

[0008] The specific manifestations of data fragmentation are as follows: clinical testing data (such as bacterial strain test results in the LIS system), environmental monitoring data (such as environmental sampling records from the logistics department), and patient treatment data (such as hospitalization records and surgical histories in the HIS system) are stored in different systems with inconsistent data formats (e.g., LIS data is in structured tables, while environmental monitoring data is in paper records), and there is no data exchange interface. A survey shows that more than 80% of hospitals in China still rely on manual data integration for the management of multidrug-resistant bacteria. On average, each infection outbreak requires 3 medical staff and 5 days to process the data, seriously delaying the time for tracing the source and intervention.

[0009] The dangers of false positive interference: Multiplex detection techniques (such as tNGS) have high detection sensitivity (capable of detecting 10^10^12 false positives). 2Microorganisms with a concentration below CFU / mL are susceptible to false positives due to reagent contamination and cross-contamination during sampling. In one case, a hospital detected 5 cases of "vancomycin-resistant enterococcal infection" using tNGS. Subsequent verification revealed that the false positives were caused by contaminated testing reagents. During this period, patients had already been isolated, which not only increased their psychological burden but also wasted medical resources (such as the occupation of isolation wards and the consumption of protective equipment) and interfered with normal infection control decisions.

[0010] (ii) The urgency of industry demand

[0011] With the increasing incidence of multidrug-resistant bacterial infections (MDR-BSA infections in tertiary hospitals in China have reached 15%-20%), traditional technologies are no longer sufficient to meet the prevention and control requirements of "early detection, early tracing, and early intervention." The World Health Organization (WHO) has listed MDR-BSA as a global public health threat, requiring countries to establish efficient monitoring and tracing systems. China's "Hospital Infection Management Regulations" and "Technical Guidelines for the Prevention and Control of Multidrug-Resistant Bacterial Hospital Infections" also clearly state that hospitals need to strengthen the integration and analysis of multi-source data to improve their ability to prevent and control the spread of drug-resistant bacteria. This technology addresses this industry pain point and fills a gap in existing technologies. Summary of the Invention

[0012] (1) Technical solution

[0013] The purpose of this invention is to provide a tracking system for multidrug-resistant bacterial infections, comprising:

[0014] Data acquisition layer module: individual tNGS detection, real-time sampling of environmental sensor network, and LiS acquisition report;

[0015] Analysis layer modules: false positive filtering, drug resistance gene-host matching, spatiotemporal correlation calculation;

[0016] Decision-making module: generates heat maps of transmission risks, visualizes source tracing paths, recommends intervention strategies, and allows for real-time dialogue between hospital infection control departments and other relevant units.

[0017] Preferably, in addition to individual tNGS detection, environmental sensor network sampling, and LIS collection reports, the data acquisition layer module adds a patient movement trajectory collection function. Through the ward access control system and mobile nursing PDA positioning, it records the patient's daily activity trajectory, providing an accurate basis for the selection of environmental sampling points; at the same time, it connects to the hospital disinfection management system data to obtain environmental cleaning and disinfection records for subsequent transmission risk analysis.

[0018] Preferably, the analysis layer module, based on false positive filtering, drug resistance gene-host matching, and spatiotemporal correlation calculation, adds a strain evolution analysis function. By comparing the gene sequences of marker strains at different time periods, it analyzes the evolutionary trend of strains and predicts changes in transmission risk. At the same time, it realizes a data quality assessment function to evaluate the completeness and accuracy of the collected data to ensure the reliability of the analysis data.

[0019] Preferably, in addition to generating a heat map of transmission risk, visualizing the source tracing path, and recommending intervention strategies, the decision-making module adds an infection trend prediction function. Based on historical data, it uses an ARIMA time series model to predict the incidence of multidrug-resistant bacteria infections in various departments within the next 1-2 weeks, allowing for the early development of prevention and control plans. It also provides data export and report generation functions, supporting the export of analysis results to Excel and PDF formats, and automatically generating monthly / quarterly multidrug-resistant bacteria management reports to facilitate hospital management in evaluating the effectiveness of prevention and control.

[0020] A method for tracking infections caused by multidrug-resistant strains, comprising the following steps:

[0021] Step 1: Dynamic extraction and filtration of individual positive microbial markers

[0022] (1) Implement individual marker points:

[0023] By utilizing the existing in-hospital LIS system, we can define the markers for multidrug-resistant microorganisms, while filtering out false positives, batch contamination, and background microorganisms to identify true positives; we can then implement marker points on this data and prepare for simultaneous and concurrent specimen sources of the same infection within the hospital.

[0024] Step 2: Spatiotemporal synchronous acquisition and matching of environmental microbial markers

[0025] (1) Implement environmental markers: For cases with high suspicion of environmental pollution or during routine environmental sampling in the ICU, environmental samples can be collected simultaneously from the patient's activity trajectory to detect the same microbial targets and generate markers;

[0026] (2) Spatiotemporal matching: Aligning individual positive markers with the same type of microorganisms in their environment according to time and space dimensions;

[0027] Step 3: Human-to-human and human-to-environment microbial homology analysis and transmission risk modeling

[0028] (1) Calculate homology similarity: Determine the homology between humans and between humans and environmental microorganisms based on the similarity of microbial genome sequences;

[0029] (2) Quantitative transmission risk index: R=

[0030] Among them, W tFor the concentration of microorganisms carried by an individual, C e This represents the corresponding microbial density in the environment.

[0031] Step 4: Automated intervention mechanism based on risk feedback

[0032] (1) Dynamic early warning: If the homology similarity of human-to-human and human-to-environment microorganisms is >90%, an environmental disinfection early warning will be triggered;

[0033] (2) Path tracing: The microbial transmission network is constructed by graph optimization algorithm to locate the source of pollution.

[0034] Preferably, step one, which involves implementing individual marker points, specifically includes:

[0035] Data extraction: The results of multidrug-resistant bacteria tests are automatically extracted from the hospital's LIS system, including strain name, drug resistance gene, test date, and patient information data. The extraction frequency is once per hour to ensure data real-time performance.

[0036] False positive filtering: First, compare the sample with the negative control sample data in the LIS system. If the difference in the Ct value of the drug resistance gene between the sample to be tested and the negative control sample is <3, it is determined to be batch contamination, and the sample is removed. Second, check the patient's testing records for the past 7 days. If multiple tests show only one positive result and the bacterial concentration is <10, the false positive is removed. 3 CFU / mL, combined with clinical symptoms, such as no fever and normal infection indicators, is judged as a false positive; finally, compared with the hospital's background microbial database, if the strain is a common background bacterium in the department and there is no drug resistance phenotype to support it, it is filtered.

[0037] Marker generation: For confirmed positive samples, individual markers are generated in the system, including unique identifiers, spatiotemporal information, and strain information;

[0038] Building upon the in-hospital LIS system for labeling multidrug-resistant microorganisms, a drug resistance phenotype verification step was added. In vitro drug susceptibility testing was performed on LIS-labeled positive strains to confirm that their drug resistance phenotype was consistent with the gene detection results, avoiding mislabeling due to gene silencing or detection errors. Simultaneously, a strain concentration grading standard was established, classifying the concentration W of microorganisms carried by individuals. t Divided into high: W t ≥10 6 CFU / mL, W t 10 3 -10 6 CFU / mL, low: W t <10 3 The CFU / mL level provides accurate data support for subsequent transmission risk calculations.

[0039] Preferably, step two, in which environmental markers are implemented, specifically includes:

[0040] Sampling trigger conditions: Environmental sampling will be automatically triggered when two or more cases of the same type of multidrug-resistant bacteria infection occur in a department, with an interval of ≤7 days, or when high-risk departments such as the ICU enter the monthly routine monitoring cycle;

[0041] Sampling Procedures: Sampling personnel must undergo professional training and use sterile cotton swabs containing 0.85% sterile saline solution to wipe the surface of objects using a "Z" shaped wiping method. One swab should be used for each sampling point. Air samples should be collected using an Anderson six-stage impactor sampler with a sampling flow rate of 28.3 L / min and a sampling time of 10 minutes. Water samples should be collected in 50 mL portions and placed in a sterile sampling bottle. All samples must be sent to the laboratory for testing within 2 hours of collection to avoid bacterial inactivation.

[0042] Marker generation: After laboratory testing is completed, environmental markers are generated for environmental samples that have detected target multidrug-resistant bacteria. These markers include information such as sampling point location, sampling time, strain concentration, and gene sequence, and are stored in association with individual markers from the same period.

[0043] Environmental marker collection and matching rules: Clearly define the "spatiotemporal synchronization" requirement for environmental sample collection, meaning the environmental sample collection time must differ from the individual positive sample collection time by ≤6 hours, and the collection area must cover all areas where the individual stayed for ≥15 minutes within the past 24 hours; for special departments such as ICU, an additional rule of prioritizing high-frequency contact surfaces, such as ventilator interfaces, infusion pump buttons, and bedside call buttons, is added, with these areas accounting for no less than 40% of the total environmental samples; when matching markers, introduce a "microbial survival cycle coefficient" to adjust the matching time window based on the environmental survival characteristics of different multidrug-resistant strains, improving the accuracy of human-environment association.

[0044] Preferably, the spatiotemporal matching in step two specifically includes:

[0045] Time alignment: Based on the sampling time of individual markers, environmental markers collected within 4 hours before and after are selected. If the sampling time of an environmental marker is within ±2 hours of the sampling time of an individual marker, it is defined as "strong temporal correlation"; if it is within ±2-4 hours, it is defined as "weak temporal correlation". Only markers with "strong temporal correlation" are analyzed later.

[0046] Spatial alignment: Using the hospital's geographic information system, the straight-line distance between the patient bed corresponding to the individual marker and the sampling location of the environmental marker is calculated. A distance of ≤2 meters is considered "strong spatial association"; 2-5 meters is considered "weak spatial association". Only the combination of individual-environment markers with "strong temporal association + strong spatial association" is retained.

[0047] Preferably, the calculation of homology similarity in step three specifically includes: extracting the core genome sequences of strains from individual marker sites and environmental marker sites, performing multiple sequence alignment using MAFFT software, and calculating SNP differences using Mega software; if the SNP difference between a patient's individual marker site strain and the environmental marker site strain from the ward faucet is ≤5;

[0048] The homology similarity is then: ×100%;

[0049] In step three, the transmission risk modeling adds a population susceptibility factor S to the original transmission risk index R calculation formula, resulting in the optimized formula: R = Among them, the population susceptibility factor S is set according to the individual's underlying diseases, immune status, and history of invasive procedures, and the value ranges from 0.3 to 1.0. The higher the susceptibility, the larger the S value.

[0050] Preferably, the dynamic early warning response process in step four is as follows: when the system detects that the homology similarity between human-to-human or human-to-environment microorganisms is >90%, an early warning is immediately triggered, and the specific process is as follows:

[0051] Warning push: The system automatically sends warning information to the hospital infection control department, relevant department directors, and head nurses. The push methods include system pop-ups, text messages, and WeChat official account notifications, ensuring that relevant personnel receive the information within 10 minutes.

[0052] On-site verification: Infection control personnel must arrive at the scene within 30 minutes to verify the implementation of patient isolation measures, environmental cleaning and disinfection records, and collect two samples for verification, one for the individual and one for the environment.

[0053] Intervention implementation: If the verification results confirm the same source, environmental disinfection should be initiated immediately. Use chlorine-containing disinfectant at a concentration of 1000 mg / L to wipe and disinfect the surfaces of objects in the relevant areas. Use ultraviolet light to disinfect the air for 30 minutes. At the same time, screen close contacts of the patient and take isolation measures if necessary.

[0054] Path tracing technical details: When constructing the propagation network using the GTSAM graph optimization algorithm, each marked point is treated as a node, and the node attributes include spatiotemporal information and strain information; the relationships between nodes are treated as edges, and the weight of the edges is the propagation risk index R; through iterative optimization of the algorithm, edges with a weight < 0.3 are removed, and high-weight edges are retained to form the propagation path.

[0055] (2) Beneficial effects

[0056] This invention provides a method and system for tracking multidrug-resistant bacterial infections. By assigning markers to multidrug-resistant bacterial infections, it enables dynamic correlation analysis between people and between people and the environment, thereby improving public health response efficiency. Compared with existing technologies, the advantages of this invention are:

[0057] (1) Precise source tracing: Through human-environment microbial homology analysis, the error in locating pollution sources is reduced to <5 meters;

[0058] (2) Real-time dynamic response: The closed-loop feedback mechanism shortens the intervention response time to within 2 hours;

[0059] (3) Multi-source data fusion: It supports parallel analysis of clinical, environmental and spatiotemporal data, and improves the efficiency of public health decision-making. Therefore, the hospital management and public health decision-making departments involved can adopt this method to implement it. Detailed Implementation

[0060] The present invention will now be described in detail with reference to the accompanying drawings and embodiments:

[0061] The multidrug-resistant strain infection tracking system described in this embodiment includes:

[0062] Data acquisition layer module: In addition to individual tNGS detection, environmental sensor network sampling, and LIS acquisition reports, a new patient movement trajectory acquisition function has been added. Through the ward access control system and mobile nursing PDA positioning, the daily activity trajectory of patients (such as the time and path from the ward to the examination room and treatment room) is recorded, providing an accurate basis for the selection of environmental sampling points; at the same time, it connects to the hospital disinfection management system data to obtain environmental cleaning and disinfection records for subsequent transmission risk analysis (such as the transmission risk index can be reduced by 20% for environmental markers after disinfection).

[0063] Analysis layer module: Based on false positive filtering, drug resistance gene-host matching, and spatiotemporal correlation calculation, it adds strain evolution analysis function. By comparing the gene sequences of marker strains at different time periods, it analyzes the evolutionary trend of strains (such as whether new drug resistance genes have emerged) and predicts changes in transmission risk. At the same time, it implements data quality assessment function to evaluate the completeness (such as whether sampling time and location information is missing) and accuracy (such as whether the repeatability error of strain concentration detection is <10%) of the collected data to ensure the reliability of the analysis data.

[0064] The decision-making module, in addition to generating heatmaps of transmission risks, visualizing source tracing paths, and recommending intervention strategies, adds an infection trend prediction function. Based on historical data (marked data from the past 6 months and the number of infected cases), it uses the ARIMA time series model to predict the incidence of multidrug-resistant bacteria infections in various departments within the next 1-2 weeks, allowing for the early development of prevention and control plans. It also provides data export and report generation functions, supporting the export of analysis results to Excel and PDF formats, and automatically generating monthly / quarterly multidrug-resistant bacteria management reports (including indicators such as infection incidence, source tracing success rate, and intervention response time), facilitating hospital management's evaluation of prevention and control effectiveness.

[0065] Individual marker implementation process:

[0066] Step 1: Data Extraction: The results of multidrug-resistant bacteria tests are automatically extracted from the hospital's LIS system, including strain name, drug resistance gene, test date, patient information (hospital number, department, bed), etc. The extraction frequency is once per hour to ensure data real-time performance.

[0067] Step 2: False Positive Filtering: First, compare the sample with the negative control sample data in the LIS system. If the difference in the Ct value of the drug resistance gene between the sample to be tested and the negative control sample is <3, it is determined to be batch contamination, and the sample is removed. Second, check the patient's testing records for the past 7 days. If multiple tests show only one positive result and the bacterial concentration is <10... 3 CFU / mL, combined with clinical symptoms (such as no fever, normal infection indicators), is judged as a false positive; finally, it is compared with the hospital background microbial database (containing common background bacteria of each department in the past 12 months). If the strain is a common background bacteria of the department and there is no drug resistance phenotype to support it, it is filtered.

[0068] Step 3: Marker generation: For confirmed positive samples, generate individual markers in the system, including unique identifiers (such as "patient ID-test date-strain name"), spatiotemporal information (sampling time accurate to the minute, sampling location accurate to the bed), and strain information (concentration, drug resistance phenotype, gene sequence).

[0069] Precise processing of individual positive markers: Building upon the in-hospital LIS system for labeling multidrug-resistant microorganisms, a drug resistance phenotype verification step is added. In vitro drug susceptibility testing is performed on LIS-labeled positive strains to confirm that their drug resistance phenotype is consistent with gene detection results, avoiding false labeling due to gene silencing or detection errors. Simultaneously, a strain concentration grading standard is established, classifying the concentration (Wt) of microorganisms carried by individuals into high (≥10) categories. 6 CFU / mL), medium (10 3 -10 6 CFU / mL, low (<10) 3 The system uses three levels (CFU / mL) to provide accurate data support for subsequent transmission risk calculations.

[0070] Details of environmental marker implementation:

[0071] Sampling trigger conditions: Environmental sampling is automatically triggered when two or more cases of the same type of multidrug-resistant bacteria infection occur in a department (with an interval of ≤7 days), or when high-risk departments such as the ICU enter the monthly routine monitoring cycle.

[0072] Sampling Procedures: Sampling personnel must undergo professional training and use sterile cotton swabs (containing 0.85% sterile saline) to wipe surfaces using a "Z" shaped wiping method (area 10cm × 10cm). One swab should be used for each sampling point. Air samples should be collected using an Anderson six-stage impactor sampler at a sampling flow rate of 28.3L / min for 10 minutes. Water samples (such as faucet water in hospital wards or condensate from central air conditioning systems) should be collected at 50mL and placed in a sterile sampling bottle. All samples must be sent to the laboratory for testing within 2 hours of collection to avoid bacterial inactivation.

[0073] Marker generation: After laboratory testing is completed, environmental markers are generated for environmental samples that have detected target multidrug-resistant bacteria. These markers include information such as sampling point location (accurate to a specific object or area, such as "the surface of the bedside table of ICU-3 bed"), sampling time, strain concentration, and gene sequence, and are stored in association with individual markers from the same period.

[0074] Environmental marker collection and matching rules: Clearly define the "spatiotemporal synchronization" requirement for environmental sample collection, meaning the time of environmental sample collection must differ from the time of individual positive specimen collection by ≤6 hours, and the collection area must cover all areas where the individual has stayed for ≥15 minutes within the past 24 hours (such as the 1.5-meter radius around the ward bed, toilets, treatment rooms, etc.). For special departments such as the ICU, an additional rule prioritizing high-frequency contact surfaces is added, such as ventilator interfaces, infusion pump buttons, bedside call buttons, etc., with these areas accounting for no less than 40% of the total environmental samples. When matching markers, introduce a "microbial survival period coefficient." Based on the environmental survival characteristics of different multidrug-resistant strains (e.g., methicillin-resistant Staphylococcus aureus can survive for more than 7 days on dry surfaces, while carbapenem-resistant Enterobacteriaceae survive for about 3 days), adjust the matching time window to improve the accuracy of human-environment association.

[0075] Specific steps for spatiotemporal matching:

[0076] Time alignment: Based on the sampling time of individual markers, environmental markers collected within 4 hours before and after are selected. If the sampling time of the environmental marker is within ±2 hours of the sampling time of the individual marker, it is defined as "strong time correlation"; if it is within ±2-4 hours, it is defined as "weak time correlation". Only markers with "strong time correlation" are analyzed later.

[0077] Spatial alignment: Using the hospital's geographic information system (GIS), the straight-line distance between the patient bed corresponding to the individual marker and the sampling location of the environmental marker is calculated. A distance of ≤2 meters is considered "strong spatial association"; 2-5 meters is considered "weak spatial association". Only the combination of individual-environment markers with "strong temporal association + strong spatial association" is retained.

[0078] Example of homology similarity calculation: Taking carbapenem-resistant Escherichia coli as an example, the core genome sequences of strains from individual and environmental marker sites are extracted, multiple sequence alignment is performed using MAFFT software, and SNP differences are calculated using Mega software. If the SNP differences between a patient's individual marker strain and the environmental marker strain from the ward faucet are 3 (≤5), then the homology similarity is (total number of SNPs in the core genome - number of differential SNPs) / total number of SNPs in the core genome × 100%. Assuming the total number of SNPs in the core genome is 1000, the homology similarity is (1000-3) / 1000 × 100% = 99.7%, and they are determined to be homologous strains.

[0079] Example of calculating the transmission risk index: The concentration (Wt) of methicillin-resistant Staphylococcus aureus (MRSA) at an individual marker site is 5 × 10^5 CFU / mL, and the density (Ce) of the same strain at the corresponding environmental marker site is 3 × 10^3 CFU / cm². If there is only one set of data, then the transmission risk index R = (5 × 10^5 × 3 × 10^3) ​​ / √[(5 × 10^5)² × (3 × 10^3)²] = (1.5 × 10^9) / (5 × 10^5 × 3 × 10^3) ​​= 1.0, indicating an extremely high transmission risk.

[0080] Optimization of Transmission Risk Modeling: Based on the original formula for calculating the transmission risk index (R), a population susceptibility factor (S) is added. The optimized formula is: R = Among them, the population susceptibility factor (S) is determined based on individual underlying diseases (such as diabetes, malignant tumors), immune status (such as use of immunosuppressants, white blood cell count <2×10⁻⁶), etc. 9 The S-value is set for the risk index (S / L) and the history of invasive procedures (such as endotracheal intubation and central venous catheterization), with a range of 0.3-1.0 (the higher the susceptibility, the larger the S-value). For example, the S-value for cancer patients receiving chemotherapy is set to 1.0, and the S-value for healthy caregivers is set to 0.3, making the risk index more consistent with actual transmission scenarios.

[0081] Dynamic early warning response process: When the system detects that the homology similarity between human-to-human or human-to-environment microorganisms is greater than 90%, an early warning is immediately triggered. The specific process is as follows:

[0082] Warning push: The system automatically sends warning information (including warning type, involved strains, associated markers, and homology similarity) to the hospital infection control department, relevant department directors, and head nurses. The push methods include system pop-ups, SMS, and WeChat official account notifications, ensuring that relevant personnel receive the information within 10 minutes.

[0083] On-site verification: Infection control personnel must arrive at the site within 30 minutes to verify the implementation of patient isolation measures, environmental cleaning and disinfection records, and collect verification samples (2 samples each for individuals and the environment).

[0084] Intervention Implementation: If the verification results confirm the same source, immediately initiate environmental disinfection (use chlorine-containing disinfectant at a concentration of 1000 mg / L to wipe and disinfect the surfaces of objects in the relevant areas, and disinfect the air with ultraviolet light for 30 minutes), and at the same time screen close contacts (such as patients in the same ward), and take isolation measures if necessary.

[0085] Path tracing technology details: When constructing the propagation network using the GTSAM (Georgia Tech Smoothing and Mapping Library) graph optimization algorithm, each marker point is treated as a node, with node attributes including spatiotemporal information and strain information; the relationships between nodes are treated as edges, with the edge weight being the propagation risk index (R). Through iterative optimization, edges with weights <0.3 are removed, and high-weight edges are retained to form the propagation path. For example, in a certain propagation network, the edge weight between "Patient A marker point" and "ward bedside table environmental marker point" is 0.9, and the edge weight between "ward bedside table environmental marker point" and "Patient B marker point" is 0.85. Therefore, the propagation path can be inferred to be "Patient A → Ward Bedside Table → Patient B," thus locating the source of contamination as Patient A or the ward bedside table.

[0086] By comparing with traditional technologies, the quantitative advantages of this technology are highlighted:

[0087] Comparison Dimensions This technology Traditional techniques (PCR / culture method + manual registration) Data integration scope Integrating multi-source data such as individual data, environmental data, patient trajectories, and disinfection records, covering the entire hospital. Integrating only individual test data, limited to a single department False positive rate Through multi-level filtering, the false positive rate is ≤3%. Without systematic filtering, the false positive rate is ≥15% (tNGS detection). Source tracing accuracy Based on gene homology and graph optimization algorithms, the source tracing accuracy is ≥90%. Relying on human experience for judgment results in a traceability accuracy rate of ≤60%. Intervention response time From early warning triggering to intervention implementation, the average response time is ≤1.5 hours. From infection detection to intervention implementation, the average response time is ≥24 hours. Infection control effectiveness It can reduce the incidence of cluster infections of multidrug-resistant bacteria by ≥40%. Cluster infections occur repeatedly and are poorly controlled. Human resource costs Automated data collection and analysis reduces manual data processing workload by 80%. Relying on manual registration and data integration results in high labor costs.

[0088] (I) Expanding application scenarios within hospitals

[0089] Departmental applications: In addition to the ICU, it can be applied to high-risk departments such as neonatology, hematology, and oncology. For example, premature infants in the neonatal ward have low immunity and are prone to multidrug-resistant bacterial infections. This technology can monitor the microbial status of the incubator environment (air and surfaces) in real time, detect contamination in a timely manner, and intervene to reduce the risk of neonatal infection. Hematology patients have a high risk of infection due to leukopenia caused by chemotherapy. This technology can track the patient's activity trajectory and environmental contact during chemotherapy, accurately locate the source of transmission, and prevent the spread of infection.

[0090] Application in hospital infection control departments: The hospital infection control department can use this technology to monitor the dynamics of multidrug-resistant bacterial infections throughout the hospital in real time, view risk heat maps for each department, and conduct key supervision of high-risk departments (such as departments with a risk index > 0.7); at the same time, it can use the source tracing reports and intervention effect data generated by the system to evaluate the quality of infection control work in each department, which can serve as a reference for department performance evaluation.

[0091] (II) Extended Applications in the Public Health Field

[0092] Regional-level multidrug-resistant bacteria surveillance: This technology integrates data from multiple hospitals into a regional public health platform, enabling cross-hospital data sharing on multidrug-resistant bacteria. For example, when multiple hospitals within a region experience cases of infection with the same type and high homology of multidrug-resistant bacteria, the platform can be used to trace the source of infection (such as a medical device supplier or a centralized disinfection service provider), allowing for timely implementation of regional-level prevention and control measures (such as suspending the use of relevant devices and strengthening disinfection supervision) to prevent cross-hospital transmission.

[0093] Emergency Response to Public Health Emergencies: In public health emergencies such as outbreaks of multidrug-resistant bacteria, this technology can quickly provide data on the spatiotemporal distribution, transmission routes, and strain characteristics of infected cases, providing a scientific basis for public health departments to formulate emergency response plans (such as delineating control areas and allocating control materials), shortening the time for handling the incident, and reducing the scope of infection spread.

[0094] (III) Refinement of Application Value

[0095] Clinical value: Through precise source tracing and rapid intervention, it can reduce the length of hospital stay (by an average of 5-7 days) and increase treatment costs (by an average of 20,000-30,000 RMB per case) caused by multidrug-resistant bacterial infections, while also reducing infection-related mortality (by 15%-20%) and improving patient prognosis.

[0096] Management Value: It helps hospitals standardize multidrug-resistant bacteria management processes, improve the informatization and intelligentization of infection control, and reduce loopholes in manual management; at the same time, by quantifying the effectiveness of prevention and control through data, it provides a basis for hospitals to optimize resource allocation (such as increasing investment in disinfection equipment in high-risk departments).

[0097] Social value: It reduces the spread and diffusion of multidrug-resistant bacteria, thereby lowering their threat to public health and safety; at the same time, it provides a referable technical paradigm for tracking and controlling other regions and other types of pathogens (such as viruses and fungi), thereby improving the overall public health prevention and control capabilities.

[0098] Although the present invention has been described in detail above with general descriptions and specific embodiments, modifications or improvements can be made to it, which will be obvious to those skilled in the art. Therefore, all such modifications or improvements made without departing from the spirit of the present invention fall within the scope of protection claimed by the present invention.

Claims

1. A tracking system for infections caused by multidrug-resistant strains, characterized in that, include: Data acquisition layer module: individual tNGS detection, real-time sampling of environmental sensor network, and LiS acquisition report; Analysis layer modules: false positive filtering, drug resistance gene-host matching, spatiotemporal correlation calculation; Decision-making module: generates heat maps of transmission risks, visualizes source tracing paths, recommends intervention strategies, and allows for real-time dialogue between hospital infection control departments and other relevant units.

2. The tracking system for multidrug-resistant bacterial infections according to claim 1, characterized in that, In addition to individual tNGS detection, environmental sensor network sampling, and LIS data acquisition reports, the data acquisition layer module adds a patient movement trajectory acquisition function. Through the ward access control system and mobile nursing PDA positioning, it records the patient's daily activity trajectory, providing an accurate basis for the selection of environmental sampling points. At the same time, it connects to the hospital's disinfection management system to obtain environmental cleaning and disinfection records for subsequent transmission risk analysis.

3. The tracking system for multidrug-resistant bacterial infections according to claim 1, characterized in that, The analysis layer module, based on false positive filtering, drug resistance gene-host matching, and spatiotemporal correlation calculation, adds a strain evolution analysis function. By comparing the gene sequences of marker strains at different time periods, it analyzes the evolutionary trend of strains and predicts changes in transmission risk. At the same time, it implements a data quality assessment function to evaluate the completeness and accuracy of the collected data to ensure the reliability of the analysis data.

4. The tracking system for multidrug-resistant bacterial infections according to claim 1, characterized in that, In addition to generating heatmaps of transmission risks, visualizing source tracing paths, and recommending intervention strategies, the decision-making module now includes an infection trend prediction function. Based on historical data and using the ARIMA time series model, it predicts the incidence of multidrug-resistant bacteria infections in various departments within the next 1-2 weeks, allowing for the early development of prevention and control plans. It also provides data export and report generation functions, supporting the export of analysis results to Excel and PDF formats, and automatically generating monthly / quarterly multidrug-resistant bacteria management reports to facilitate hospital management's evaluation of prevention and control effectiveness.

5. A method for tracking infections caused by multidrug-resistant strains, characterized in that, Includes the following steps: Step 1: Dynamic extraction and filtration of individual positive microbial markers (1) Implement individual marker points: By utilizing the existing in-hospital LIS system, we can define the markers for multidrug-resistant microorganisms, while filtering out false positives, batch contamination, and background microorganisms to identify true positives; we can then implement marker points on this data and prepare for simultaneous and concurrent specimen sources of the same infection within the hospital. Step 2: Spatiotemporal synchronous acquisition and matching of environmental microbial markers (1) Implement environmental markers: For cases with high suspicion of environmental pollution or during routine environmental sampling in the ICU, environmental samples can be collected simultaneously from the patient's activity trajectory to detect the same microbial targets and generate markers; (2) Spatiotemporal matching: Aligning individual positive markers with the same type of microorganisms in their environment according to time and space dimensions; Step 3: Human-to-human and human-to-environment microbial homology analysis and transmission risk modeling (1) Calculate homology similarity: Determine the homology between humans and between humans and environmental microorganisms based on the similarity of microbial genome sequences; (2) Quantitative transmission risk index: R= ; Among them, W t For the concentration of microorganisms carried by an individual, C e This represents the corresponding microbial density in the environment. Step 4: Automated intervention mechanism based on risk feedback (1) Dynamic early warning: If the homology similarity of human-to-human and human-to-environment microorganisms is >90%, an environmental disinfection early warning will be triggered; (2) Path tracing: The microbial transmission network is constructed by graph optimization algorithm to locate the source of pollution.

6. The method for tracking infections caused by multidrug-resistant strains according to claim 5, characterized in that, Step one, implementing individual marker points, specifically includes: Data extraction: The results of multidrug-resistant bacteria tests are automatically extracted from the hospital's LIS system, including strain name, drug resistance gene, test date, and patient information data. The extraction frequency is once per hour to ensure data real-time performance. False positive filtering: First, compare the sample with the negative control sample data in the LIS system. If the difference in the Ct value of the drug resistance gene between the sample to be tested and the negative control sample is <3, it is determined to be batch contamination, and the sample is removed. Second, check the patient's testing records for the past 7 days. If multiple tests show only one positive result and the bacterial concentration is <10, the false positive is removed. 3 CFU / mL, combined with clinical symptoms, such as no fever and normal infection indicators, is judged as a false positive; finally, compared with the hospital's background microbial database, if the strain is a common background bacterium in the department and there is no drug resistance phenotype to support it, it is filtered. Marker generation: For confirmed positive samples, individual markers are generated in the system, including unique identifiers, spatiotemporal information, and strain information; Building upon the in-hospital LIS system for labeling multidrug-resistant microorganisms, a drug resistance phenotype verification step was added. In vitro drug susceptibility testing was performed on LIS-labeled positive strains to confirm that their drug resistance phenotype was consistent with the gene detection results, avoiding mislabeling due to gene silencing or detection errors. Simultaneously, a strain concentration grading standard was established, classifying the concentration W of microorganisms carried by individuals. t Divided into high: W t ≥10 6 CFU / mL, W t 10 3 -10 6 CFU / mL, low: W t <10 3 The CFU / mL level provides accurate data support for subsequent transmission risk calculations.

7. The method for tracking infections caused by multidrug-resistant strains according to claim 5, characterized in that, Step two, in implementing environmental markers, specifically includes: Sampling trigger conditions: Environmental sampling will be automatically triggered when two or more cases of the same type of multidrug-resistant bacteria infection occur in a department, with an interval of ≤7 days, or when high-risk departments such as the ICU enter the monthly routine monitoring cycle; Sampling Procedures: Sampling personnel must undergo professional training and use sterile cotton swabs containing 0.85% sterile saline solution to wipe the surface of objects using a "Z" shaped wiping method. One swab should be used for each sampling point. Air samples should be collected using an Anderson six-stage impactor sampler with a sampling flow rate of 28.3 L / min and a sampling time of 10 minutes. Water samples should be collected in 50 mL portions and placed in a sterile sampling bottle. All samples must be sent to the laboratory for testing within 2 hours of collection to avoid bacterial inactivation. Marker generation: After laboratory testing is completed, environmental markers are generated for environmental samples that have detected target multidrug-resistant bacteria. These markers include information such as sampling point location, sampling time, strain concentration, and gene sequence, and are stored in association with individual markers from the same period. Environmental marker collection and matching rules: Clearly define the "spatiotemporal synchronization" requirement for environmental sample collection, meaning the environmental sample collection time must differ from the individual positive sample collection time by ≤6 hours, and the collection area must cover all areas where the individual stayed for ≥15 minutes within the past 24 hours; for special departments such as ICU, an additional rule of prioritizing high-frequency contact surfaces, such as ventilator interfaces, infusion pump buttons, and bedside call buttons, is added, with these areas accounting for no less than 40% of the total environmental samples; when matching markers, introduce a "microbial survival cycle coefficient" to adjust the matching time window based on the environmental survival characteristics of different multidrug-resistant strains, improving the accuracy of human-environment association.

8. The method for tracking infections caused by multidrug-resistant strains according to claim 5, characterized in that, Step two, spatiotemporal matching, specifically includes: Time alignment: Based on the sampling time of individual markers, environmental markers collected within 4 hours before and after are selected. If the sampling time of an environmental marker is within ±2 hours of the sampling time of an individual marker, it is defined as "strong temporal correlation"; if it is within ±2-4 hours, it is defined as "weak temporal correlation". Only markers with "strong temporal correlation" are analyzed later. Spatial alignment: Using the hospital's geographic information system, the straight-line distance between the patient bed corresponding to the individual marker and the sampling location of the environmental marker is calculated. A distance of ≤2 meters is considered "strong spatial association"; 2-5 meters is considered "weak spatial association". Only the combination of individual-environment markers with "strong temporal association + strong spatial association" is retained.

9. The method for tracking infections caused by multidrug-resistant strains according to claim 5, characterized in that, Step 3, calculating homology similarity, specifically includes: extracting the core genome sequences of strains from individual marker sites and environmental marker sites, performing multiple sequence alignment using MAFFT software, and calculating SNP differences using Mega software; if the SNP difference between a patient's individual marker site strain and the environmental marker site strain from the ward faucet is ≤5; The homology similarity is then: ×100%; In step three, the transmission risk modeling adds a population susceptibility factor S to the original transmission risk index R calculation formula, resulting in the optimized formula: R = Among them, the population susceptibility factor S is set according to the individual's underlying diseases, immune status, and history of invasive procedures, and the value ranges from 0.3 to 1.

0. The higher the susceptibility, the larger the S value.

10. The method for tracking infections caused by multidrug-resistant strains according to claim 5, characterized in that, Step four, dynamic early warning response process: When the system detects that the homology similarity between human-to-human or human-to-environment microorganisms is >90%, an early warning is immediately triggered. The specific process is as follows: Warning push: The system automatically sends warning information to the hospital infection control department, relevant department directors, and head nurses. The push methods include system pop-ups, text messages, and WeChat official account notifications, ensuring that relevant personnel receive the information within 10 minutes. On-site verification: Infection control personnel must arrive at the scene within 30 minutes to verify the implementation of patient isolation measures, environmental cleaning and disinfection records, and collect two samples for verification, one for the individual and one for the environment. Intervention implementation: If the verification results confirm the same source, environmental disinfection should be initiated immediately. Use chlorine-containing disinfectant at a concentration of 1000 mg / L to wipe and disinfect the surfaces of objects in the relevant areas. Use ultraviolet light to disinfect the air for 30 minutes. At the same time, screen close contacts of the patient and take isolation measures if necessary. Path tracing technical details: When constructing the propagation network using the GTSAM graph optimization algorithm, each marked point is treated as a node, and the node attributes include spatiotemporal information and strain information; the relationships between nodes are treated as edges, and the weight of the edges is the propagation risk index R; Through iterative optimization of the algorithm, edges with weights less than 0.3 are removed, while high-weight edges are retained to form a propagation path.

Images