Method and system for evaluating effectiveness of an anti-infective drug

Abstract

The application discloses an effectiveness evaluation method and system of anti-infective drugs, and relates to the technical field of data analysis, which comprises the following steps: collecting oral microbial samples of patients in the whole treatment cycle, generating time-characteristic personalized dynamic atlas of oral microbiome of the patients, capturing the composition structure and dynamic evolution law of oral microbial community; integrating multi-dimensional life habit data of the patients, extracting life behavior patterns and individual preference characteristics of the patients, coupling analysis of the time-characteristic personalized dynamic atlas of oral microbiome of the patients, establishing a patient behavior-microorganism correlation model, and obtaining the regulation path of external behavior to the microbiome; based on the regulation path of external behavior to the microbiome, introducing the physical fitness parameters of the patients as compensation factors, establishing a comprehensive evaluation function of the anti-infective drug efficacy of the patients, and realizing personalized effective evaluation of the patients for the anti-infective drugs. The application improves the accuracy, forward-looking and clinical practicability of the efficacy evaluation.

Description

Technical Field

[0001] This invention relates to the field of data analysis technology, specifically to a method and system for evaluating the effectiveness of anti-infective drugs. Background Technology

[0002] Traditional methods for evaluating the effectiveness of anti-infective drugs mainly rely on single and lagging clinical endpoint indicators such as pathogen clearance rate and symptom relief time. The essential flaw lies in the use of a static and isolated analytical perspective. Most existing technologies are based on microbial sampling at a few time points before and after treatment, which cannot capture the continuous and dynamic evolution of the oral microbiome throughout the entire treatment cycle, resulting in the loss of key temporal information. Evaluation systems generally ignore the dynamic differences in individual patient behavior patterns and their temporal regulatory effects on the microbiome, and also fail to include the patient's basic immune and metabolic status as a systemic compensatory factor in efficacy analysis. This leads to the current methods having a one-sided analytical dimension, weak causal inference ability, and an inability to link clinical events with microbial functional evolution. It is difficult to achieve truly personalized and prospective efficacy assessment and intervention guidance, and there are significant limitations in the context of precision medicine. Summary of the Invention

[0003] To address the aforementioned technical problems, this paper provides a method and system for evaluating the effectiveness of anti-infective drugs. This technical solution solves the problem of the difficulty in achieving truly personalized and prospective efficacy assessment.

[0004] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A method for evaluating the effectiveness of an anti-infective drug, comprising: S1. Collect oral microbial samples from patients throughout the entire treatment cycle, generate personalized dynamic maps of the patient's oral microbiome with time characteristics, and capture the composition structure and dynamic evolution of the oral microbial community.

[0005] S2. Integrate multidimensional lifestyle data of patients, extract patients’ lifestyle behavior patterns and personality preference characteristics, and couple them with the personalized dynamic map of patients’ oral microbiome with time characteristics to establish a patient behavior-microbiome association model and obtain the regulatory path of external behavior on the microbiome. S3. Based on the regulatory pathways of external behavior on the microbiome, patient physical fitness parameters are introduced as compensating factors to establish a comprehensive evaluation function for the efficacy of anti-infective drugs, thereby achieving personalized and effective evaluation of anti-infective drugs for patients.

[0006] Preferably, step S1 specifically includes: Using a standardized, non-irritating saliva collector, non-irritating whole saliva from patients is collected. The sample taken before the start of treatment is used as the baseline sample for initial oral microbial sample collection. Samples are collected at fixed time points on days 1, 3, and 7 after the patient starts medication. Samples are collected again at the predetermined follow-up time point after the patient finishes treatment. If a specific clinical event occurs, additional samples are collected within 24-48 hours after the event, so as to achieve dynamic monitoring of oral microbial samples throughout the entire treatment cycle of the patient. The specific clinical events include: patients reporting significant changes in oral symptoms, clinicians observing changes in the oral mucosa, completion of chemotherapy drug infusion, initiation of prophylactic antibiotics, and significant changes in patients' lifestyle habits.

[0007] Preferably, step S1 further includes: Total DNA was extracted from the collected samples using a DNA extraction kit validated for low biomass oral samples. The concentration and integrity of the DNA were assessed using a Qubit fluorescence quantitative quantification system and an Agilent Bioanalyzer, respectively. Qualified DNA samples were screened and libraries were constructed. For qualified DNA libraries, shotgun metagenomic sequencing was performed on the Illumina NovaSeq platform, aiming to produce 10-20 million raw paired-end reads per sample. FastP was used to perform quality control on the raw data to remove low-quality bases, adapters and excessively short reads. Bowtie2 was used to align the quality-controlled reads to the human reference genome, remove host-derived reads, and retain non-host-derived microbial reads for data preprocessing. The Kraken2 method was used to classify and quantify the abundance of microbial species, enabling rapid identification of microbial species present in patient samples. Based on the HUMAnN3 process, read lengths are aligned to the UniRef90 integrated gene database to quantify the abundance of microbial gene families. These are then mapped to the MetaCyc metabolic pathway database to calculate pathway coverage and abundance, generate a metabolic pathway abundance table, and quantitatively assess the functional potential of microbial communities.

[0008] Preferably, step S1 further includes: Based on the metabolic pathway abundance table, the abundance of microbial species at each time point is regarded as a point in a high-dimensional species space. A time-series point cloud is constructed during the patient's treatment. The persistent homology algorithm is applied to identify the topological stability features that persist in the point cloud at different time scales. The features that persist in multiple time windows are used as the personalized core microbial components of the patient, thus obtaining the personalized oral microbial characteristics of the patient during the treatment. Based on the abundance of microbial species at each time point of the collected samples, the data at each time point was smoothed using cubic spline interpolation. The smoothing parameters were automatically selected through generalized cross-validation to complete the smoothing of the time series data of the samples. Using a change point detection algorithm, a fixed time window length is set, and the statistical characteristics of microbial species at each time point of the collected samples within the time window are calculated. The points where the statistical characteristics of microbial species change are automatically identified as the points where the microbial community status changes. Using time points, sample IDs, species abundance, metabolic pathway abundance, species diversity index, and specific clinical events as multi-dimensional data, a personalized dynamic map of the patient's oral microbiome with time characteristics is generated to capture the composition structure and dynamic evolution of the patient's oral microbiome.

[0009] Preferably, step S2 specifically includes: Based on the patient diary APP, we can obtain patients' daily dietary records, oral hygiene practices, and records of the time and dosage of anti-infective drugs. Based on wearable devices for patients, the system monitors patients' sleep status in real time and automatically synchronizes patients' daily step count and duration of moderate to high intensity exercise. Based on the smart toothbrush backend data, data on the duration, frequency and coverage area of ​​each brushing session are collected from the patient. By integrating patient diary app, patient wearable device and smart toothbrush backend data, multidimensional patient lifestyle data is obtained, unified to the same timestamp, and patient discrete behavioral events are transformed into frequency indicators, and patient data is processed in a structured manner. Based on the patient's daily dietary records, the types and quantities of food consumed by the patient were obtained, and a frequency matrix of the patient's daily food categories was established.

[0010] Preferably, step S2 further includes: Using the patient's daily food category frequency matrix as input, the latent Dirichlet Allocation Algorithm is used to randomly select any patient's daily food category distribution to minimize perplexity, determine the number of topics, identify 3-5 potential dietary topics and their probability distributions, and obtain the patient's dietary pattern characteristics. For wearable devices used by patients, the mean, variance, and regularity of patients' sleep states over 24 hours were calculated, and statistical characteristics of patients' sleep states were extracted. Calculate the patient's weekly exercise duration, frequency, and regularity, and extract statistical characteristics of the patient's exercise data; Calculate the brushing time, frequency, and uniformity of coverage area for patients, and extract statistical characteristics of oral hygiene. By integrating statistical characteristics of patients' sleep status, sleep status, and oral hygiene, we can obtain patients' lifestyle behavior patterns and personality preferences. Based on each microbial sample collection point, different behavioral impact lag time windows are defined. For each microbial sample, specific behavioral characteristics of the patient within the behavioral impact lag time window are extracted. A corresponding lag behavior feature vector is constructed for each microbial sample, and a lag behavior feature matrix is ​​established. The time windows for the effects of different behaviors include: 0-6 hours as the immediate effect, applicable to oral hygiene behaviors; 6-24 hours as the short-term effect, applicable to a single eating event; 24-72 hours as the medium-term effect, applicable to eating patterns and sleep states; and more than 72 hours as the long-term effect, applicable to lifestyle habits.

[0011] Preferably, step S2 further includes: Based on the abundance of microbial species, metabolic pathways, and species diversity index of the samples, normalization was performed to establish a microbial characteristic matrix. Using the lagged behavioral feature matrix as the predictor variable and the microbial feature matrix as the response variable, the optimal regularization parameter was selected through 10-fold cross-validation. Behavioral variables related to microbial features were screened. Bootstrap resampling was used to calculate the 95% confidence interval of the regression coefficients. Stable associations with confidence intervals not containing zero were retained. Coupled with the personalized dynamic map of the patient's oral microbiome with time characteristics, a patient behavior-microbial association model was established to obtain standardized regression coefficients and quantitatively assess the direction and magnitude of the influence of each behavioral feature on specific microbial features. Using lagging behavioral characteristics and microbial characteristics as nodes and standardized regression coefficients as edges, a patient behavior-microbial regulation network diagram is established. The sum of the absolute values ​​of the outgoing edge coefficients of each patient behavior node is calculated, and the microbial regulation potential index of the behavior is defined to obtain the regulation path of external behavior on the microbiome.

[0012] Preferably, step S3 specifically includes: Collect patients' pre-treatment immune status indicators and baseline values ​​of liver and kidney function metabolism; Based on the baseline sample of patients before the start of treatment and the key node sample of patients on the 7th day after medication, combined with the oral mucositis assessment scale, the clinical inflammation dimension indicators of patients were calculated. Based on the personalized dynamic map of the patient's oral microbiome with time characteristics, the improvement rate of the patient's oral microbial ecological health index on day 7 compared with the baseline sample was calculated, and the patient's microbial ecological improvement dimension index was obtained.

[0013] Preferably, step S3 further includes: Based on the metabolic pathway abundance table, the increase rate of specific microbial metabolic pathways related to mucosal repair on day 7 was calculated to obtain the patient's metabolic function recovery dimension index. Standardize the indicators based on the patient's clinical inflammation dimension, microbial ecology improvement dimension, and metabolic function recovery dimension; Using principal component analysis, the first principal component with the highest variance explanation rate is extracted as a comprehensive index of the efficacy of anti-infective drugs for patients. Based on the linear expression of this principal component, a comprehensive evaluation function of the efficacy of anti-infective drugs is obtained, generating a multidimensional efficacy vector for different medication regimens, thereby realizing personalized and effective evaluation of anti-infective drugs for patients.

[0014] Furthermore, an efficacy evaluation system for anti-infective drugs includes: Data acquisition module, adjustment path module, and personalized effective evaluation module; Among them, the adjustment path module is electrically connected to the data acquisition module, and the personalized effective evaluation module is electrically connected to the adjustment path module; The data acquisition module collects oral microbial samples from patients throughout the entire treatment cycle, generates personalized dynamic maps of the patient's oral microbiome with time characteristics, and captures the composition and dynamic evolution of the oral microbial community. The regulation pathway module integrates multidimensional lifestyle data of patients, extracts patients' lifestyle behavior patterns and personality preference characteristics, and performs coupled analysis with a personalized dynamic map of the patient's oral microbiome with time characteristics to establish a patient behavior-microbiome association model and obtain the regulation pathway of external behavior on the microbiome. The personalized effective evaluation module, based on the obtained regulatory pathway of external behavior on the microbiome, introduces the patient's physical fitness parameters as a compensating factor, establishes a comprehensive evaluation function for the efficacy of anti-infective drugs, and realizes personalized effective evaluation of patients for anti-infective drugs.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention proposes a method and system for evaluating the effectiveness of anti-infective drugs. The beneficial effects of this invention lie in the systematic integration and coupling analysis of high-frequency, event-triggered, full-cycle dynamic monitoring of oral microbiota with multidimensional real-time behavioral data and patient physical fitness parameters. This constructs a comprehensive evaluation system integrating a dynamic microbial map, behavioral regulation pathways, and individualized efficacy functions. It overcomes the limitations of traditional static, single-dimensional evaluations, achieving continuous and accurate capture and quantitative analysis of the oral microecological structure, functional dynamics, and their interaction with clinical events and individual behaviors during treatment. By establishing a behavior-microbial correlation model and introducing physical fitness compensation factors, it can not only comprehensively and personally evaluate the efficacy of anti-infective drugs from multiple dimensions such as microbial ecological health, metabolic function recovery, and clinical inflammation relief, but also identify key behavioral regulation pathways. This provides interpretable and operable dynamic decision-making basis for real-time microecological intervention, medication regimen optimization, and personalized health management during treatment, improving the accuracy, foresight, and clinical applicability of efficacy evaluation. Attached Figure Description

[0016] Figure 1 This is a flowchart of a method for evaluating the effectiveness of an anti-infective drug. Figure 2 This is a framework diagram of an efficacy evaluation system for an anti-infective drug. Detailed Implementation

[0017] The following description is intended to disclose the invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious variations will occur to those skilled in the art.

[0018] Reference Figure 1 As shown, a method for evaluating the effectiveness of an anti-infective drug includes: S1. Collect oral microbial samples from patients throughout the entire treatment cycle, generate personalized dynamic maps of the patient's oral microbiome with time characteristics, and capture the composition structure and dynamic evolution of the oral microbial community. Step S1 specifically includes: Using a standardized, non-irritating saliva collector, non-irritating whole saliva from patients is collected. The sample taken before the start of treatment is used as the baseline sample for initial oral microbial sample collection. Samples are collected at fixed time points on days 1, 3, and 7 after the patient starts medication. Samples are collected again at the predetermined follow-up time point after the patient finishes treatment. If a specific clinical event occurs, additional samples are collected within 24-48 hours after the event, so as to achieve dynamic monitoring of oral microbial samples throughout the entire treatment cycle of the patient. The specific clinical events include: patients reporting significant changes in oral symptoms, clinicians observing changes in the oral mucosa, completion of chemotherapy drug infusion, initiation of prophylactic antibiotics, and significant changes in patients' lifestyle habits.

[0019] Step S1 also includes: Total DNA was extracted from the collected samples using a DNA extraction kit validated for low biomass oral samples. The concentration and integrity of the DNA were assessed using a Qubit fluorescence quantitative quantification system and an Agilent Bioanalyzer, respectively. Qualified DNA samples were screened and libraries were constructed. For qualified DNA libraries, shotgun metagenomic sequencing was performed on the Illumina NovaSeq platform, aiming to produce 10-20 million raw paired-end reads per sample. FastP was used to perform quality control on the raw data to remove low-quality bases, adapters and excessively short reads. Bowtie2 was used to align the quality-controlled reads to the human reference genome, remove host-derived reads, and retain non-host-derived microbial reads for data preprocessing. The Kraken2 method was used to classify and quantify the abundance of microbial species, enabling rapid identification of microbial species present in patient samples. Based on the HUMAnN3 process, read lengths are aligned to the UniRef90 integrated gene database to quantify the abundance of microbial gene families. These are then mapped to the MetaCyc metabolic pathway database to calculate pathway coverage and abundance, generate a metabolic pathway abundance table, and quantitatively assess the functional potential of microbial communities.

[0020] Step S1 also includes: Based on the metabolic pathway abundance table, the abundance of microbial species at each time point is regarded as a point in a high-dimensional species space. A time-series point cloud is constructed during the patient's treatment. The persistent homology algorithm is applied to identify the topological stability features that persist in the point cloud at different time scales. The features that persist in multiple time windows are used as the personalized core microbial components of the patient, thus obtaining the personalized oral microbial characteristics of the patient during the treatment. Based on the abundance of microbial species at each time point of the collected samples, the data at each time point was smoothed using cubic spline interpolation. The smoothing parameters were automatically selected through generalized cross-validation to complete the smoothing of the time series data of the samples. Using a change point detection algorithm, a fixed time window length is set, and the statistical characteristics of microbial species at each time point of the collected samples within the time window are calculated. The points where the statistical characteristics of microbial species change are automatically identified as the points where the microbial community status changes. Using time points, sample IDs, species abundance, metabolic pathway abundance, species diversity index, and specific clinical events as multi-dimensional data, a personalized dynamic map of the patient's oral microbiome with time characteristics is generated to capture the composition structure and dynamic evolution of the patient's oral microbiome.

[0021] When using it, please refer to the steps outlined above: Existing oral microbiome monitoring technologies mostly rely on static sampling at a few time points, making it difficult to capture the continuous dynamic changes of the microbiota throughout the entire treatment cycle. Their analytical dimensions are limited, failing to effectively link clinical events with microbial functional evolution. This approach, through high-frequency, event-triggered dynamic sampling, combined with metagenomic time-series analysis, topological stability identification, and mutation point detection, constructs a multi-dimensional and personalized dynamic map of the oral microbiota. This method not only achieves continuous and accurate capture of the microbial community structure and functional evolution but also correlates it with key clinical events, providing real-time and accurate scientific evidence for microecological monitoring and personalized intervention during treatment.

[0022] S2. Integrate multidimensional lifestyle data of patients, extract patients’ lifestyle behavior patterns and personality preference characteristics, and couple them with the personalized dynamic map of patients’ oral microbiome with time characteristics to establish a patient behavior-microbiome association model and obtain the regulatory path of external behavior on the microbiome. Step S2 specifically includes: Based on the patient diary APP, we can obtain patients' daily dietary records, oral hygiene practices, and records of the time and dosage of anti-infective drugs. Based on wearable devices for patients, the system monitors patients' sleep status in real time and automatically synchronizes patients' daily step count and duration of moderate to high intensity exercise. Based on the smart toothbrush backend data, data on the duration, frequency and coverage area of ​​each brushing session are collected from the patient. By integrating patient diary app, patient wearable device and smart toothbrush backend data, multidimensional patient lifestyle data is obtained, unified to the same timestamp, and patient discrete behavioral events are transformed into frequency indicators, and patient data is processed in a structured manner. Based on the patient's daily dietary records, the types and quantities of food consumed by the patient were obtained, and a frequency matrix of the patient's daily food categories was established.

[0023] Step S2 also includes: Using the patient's daily food category frequency matrix as input, the latent Dirichlet Allocation Algorithm is used to randomly select any patient's daily food category distribution to minimize perplexity, determine the number of topics, identify 3-5 potential dietary topics and their probability distributions, and obtain the patient's dietary pattern characteristics. For wearable devices used by patients, the mean, variance, and regularity of patients' sleep states over 24 hours were calculated, and statistical characteristics of patients' sleep states were extracted. Calculate the patient's weekly exercise duration, frequency, and regularity, and extract statistical characteristics of the patient's exercise data; Calculate the brushing time, frequency, and uniformity of coverage area for patients, and extract statistical characteristics of oral hygiene. By integrating statistical characteristics of patients' sleep status, sleep status, and oral hygiene, we can obtain patients' lifestyle behavior patterns and personality preferences. Based on each microbial sample collection point, different behavioral impact lag time windows are defined. For each microbial sample, specific behavioral characteristics of the patient within the behavioral impact lag time window are extracted. A corresponding lag behavior feature vector is constructed for each microbial sample, and a lag behavior feature matrix is ​​established. The time windows for the effects of different behaviors include: 0-6 hours as the immediate effect, applicable to oral hygiene behaviors; 6-24 hours as the short-term effect, applicable to a single eating event; 24-72 hours as the medium-term effect, applicable to eating patterns and sleep states; and more than 72 hours as the long-term effect, applicable to lifestyle habits.

[0024] Step S2 also includes: Based on the abundance of microbial species, metabolic pathways, and species diversity index of the samples, normalization was performed to establish a microbial characteristic matrix. Using the lagged behavioral feature matrix as the predictor variable and the microbial feature matrix as the response variable, the optimal regularization parameter was selected through 10-fold cross-validation. Behavioral variables related to microbial features were screened. Bootstrap resampling was used to calculate the 95% confidence interval of the regression coefficients. Stable associations with confidence intervals not containing zero were retained. Coupled with the personalized dynamic map of the patient's oral microbiome with time characteristics, a patient behavior-microbial association model was established to obtain standardized regression coefficients and quantitatively assess the direction and magnitude of the influence of each behavioral feature on specific microbial features. Using lagging behavioral characteristics and microbial characteristics as nodes and standardized regression coefficients as edges, a patient behavior-microbial regulation network diagram is established. The sum of the absolute values ​​of the outgoing edge coefficients of each patient behavior node is calculated, and the behavior is defined as the microbial regulation potential index to obtain the regulation path of external behavior on the microbiome. The specific expression for the microbial regulatory potential index is as follows: In the formula, Patient behavior nodes Sum of the absolute values ​​of the outgoing edge coefficients For behavior nodes To microbial nodes The standardized regression coefficients, For behavior nodes The set of all connected outgoing edges.

[0025] When using it, please refer to the steps outlined above: Existing research on the association between oral microbiome and lifestyle habits largely relies on cross-sectional data or behavioral records at single time points, making it difficult to capture the dynamic changes in behavioral patterns and their temporal impact on the microbiome. Furthermore, the lack of a systematic classification of different behavioral lag effects limits the ability to make causal inferences between behavior and microbes. This step integrates multi-source real-time behavioral data with high temporal resolution microbial sampling to construct a behavior-microbe association model coupled with temporal features. This enables multidimensional quantitative analysis of behavioral lag effects based on personalized dynamic maps, identifying specific regulatory pathways of behavior on the microbiome at different time scales. The microbial regulatory potential index provides a direct assessment of key behavioral intervention targets, offering dynamic and interpretable predictive model support for personalized oral health management.

[0026] S3. Based on the regulatory pathways of external behavior on the microbiome, patient physical fitness parameters are introduced as compensating factors to establish a comprehensive evaluation function for the efficacy of anti-infective drugs, thereby achieving personalized and effective evaluation of anti-infective drugs for patients. Step S3 specifically includes: Collect patients' pre-treatment immune status indicators and baseline values ​​of liver and kidney function metabolism; Based on the baseline sample of patients before the start of treatment and the key node sample of patients on the 7th day after medication, combined with the oral mucositis assessment scale, the clinical inflammation dimension indicators of patients were calculated. Based on the personalized dynamic map of the patient's oral microbiome with time characteristics, the improvement rate of the patient's oral microbiome health index on day 7 compared with the baseline sample was calculated, and the patient's microbiome improvement dimension index was obtained. The specific expression for the patient's oral microbiome health index on day 7 is as follows: In the formula, The oral microbiome health index of the patient on day 7. This represents the total number of microbial ecological indicators. For the first One microbial ecological indicator, For the patient on the 7th day Normalized values ​​of microbial ecological indicators For the first The weights of each microbial ecological indicator.

[0027] Step S3 also includes: Based on the metabolic pathway abundance table, the increase rate of specific microbial metabolic pathways related to mucosal repair on day 7 was calculated to obtain the patient's metabolic function recovery dimension index. Standardize the indicators based on the patient's clinical inflammation dimension, microbial ecology improvement dimension, and metabolic function recovery dimension; Using principal component analysis, the first principal component with the highest variance explanation rate is extracted as the comprehensive index of the efficacy of anti-infective drugs for patients. Based on the linear expression of this principal component, the comprehensive evaluation function of anti-infective drug efficacy is obtained, and a multidimensional efficacy vector for different medication regimens is generated to realize the personalized and effective evaluation of anti-infective drugs for patients. The specific expression for the comprehensive index of patient anti-infective drug efficacy is as follows: In the formula, For the patient's clinical inflammation dimension indicators, As an indicator of microbial ecology improvement, As an indicator of metabolic function recovery, , , The scores are standardized clinical inflammation score, microbial ecology improvement score, and metabolic function recovery score. , , The mean values ​​are the clinical inflammatory markers, the mean values ​​of the indicators for improvement in the microbial ecology dimension, and the mean values ​​of the indicators for recovery of metabolic function. , , The standard deviations are the standard deviations of clinical inflammation indicators, the standard deviations of indicators related to improvement in the microbial ecology dimension, and the standard deviations of indicators related to recovery of metabolic function. , , , Principal component coefficients, As a compensating factor for physical fitness, This is a comprehensive index of the efficacy of anti-infective drugs for patients.

[0028] Reference Figure 2 As shown, an efficacy evaluation system for anti-infective drugs includes: Data acquisition module, adjustment path module, and personalized effective evaluation module; Among them, the adjustment path module is electrically connected to the data acquisition module, and the personalized effective evaluation module is electrically connected to the adjustment path module; The data acquisition module collects oral microbial samples from patients throughout the entire treatment cycle, generates personalized dynamic maps of the patient's oral microbiome with time characteristics, and captures the composition and dynamic evolution of the oral microbial community. The regulation pathway module integrates multidimensional lifestyle data of patients, extracts patients' lifestyle behavior patterns and personality preference characteristics, and performs coupled analysis with a personalized dynamic map of the patient's oral microbiome with time characteristics to establish a patient behavior-microbiome association model and obtain the regulation pathway of external behavior on the microbiome. The personalized effective evaluation module, based on the obtained regulatory pathway of external behavior on the microbiome, introduces the patient's physical fitness parameters as a compensating factor, establishes a comprehensive evaluation function for the efficacy of anti-infective drugs, and realizes personalized effective evaluation of patients for anti-infective drugs.

[0029] When using it, please refer to the steps outlined above: At the background technology level, existing anti-infective drug efficacy evaluation systems mostly rely on single clinical endpoints such as pathogen clearance rate and symptom relief time, lacking a systematic integration of dynamic changes in oral microbiome ecology, host immune metabolic function, and individual physiological basis during patient treatment. This results in one-sided evaluation dimensions and insufficient consideration of individual differences, making it difficult to achieve truly personalized efficacy prediction and optimization. This step integrates patient physical fitness parameters as compensating factors to construct a multi-level comprehensive evaluation function that integrates clinical inflammation dimension, microbiome ecology improvement dimension, and metabolic function recovery dimension, realizing a comprehensive, personalized, and dynamic assessment of the efficacy of anti-infective drugs for patients. It not only quantifies the functional recovery and ecological health improvement of the microbiome, but also extracts core efficacy indices through principal component analysis to generate multi-dimensional efficacy vectors for different medication regimens, providing interpretable and operable quantitative evidence for clinical precision medication decisions and improving the individualization and effectiveness of treatment strategies.

[0030] Based on the above, the specific implementation method is as follows: Taking a cancer patient undergoing chemotherapy as an example, the oral microbiome was dynamically monitored during the anti-infective treatment. Non-irritating whole saliva samples were collected using a standardized saliva collector at five fixed time points: before the start of treatment, on days 1, 3, and 7 after taking the medication, and on day 14 after the end of treatment. On the 5th day of treatment, the patient reported oral mucosal pain, and an additional sample was collected within 24 hours. All samples underwent standardized DNA extraction, quality control, and host gene knockout before shotgun metagenomic sequencing. By analyzing the sequencing data, Kraken2 was used to identify the main genera of Streptococcus and Veillonella in the patient's oral cavity and their abundance over time. Using the HUMAnN3 process, the abundance of the L-arginine biosynthesis II metabolic pathway, which is related to mucosal barrier repair, was quantified. Topological data analysis revealed that species abundance fluctuated, but a core module composed of specific symbiotic streptococci remained stable in multiple time windows and was identified as the patient's personalized core microbiome. By using a change point detection algorithm, statistically significant mutations in the microbial community diversity index and pathogenic bacteria abundance were accurately located on the 3rd day of medication and after the occurrence of mucosal pain events. By integrating species, function, diversity and clinical event data from all time points, a personalized dynamic map of the patient's entire treatment cycle was generated, revealing the continuous and dynamic evolution trajectory of oral microbial community structure and functional pathways under the impact of drug intervention and clinical events.

[0031] During microbial sampling, the patient's daily diet was collected simultaneously through the patient diary APP, recording data on spicy food consumption for lunch on the second day, oral hygiene, mouthwash use, and medication adherence; daily sleep duration and quality, and step count data were obtained through smart bracelets; the duration and coverage of each brushing session were recorded through smart toothbrushes; and the multi-source behavioral data were unified with timestamps and structured. Researchers defined the lag time windows of the effects of different behaviors on microbes, classifying brushing behavior as an immediate effect, a single high-sugar diet as a short-term effect, and sleep patterns over several consecutive days as a medium-term effect. For each microbial sample, behavioral characteristics within the corresponding time window before the collection point are extracted, and a lagged behavioral feature vector is constructed. When analyzing the microbial sample on day 3, the dietary patterns of the previous 24-72 hours and oral cleaning behaviors of the previous 0-6 hours are correlated. Using the lagged behavioral characteristics at all time points as predictors and the corresponding microbial characteristics as response variables, an elastic network regression model was constructed. The analysis revealed that the decrease in sleep efficiency within the mid-term effect window was significantly positively correlated with the increase in Porphyromonas gingivalis abundance, with a standardized regression coefficient of 0.32. Effective brushing behavior within the immediate effect window was significantly correlated with the increase in nitrate reduction pathway abundance. These stable associations were visualized to construct a personalized behavior-microbe regulation network diagram for the patient. The microbial regulation potential index of the patient's sleep quality in this network was calculated to be the highest, identifying that improving sleep is a key potential intervention path for regulating the oral microecology and promoting health functions. Pre-treatment physical condition compensatory factors, including basic immune function and liver function indicators, were collected. Efficacy indicators were calculated from three dimensions: clinical, microbial ecology, and functional. Clinical inflammation dimension: Compared to baseline, on day 7, the oral mucositis grade decreased from level 2 to level 1 according to the oral mucositis assessment scale. Microbial ecology improvement dimension: The oral microbial ecology health index on day 7 improved by 25% compared to baseline. Metabolic function recovery dimension: The abundance of the L-arginine biosynthesis II pathway, related to mucosal repair, increased by 40% on day 7. These standardized indicators, along with physical fitness compensation factors, were input into the principal component analysis model; the first principal component extracted by the model explained 85% of the variance, constituting the comprehensive index of the patient's anti-infective drug efficacy. Based on the linear expression of the principal component, a personalized comprehensive efficacy evaluation function was generated. The calculation results showed that the patient's comprehensive efficacy index was 0.76, indicating that the current medication regimen was generally effective. Analysis of the multidimensional efficacy vector revealed that the patient scored particularly well in the dimension of metabolic function recovery, but the improvement in the dimension of microbial community stability was relatively limited, which is consistent with the impact of sleep disturbance behavior experienced during treatment. This evaluation result provides clinicians with precise insights. The current medication regimen is effective in inhibiting pathogens and restoring metabolic function, but to further improve the overall efficacy and prevent relapse, it is necessary to combine behavioral interventions to enhance the stability of the microbial ecosystem.

[0032] To further verify the rationality and feasibility of the technical solution of this invention, and to demonstrate a clear physiological link between changes in the oral microbiome and the efficacy of anti-infective drugs, thereby illustrating that evaluating the effectiveness of anti-infective drugs based on dynamic changes in the oral microbiome has a scientific basis: Existing research indicates that the oral cavity is not only a gateway for the colonization and invasion of many pathogenic microorganisms, but also a window into the systemic immune status. During anti-infective treatment, changes in the composition of the oral microbiota often precede the appearance of clinical symptoms. Publicly available literature (related research in Microbiome, 2021) indicates that oral microbiota dysbiosis induced by chemotherapy or antibiotic treatment is significantly positively correlated with the level of systemic inflammation and the incidence of bacteremia. Therefore, monitoring the dynamic changes in the oral microbiota can provide sensitive biomarkers for early assessment of the efficacy of systemic anti-infective therapy. Multiple clinical studies have confirmed that specific commensal bacteria (Streptococcus, Veillonella) and their metabolic pathways (L-arginine biosynthesis, nitrate reduction pathway) in the oral cavity are closely related to the host mucosal barrier function and local immune response. When anti-infective drugs are effective, the abundance of pathogens decreases, while the abundance of beneficial bacteria and their restorative metabolic pathways increases. This functional recovery is an important dimension for evaluating efficacy. Based on this mechanism, this invention quantifies functional recovery rather than merely focusing on pathogen clearance through metagenomic sequencing and metabolic pathway analysis. Recent digital health research (a review in Nature Digital Medicine in 2023) has shown that integrating wearable device behavioral data with microbiome time-series data can construct an exposure-response model that can effectively predict an individual's response trajectory to drugs. This invention, by introducing lag time window analysis and coupling it with personalized behavior, is in line with the mainstream technical trend of multimodal dynamic monitoring in current precision medicine. In summary, although relying solely on oral microbial indicators is insufficient to completely replace clinical endpoint indicators, combining behavioral data, physical fitness compensating factors, and multi-dimensional microbial function analysis has constituted a comprehensive evaluation system with clinical interpretability and predictive power. This invention solves the technical problems of static, single-dimensional, and uncorrelated clinical events in traditional methods through the above-mentioned technical means, and achieves a technical leap from microbial structure changes to efficacy and function evaluation.

[0033] This invention does not rely solely on changes in the structure of the oral microbiome as the sole basis for efficacy evaluation. Instead, it constructs a comprehensive efficacy evaluation model integrating a dynamic microbiome map, behavioral regulation pathways, and physical fitness compensation. By introducing behavioral lag effect analysis and metabolic function recovery dimensions, it overcomes the uncertainty of evaluation based on a single biological indicator. Existing medical literature (a related review in Cell Host & Microbe, 2022) has clearly pointed out that the oral microbiome plays a dual role in early warning and functional indication during anti-infective treatment, and its change patterns are highly coupled with clinical efficacy in time. This invention transforms this scientific discovery into an operable technical solution through multi-source data fusion and personalized modeling, significantly improving the accuracy and clinical applicability of efficacy evaluation.

[0034] To further verify the rationality and feasibility of the technical solution of this invention, and to prove that the evaluation system of this invention does not rely solely on changes in the oral microbiome, but rather achieves a comprehensive evaluation of the effectiveness of anti-infective drugs through the fusion of multi-dimensional indicators: Existing research indicates that the oral cavity is not only a gateway for the colonization and invasion of many pathogenic microorganisms, but also a window into the systemic immune status. During anti-infective treatment, changes in the composition of the oral microbiome often precede the appearance of clinical symptoms, and single-dimensional microbial indicators are insufficient to fully and accurately reflect the overall therapeutic effect. Based on the dynamic oral microbiome mapping in step S1, this invention further introduces a temporal coupling analysis between patient lifestyle patterns and the microbiome in step S2, quantifying the regulatory pathways of external behavior on the microbiome, thus elevating therapeutic effect evaluation from simple biomarker monitoring to an exposure-response causal inference level. In step S3, the patient's physical fitness parameters are introduced as compensation factors, and a multi-level comprehensive evaluation function is constructed by integrating the dimensions of clinical inflammation, microbial ecology improvement, and metabolic function recovery.

[0035] In summary, this invention constructs a comprehensive therapeutic effect evaluation model integrating microbial dynamic map, behavioral regulation pathway, and physical fitness compensation, overcoming the uncertainty of single biological indicator evaluation and realizing a technological leap from microbial structural changes to therapeutic function evaluation. Those skilled in the art can fully implement this invention based on the description in the specification.

[0036] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention. The scope of protection claimed by the appended claims and their equivalents is defined.

Claims

1. A method for evaluating the effectiveness of an anti-infective drug, characterized in that, include: S1. Collect oral microbial samples from patients throughout the entire treatment cycle, generate personalized dynamic maps of the patient's oral microbiome with time characteristics, and capture the composition structure and dynamic evolution of the oral microbial community. S2. Integrate multidimensional lifestyle data of patients, extract patients’ lifestyle behavior patterns and personality preference characteristics, and couple them with the personalized dynamic map of patients’ oral microbiome with time characteristics to establish a patient behavior-microbiome association model and obtain the regulatory path of external behavior on the microbiome. S3. Based on the regulatory pathways of external behavior on the microbiome, patient physical fitness parameters are introduced as compensating factors to establish a comprehensive evaluation function for the efficacy of anti-infective drugs, thereby achieving personalized and effective evaluation of anti-infective drugs for patients.

2. The method for evaluating the effectiveness of an anti-infective drug according to claim 1, characterized in that, Step S1 specifically includes: Using a standardized, non-irritating saliva collector, non-irritating whole saliva from patients is collected. The sample taken before the start of treatment is used as the baseline sample for initial oral microbial sample collection. Samples are collected at fixed time points on days 1, 3, and 7 after the patient starts medication. Samples are collected again at the predetermined follow-up time point after the patient finishes treatment. If a specific clinical event occurs, additional samples are collected within 24-48 hours after the event, so as to achieve dynamic monitoring of oral microbial samples throughout the entire treatment cycle of the patient. The specific clinical events include: patients reporting significant changes in oral symptoms, clinicians observing changes in the oral mucosa, completion of chemotherapy drug infusion, initiation of prophylactic antibiotics, and significant changes in patients' lifestyle habits.

3. The method for evaluating the effectiveness of an anti-infective drug according to claim 2, characterized in that, Step S1 also includes: Total DNA was extracted from the collected samples using a DNA extraction kit validated for low biomass oral samples. The concentration and integrity of the DNA were assessed using a Qubit fluorescence quantitative quantification system and an Agilent Bioanalyzer, respectively. Qualified DNA samples were screened and libraries were constructed. For qualified DNA libraries, shotgun metagenomic sequencing was performed on the Illumina NovaSeq platform, aiming to produce 10-20 million raw paired-end reads per sample. FastP was used to perform quality control on the raw data to remove low-quality bases, adapters and excessively short reads. Bowtie2 was used to align the quality-controlled reads to the human reference genome, remove host-derived reads, and retain non-host-derived microbial reads for data preprocessing. The Kraken2 method was used to classify and quantify the abundance of microbial species, enabling rapid identification of microbial species present in patient samples. Based on the HUMAnN3 process, read lengths are aligned to the UniRef90 integrated gene database to quantify the abundance of microbial gene families. These are then mapped to the MetaCyc metabolic pathway database to calculate pathway coverage and abundance, generate a metabolic pathway abundance table, and quantitatively assess the functional potential of microbial communities.

4. The method for evaluating the effectiveness of an anti-infective drug according to claim 3, characterized in that, Step S1 also includes: Based on the metabolic pathway abundance table, the abundance of microbial species at each time point is regarded as a point in a high-dimensional species space. A time-series point cloud is constructed during the patient's treatment. The persistent homology algorithm is applied to identify the topological stability features that persist in the point cloud at different time scales. The features that persist in multiple time windows are used as the personalized core microbial components of the patient, thus obtaining the personalized oral microbial characteristics of the patient during the treatment. Based on the abundance of microbial species at each time point of the collected samples, the data at each time point was smoothed using cubic spline interpolation. The smoothing parameters were automatically selected through generalized cross-validation to complete the smoothing of the time series data of the samples. Using a change point detection algorithm, a fixed time window length is set, and the statistical characteristics of microbial species at each time point of the collected samples within the time window are calculated. The points where the statistical characteristics of microbial species change are automatically identified as the points where the microbial community status changes. Using time points, sample IDs, species abundance, metabolic pathway abundance, species diversity index, and specific clinical events as multi-dimensional data, a personalized dynamic map of the patient's oral microbiome with time characteristics is generated to capture the composition structure and dynamic evolution of the patient's oral microbiome.

5. The method for evaluating the effectiveness of an anti-infective drug according to claim 1, characterized in that, Step S2 specifically includes: Based on the patient diary APP, we can obtain patients' daily dietary records, oral hygiene practices, and records of the time and dosage of anti-infective drugs. Based on wearable devices for patients, the system monitors patients' sleep status in real time and automatically synchronizes patients' daily step count and duration of moderate to high intensity exercise. Based on the smart toothbrush backend data, data on the duration, frequency and coverage area of ​​each brushing session are collected from the patient. By integrating patient diary app, patient wearable device and smart toothbrush backend data, multidimensional patient lifestyle data is obtained, unified to the same timestamp, and patient discrete behavioral events are transformed into frequency indicators, and patient data is processed in a structured manner. Based on the patient's daily dietary records, the types and quantities of food consumed by the patient were obtained, and a frequency matrix of the patient's daily food categories was established.

6. The method for evaluating the effectiveness of an anti-infective drug according to claim 5, characterized in that, Step S2 also includes: Using the patient's daily food category frequency matrix as input, the latent Dirichlet Allocation Algorithm is used to randomly select any patient's daily food category distribution to minimize perplexity, determine the number of topics, identify 3-5 potential dietary topics and their probability distributions, and obtain the patient's dietary pattern characteristics. For wearable devices used by patients, the mean, variance, and regularity of patients' sleep states over 24 hours were calculated, and statistical characteristics of patients' sleep states were extracted. Calculate the patient's weekly exercise duration, frequency, and regularity, and extract statistical characteristics of the patient's exercise data; Calculate the brushing time, frequency, and uniformity of coverage area for patients, and extract statistical characteristics of oral hygiene. By integrating statistical characteristics of patients' sleep status, sleep status, and oral hygiene, we can obtain patients' lifestyle behavior patterns and personality preferences. Based on each microbial sample collection point, different behavioral impact lag time windows are defined. For each microbial sample, specific behavioral characteristics of the patient within the behavioral impact lag time window are extracted. A corresponding lag behavior feature vector is constructed for each microbial sample, and a lag behavior feature matrix is ​​established. The time windows for the effects of different behaviors include: 0-6 hours as the immediate effect, applicable to oral hygiene behaviors; 6-24 hours as the short-term effect, applicable to a single eating event; 24-72 hours as the medium-term effect, applicable to eating patterns and sleep states; and more than 72 hours as the long-term effect, applicable to lifestyle habits.

7. The method for evaluating the effectiveness of an anti-infective drug according to claim 6, characterized in that, Step S2 also includes: Based on the abundance of microbial species, metabolic pathways, and species diversity index of the samples, normalization was performed to establish a microbial characteristic matrix. Using the lagged behavioral feature matrix as the predictor variable and the microbial feature matrix as the response variable, the optimal regularization parameter was selected through 10-fold cross-validation. Behavioral variables related to microbial features were screened. Bootstrap resampling was used to calculate the 95% confidence interval of the regression coefficients. Stable associations with confidence intervals not containing zero were retained. Coupled with the personalized dynamic map of the patient's oral microbiome with time characteristics, a patient behavior-microbial association model was established to obtain standardized regression coefficients and quantitatively assess the direction and magnitude of the influence of each behavioral feature on specific microbial features. Using lagging behavioral characteristics and microbial characteristics as nodes and standardized regression coefficients as edges, a patient behavior-microbial regulation network diagram is established. The sum of the absolute values ​​of the outgoing edge coefficients of each patient behavior node is calculated, and the microbial regulation potential index of the behavior is defined to obtain the regulation path of external behavior on the microbiome.

8. The method for evaluating the effectiveness of an anti-infective drug according to claim 7, characterized in that, Step S3 specifically includes: Collect patients' pre-treatment immune status indicators and baseline values ​​of liver and kidney function metabolism; Based on the baseline sample of patients before the start of treatment and the key node sample of patients on the 7th day after medication, combined with the oral mucositis assessment scale, the clinical inflammation dimension indicators of patients were calculated. Based on the personalized dynamic map of the patient's oral microbiome with time characteristics, the improvement rate of the patient's oral microbial ecological health index on day 7 compared with the baseline sample was calculated, and the patient's microbial ecological improvement dimension index was obtained.

9. The method for evaluating the effectiveness of an anti-infective drug according to claim 8, characterized in that, Step S3 also includes: Based on the metabolic pathway abundance table, the increase rate of specific microbial metabolic pathways related to mucosal repair on day 7 was calculated to obtain the patient's metabolic function recovery dimension index. Standardize the indicators based on the patient's clinical inflammation dimension, microbial ecology improvement dimension, and metabolic function recovery dimension; Using principal component analysis, the first principal component with the highest variance explanation rate is extracted as a comprehensive index of the efficacy of anti-infective drugs for patients. Based on the linear expression of this principal component, a comprehensive evaluation function of the efficacy of anti-infective drugs is obtained, generating a multidimensional efficacy vector for different medication regimens, thereby realizing personalized and effective evaluation of anti-infective drugs for patients.

10. An efficacy evaluation system for anti-infective drugs, characterized in that, A method for evaluating the effectiveness of an anti-infective drug as described in claims 1-9 includes: Data acquisition module, adjustment path module, and personalized effective evaluation module; Among them, the adjustment path module is electrically connected to the data acquisition module, and the personalized effective evaluation module is electrically connected to the adjustment path module; The data acquisition module collects oral microbial samples from patients throughout the entire treatment cycle, generates personalized dynamic maps of the patient's oral microbiome with time characteristics, and captures the composition and dynamic evolution of the oral microbial community. The regulation pathway module integrates multidimensional lifestyle data of patients, extracts patients' lifestyle behavior patterns and personality preference characteristics, and performs coupled analysis with a personalized dynamic map of the patient's oral microbiome with time characteristics to establish a patient behavior-microbiome association model and obtain the regulation pathway of external behavior on the microbiome. The personalized effective evaluation module, based on the obtained regulatory pathway of external behavior on the microbiome, introduces the patient's physical fitness parameters as a compensating factor, establishes a comprehensive evaluation function for the efficacy of anti-infective drugs, and realizes personalized effective evaluation of patients for anti-infective drugs.

Images