A method and device for assisting individualized administration of antibacterial drugs to infected patients

Abstract

The application discloses an infection patient antibacterial drug individualized administration auxiliary method and device, through comprehensive pathogen detection result and dynamic comparison of inflammation index, false negative caused by administration exposure, sample processing and other factors in detection link is recognized, and according to this, potential pathogen coverage gap and drug resistance induction risk are investigated;Combined with body weight and liver and kidney function index, the masking abnormality caused by factors such as body fluid retention and decreased creatinine generation capacity is used to grade correct the initial administration dose, and form individualized target dose;Further, through the continuous tracking of blood drug concentration peak and valley fluctuation trend and administration execution time point, the real existing drug accumulation and dose deficiency phenomenon are distinguished, and the administration scheme is dynamically adjusted combined with feedback information such as disease conversion opportunity in treatment process, and finally the individualized administration scheme considering pathogen coverage, dose safety and recurrence risk control is output, which provides support for the precision diagnosis and treatment of infectious diseases.

Description

Technical Field

[0001] This invention relates to the field of medical information processing technology, and in particular to a method and device for individualized administration of antibacterial drugs to infected patients. Background Technology

[0002] The efficacy and safety of antimicrobial therapy are highly dependent on whether the dosing regimen matches the patient's true infection status and metabolic capacity. Clinically, pathogen detection results may be false negatives due to factors such as the timing of sample collection and previous drug exposure. If treatment decisions are based solely on a single test result, the need for pathogen coverage may be underestimated or the potential risk of drug resistance induction may be overlooked. At the same time, routine renal function assessment indicators are easily affected by individual differences such as fluid retention and low muscle mass. The dosage calculated based on routine test values ​​often deviates from the patient's actual clearance capacity, leading to the dual risks of cumulative toxicity or insufficient therapeutic concentration.

[0003] Existing methods for monitoring blood drug concentrations are mostly static models involving single sampling and post-event verification. They are difficult to capture the slow-developing accumulation trend during drug administration and to identify the true source of abnormalities from the time fluctuations in drug administration. In addition, once the dosing regimen is determined, it often lacks a dynamic feedback mechanism based on the treatment response, and it is even more difficult to incorporate potential relapse signals, such as disease relief that occurs earlier than expected, into the basis for adjusting the regimen. Summary of the Invention

[0004] This invention discloses a method and device for individualized administration of antibacterial drugs to infected patients. It aims to integrate multidimensional information such as infection characteristic data, liver and kidney function indicators, and body weight parameters to identify and correct interfering factors in the pathogen detection and drug dosage calculation processes, and to investigate the risk of drug resistance induction. Simultaneously, it dynamically monitors and adjusts blood drug concentration changes and disease progression during treatment, ultimately forming an individualized dosing regimen that balances pathogen coverage, dosage safety, and relapse risk control, providing practical technical support for rational individualized drug use in clinical practice.

[0005] The first aspect of this invention provides a method for individualized administration of antimicrobial drugs to infected patients, comprising: Acquire infection characteristic data, liver and kidney function indicators, and weight parameters; based on the false negative results of the infection characteristic data, identify interfering factors and form antibacterial interference records. Based on the antibacterial interference record, a baseline rule is determined by matching the preset medication rule. The baseline rule is used to screen for missed pathogens and obtain pathogen gap markers. Drug resistance induction correlation analysis is performed on the pathogen gap markers and the overlapping medication clauses in the baseline rule to obtain drug resistance induction records. The initial dosing dose is obtained by matching the baseline rule with the body weight parameter. Compensatory manifestations at the lower limit of normal are identified from the liver and kidney function indicators to form a reduction factor. The target dosing dose is determined by calculating the initial dosing dose using the reduction factor. Obtain blood drug concentration data, compare the blood drug concentration data with the expected concentration range calculated based on the target dosage to establish an execution deviation record, identify the peak-to-valley fluctuation narrowing feature based on the execution deviation record to obtain an accumulation tendency marker, and eliminate interference based on the accumulation tendency marker to generate a dosage insufficiency record; The initial adaptation plan is established by frequency compensation using the insufficient dosage record and the drug resistance induction record. Based on the initial adaptation plan, negative conversion cases earlier than the preset assessment time point are identified to generate relapse risk points. The relapse risk points are used to adjust the initial adaptation plan and output an individualized dosing plan.

[0006] A second aspect of the present invention provides a device for personalized administration of antimicrobial drugs to infected patients, comprising: An interference identification unit is used to acquire infection characteristic data, liver and kidney function indicators and weight parameters, and to identify interference factors based on the false negative results of the infection characteristic data to form an antibacterial interference record. The induction analysis unit is used to determine the baseline rule based on the antibacterial interference record matching the preset medication rule, use the baseline rule to screen for missed pathogens to obtain pathogen gap markers, and perform drug resistance induction correlation analysis on the pathogen gap markers and the overlapping medication clauses in the baseline rule to obtain drug resistance induction records. The dosage determination unit is used to obtain an initial drug dose by matching the baseline rule with the body weight parameter, identify compensatory manifestations at the lower limit of normal from the liver and kidney function indicators to form a reduction coefficient, and use the reduction coefficient to calculate the target drug dose based on the initial drug dose. The deviation identification unit is used to acquire blood drug concentration data, compare the blood drug concentration data with the expected concentration range calculated based on the target dosage to establish an execution deviation record, identify the peak-valley fluctuation narrowing feature based on the execution deviation record to obtain an accumulation tendency mark, and eliminate interference based on the accumulation tendency mark to generate a dosage insufficiency record. The protocol output unit is used to establish an initial adaptation protocol by using the insufficient dosage record and the drug resistance induction record for frequency compensation, identify negative conversion cases earlier than the preset assessment time point based on the initial adaptation protocol to generate relapse risk points, and use the relapse risk points to adjust the initial adaptation protocol and output an individualized dosing protocol.

[0007] The beneficial effects of this invention are reflected in the following points: First, the joint comparison of infection characteristic data with previous medication records allows for the identification of potential interference in etiological detection results. The identification results are further used to investigate the drug resistance induction association between empirical medication guidelines and potential undetected pathogens. A complete deductive chain is formed from detection interpretation to risk screening, providing a basis for the coverage and adjustment timing of empirical medication regimens, rather than relying solely on a single test result. Second, fluid retention may lead to inflated weight readings, and decreased creatinine production may cause laboratory readings to appear normal. If these two types of masking biases are not identified, they can easily be directly incorporated into the dosage calculation process. This invention identifies and corrects these two types of biases separately in the dosage calculation process, thus establishing dosage calculations based on values ​​that more closely reflect the patient's true fluid state and compensatory level, reducing dosage deviations caused by indicator distortion. Finally, after the dosing regimen is determined, the peak and trough trends of blood drug concentration and the actual dosing time are continuously recorded and periodically reviewed to identify two different types of deviations: drug accumulation and insufficient dosage. Situations such as premature relief of the condition during treatment are also included in the observation and used as feedback information to further adjust the dosing regimen, so that the individualized dosing regimen can be continuously updated along with the treatment process, rather than remaining unchanged in the initially set state. Attached Figure Description

[0008] Figure 1 This is a flowchart of an adjunct method for individualized administration of antibacterial drugs to infected patients according to the present invention.

[0009] Figure 2 This is a time series diagram for determining the narrowing trend and accumulation tendency of peak and valley fluctuations in this invention.

[0010] Figure 3 This is a structural block diagram of an auxiliary device for personalized administration of antibacterial drugs to infected patients according to the present invention.

[0011] Among them: 1-Volatility sequence; 2-Early stable segment; 3-Regression window; 4-Narrowing rate regression line; 5-First hysteresis level trigger point; 6-Second hysteresis level trigger point; 7-Accumulation tendency marker effective interval; 8-Volatility narrowing endpoint; 9-Boundary between no hysteresis and mild hysteresis; 10-Boundary between mild and severe hysteresis; 300-Personalized antimicrobial drug administration aid device for infected patients; 301-Interference identification unit; 302-Induction analysis unit; 303-Dosage verification unit; 304-Bias screening unit; 305-Protocol output unit. Detailed Implementation

[0012] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0013] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0014] The technical solutions of the embodiments of this application are described below.

[0015] like Figure 1 As shown, this embodiment of the invention provides a method for individualized administration of antibacterial drugs to infected patients, including the following steps S101-S105: Step S101: Obtain infection characteristic data, liver and kidney function indicators and weight parameters, and determine interfering factors based on false negative results of infection characteristic data to form antibacterial interference records.

[0016] Specifically, infection characteristic data, liver and kidney function indicators, and weight parameters are obtained. Infection characteristic data includes pathogen detection and inflammation monitoring sub-items: Pathogen detection sub-items are obtained from positive / negative interpretations provided by blood culture or pathogen nucleic acid detection systems. Blood culture is considered positive when colony growth appears within the standard positive reporting window after specimen incubation, and the actual incubation time is recorded. Nucleic acid detection is considered positive when the amplification curve reaches a preset cycle threshold, and the cycle number at the threshold is recorded. The two methods have different result times; blood culture requires the pathogen to proliferate to an identifiable level, typically taking 48 to 72 hours, while nucleic acid detection is completed within 24 hours. Infection characteristic data simultaneously records the detection method corresponding to the current result, and the name of the detected pathogen is also read when positive. Inflammation monitoring sub-items are obtained from body temperature monitoring records and procalcitonin and C-reactive protein readings simultaneously provided by the biochemical testing system. All three items share the same sampling time label. The two biochemical indicators are obtained from separate tests within the same venous blood sample. Liver and kidney function indicators were obtained from concurrent biochemical tests, including serum creatinine, estimated glomerular filtration rate, and alanine aminotransferase (ALT) and aspartate aminotransferase (AST) readings. The first two reflect the renal compensation status, and the latter two reflect the liver compensation status. These four indicators, along with procalcitonin and C-reactive protein (CRP), were collected from the same blood sample. Weight parameters were obtained from the initial weighing upon admission and subsequent daily morning weighings after emptying the bladder. The initial weight reading was marked as the baseline weight. If a measurement could not be completed on the same day, it was recorded at the nearest available time, and the recorded reading was marked with the actual measurement time. Infection characteristic data and liver and kidney function indicators were collected irregularly, while weight parameters were collected daily. All three types of data included their respective sampling times and were simultaneously entered into the same medical record system.

[0017] False negative results based on infection characteristic data identify interfering factors and form antimicrobial interference records. If the etiological testing sub-item in the infection characteristic data reports negative, but in the inflammation monitoring sub-item, at least two of the three parameters—body temperature, procalcitonin, and C-reactive protein—show an increase in three consecutive retests compared to the previous one, with the retest interval corresponding to the routine 72-hour retest cycle, then that etiological testing sub-item is marked as a false negative result. The determination is based solely on whether the negative result deviates from the trend of the inflammatory markers, regardless of the absolute level of any single inflammatory marker reading. Once a false negative result is confirmed, the sampling time of this determination is used as a baseline to trace back to the most recent medication record of the patient. The medication record is taken from the confirmed medication entries in the medical order execution system. The corresponding half-life is found from the preset pharmacokinetic parameter table by drug name. This table also records other pharmacokinetic parameters such as dosing frequency limit and peak time. The interval between the dosing time and the current sampling time is calculated: if the interval is within two half-lives, the drug concentration in the body is not less than one-quarter of the peak concentration. The residual drug enters the culture medium with the sample and inhibits proliferation. It is determined that the pathogen growth is inhibited by sampling after administration. The interfering factor is recorded as drug exposure and associated with the drug name. If the interval exceeds this range or no administration was performed before, the cause of drug inhibition is excluded. The interfering factor is recorded as related to sample processing. If the sampling volume is insufficient, even if bacteria are present, the absolute bacterial count after inoculation will not proliferate to a detectable level. The detection limit is the lowest detectable concentration of the method used. The antimicrobial interference record records the interfering factors, drug names involved, and sampling time corresponding to each false negative determination. The sampling time identifies the determination, and the drug name is only filled in at the time of drug administration exposure. Multiple determinations within the same treatment cycle are included in the antimicrobial interference record, and the entries in the record correspond one-to-one with the detection sub-items of the infection characteristic data.

[0018] Step S102: Based on the antibacterial interference record, match the preset medication rules to determine the baseline rules, use the baseline rules to screen for missed pathogens and obtain pathogen gap markers, and perform drug resistance induction correlation analysis on the overlapping medication clauses in the pathogen gap markers and the baseline rules to obtain drug resistance induction records.

[0019] Specifically, baseline rules are determined by matching pre-set medication rules with antimicrobial interference records. Baseline rules are the pathogen categories and clause sets used in this study. The type of interfering factor carried by the antimicrobial interference record uniquely determines which subset the match falls into. Pre-set medication rules establish a mapping between pathogen categories and corresponding medication clauses. The mapping is divided into two subsets based on this type: the subset corresponding to drug exposure includes all pathogen categories and their clauses appearing in the annual monitoring; the subset corresponding to specimen processing only includes categories and their clauses that account for 80% or more of the cumulative total, i.e., the routine pathogen spectrum. The difference in the width of the two subsets is because residual drugs may inhibit any type of pathogen, and insufficient specimen quantity will only result in missed detections within the routine spectrum. The mapping relationship is compiled based on annual pathogen spectrum monitoring data, and a new version number is generated with each update. When the number of pathogen categories corresponding to the same clause reaches two or more, the clause is marked as an overlapping medication clause during the compilation stage. This marking is independent of each specific match, and once determined, it remains valid indefinitely, not becoming invalid due to not being used in a particular match, and can be called upon in subsequent steps at any time. If multiple antimicrobial interference records accumulate within the current period and their types are inconsistent, they are matched according to the majority of the types. If the number of occurrences is equal, they are matched according to the drug exposure subset. If the types are consistent or there is only one record, they are directly matched. When determining the baseline rule, the mapping version number that takes effect on the day the matching is triggered is fixed and retained with the rule. This step triggers a matching once for each new antimicrobial interference record, and the matching result overwrites the existing baseline rule for the current period. All pathogen categories, clauses, and overlap markers of the corresponding subset are taken, along with the mapping version number used to trigger this matching, and included in the baseline rule.

[0020] Pathogen gap markers were obtained by using baseline rules to identify missed pathogens. All pathogen categories in the baseline rules constituted the comparison benchmark for this screening, and were checked one by one against the names of pathogens that reported positive results in all pathogen testing sub-items for this patient during this treatment cycle. The scope of verification was limited to test records submitted after the diagnosis of this infection. Verification used standardized codes corresponding to the pathogen names for comparison. Codes were grouped by pathogen category, and different specific species within the same category were assigned the same code. Positive results for different specific species within the same category were considered as coverage of that category, because the coverage relationship in medication guidelines is based on pathogen category rather than specific species. Verification was uniformly performed using standardized codes, regardless of the specific testing method. The start and end of this treatment cycle was defined as the period from the diagnosis of this infection to the clinical determination of infection clearance. The specific criteria for determining infection clearance followed existing treatment guidelines. If no positive results were found in any records for a certain pathogen category from the first submission during this treatment cycle until the submission that reported a negative result, this category is determined to be a missed detection category and included in the pathogen gap marker. This determination is anchored to the point in time of the negative submission. Subsequent positive results do not affect this determination. Categories that have not produced any positive results are not carried over to subsequent missed detection counts. The drug administration exposure subset contains more pathogen categories than the specimen processing-related subset. The scope of verification expands as the subset used in the baseline rule expands, while maintaining consistent verification logic. The pathogen gap marker records the name of the missed detection category and carries the version number used in the baseline rule. If the mapping table is subsequently revised, it only indicates a change in the scope defined at that time and does not affect the currently established missed detection conclusion.

[0021] In some embodiments, resistance induction records are obtained by performing a resistance induction association analysis on the pathogen gap markers and overlapping drug use clauses in the baseline rules. This includes: obtaining a gap frequency distribution based on the missed detection frequency count of the pathogen gap markers; classifying overlapping drug use clauses in the baseline rules according to their coverage relationships to form an overlapping drug use list; comparing the gap frequency distribution with the overlapping drug use list to determine induction risk points; and using the induction risk points to perform a resistance induction association analysis to obtain resistance induction records.

[0022] For example, obtaining the gap frequency distribution based on the pathogen gap marker to count the missed detection frequency includes: tracing back late-positive samples based on the pathogen gap marker to obtain the number of missed bacterial species; tracking consecutive negative samples from the number of missed bacterial species to obtain a negative consecutive publication; establishing a consecutive publication statistics table based on the consecutive publication length; and using the consecutive publication statistics table to aggregate the missed detection frequency to obtain the gap frequency distribution.

[0023] The number of missed pathogens is determined by retrospectively analyzing late-positive submissions based on pathogen gap markers. A late-positive result refers to a positive result at a later time. For each missed category in the pathogen gap markers, the complete submission records for the corresponding pathogen detection sub-item within the current treatment cycle are checked. The search is performed by comparing result types in ascending order of submission time until all records for that category within the current treatment cycle have been traversed or a later-time positive result is found. If a later-time positive result exists after the negative result identified as a missed detection, the previous negative result is considered a genuine missed detection. This determination is supported by the actual occurrence of the later positive result; the subsequent detection of the same pathogen contradicts the previous negative result, which is different from direct inference based solely on a single negative result. Submission records are arranged chronologically by sampling time. The later-time positive result, whether from blood culture or nucleic acid testing, can serve as supporting evidence. If multiple pathogens are detected in a single nucleic acid test, the missed detection category corresponding to each detected pathogen is supported by this result. When the same missed detection category corresponds to more than one positive result at a later time point, the earliest occurrence in time is taken as supporting evidence, and the name of the pathogen detected in that result is also recorded. For categories determined to be true missed detections, the specific pathogen name is taken from that later positive result; the same category may have different names in different cycles. When the same category is included multiple times by pathogen gap markers within the same treatment cycle, as long as there is one instance where a later positive result can be found, the number of missed bacterial species is counted as one category. The number of missed bacterial species is counted separately for each treatment cycle, and they are not accumulated across cycles.

[0024] For batches of samples with consecutive negative results among the missed bacterial species counts, a negative consecutive report is obtained. For each pathogen category included in the missed bacterial species count, a separate negative consecutive report entry must be created in this step. The submission records for the corresponding pathogen detection sub-item of that category within the current treatment cycle are taken, and counted chronologically from the negative result determined to be a missed detection until the last negative result before the later positive result appears. All submissions covered within this interval are considered consecutive negative batches for that category. If more than one submission for that category occurs on the same day, the order is determined by the time (minute, second) precision of the submission records. The starting point of the interval is fixed at the submission with the negative result determined to be a missed detection, and the ending point is fixed at the submission immediately preceding the later positive result. The positions of both endpoints are determined by the submission record of the later positive result identified in the previous sub-step, and once the endpoints are determined, they do not move with subsequent submission records. Within a consecutive negative test batch, the testing methods used for each batch are not required to be consistent. If both blood culture and nucleic acid testing batches exist within the same interval, both types of batches are counted together in the same consecutive test interval, without affecting the interval definition. If a certain category of missed bacterial species corresponds to more than one consecutive test interval that meets the criteria, a separate negative consecutive test record is generated for each interval. The records are arranged chronologically according to the start time of the interval, and each record independently carries a complete batch list, not sharing batches with other records under the same category. The negative consecutive test record records the corresponding consecutive negative test batch list, the submission time of each batch, and the testing method used for that batch, according to the pathogen category. These three pieces of information jointly determine the unique attribution of this list within the current treatment cycle.

[0025] A serialization statistics table is established based on the serialization length of negative consecutive reports. For each pathogen category in the negative consecutive reports, a list of consecutive negative batches is created, and the number of batches contained in this list represents the serialization length for that category. Each time a length value is calculated, a corresponding entry is added to the serialization statistics table. The length is calculated based on the number of submitted batches, not calendar days, to avoid miscounting multiple submissions on the same day as a single length unit. Multiple submissions on the same day are each counted as a single batch, and batches with different testing methods are also counted as a single batch. Intervals where blood culture and nucleic acid testing are mixed are not split in the statistics but are counted as a single interval. When the same pathogen gap marker is included multiple times within the same treatment cycle, and a later positive result is found each time, the negative continuation report generates multiple consecutive negative batch lists. The number of batches in each list constitutes a continuation length value for that category. Multiple continuation length values ​​are thus retained under that category, with the number of values ​​equal to the number of times the category is included. If the time intervals of multiple lists overlap, they are merged into one list, and a length value is calculated based on the number of batches in the merged list. Merging is to avoid the same consecutive negative results being counted twice, resulting in duplicate length calculations. Continuation length values ​​generated from different treatment cycles are included in separate continuation statistics tables for each cycle, even if they involve the same pathogen category. A continuation length value of one corresponds to a test record that turned positive after only one negative result; this value is also included under that category, with the lower limit of the value range being one. The continuation statistics table records all continuation length values ​​under each pathogen category, with the number of values ​​equal to the number of lists corresponding to that category in the negative continuation report.

[0026] The frequency distribution of missed detections was obtained by using a serial number statistics table to collect the missed detection frequency. The missed detection frequency characterizes how often the category is missed. This frequency is a dimensionless ratio. The closer the ratio is to 1, the more likely that the category is almost always missed when submitted for testing, and the more likely it is to be a long-term missed detection rather than an occasional missed detection. In the consecutive occurrence statistics table, the consecutive occurrence length values ​​under a certain pathogen category are aggregated into the missed detection frequency using the formula F=ΣL_i / N, where i is the sequential number of each consecutive occurrence under that category, numbered sequentially from 1 to the total number of consecutive occurrences in that category; L_i is the consecutive occurrence length value corresponding to the i-th consecutive occurrence; F is the missed detection frequency for that category; ΣL_i is the sum of multiple consecutive occurrence length values, and the value of ΣL_i is used when there is only one consecutive occurrence length value; N is the total number of pathogen tests submitted for that category during the current treatment cycle; the consecutive occurrence intervals do not overlap, ΣL_i does not exceed N, and F takes a value between 0 and 1, rounded to two decimal places, which is the missed detection frequency for that category in the gap frequency distribution. The F value and the 1 / 3 threshold used for subsequent comparison are both dimensionless ratios between 0 and 1, and the two can be directly compared without conversion. The total number of submissions, N, is taken from all records of the corresponding pathogen detection sub-item within the current period, regardless of whether they are positive or negative. When N is 1, F is equivalent to the proportion of the current consecutive detection length, and it tends to stabilize as N increases. The more submissions, the better this ratio reflects the true frequency of missed detections for that category, rather than random fluctuations. When a category is first included in the pathogen gap marker, its corresponding position is established in the gap frequency distribution. Subsequently, every time the consecutive detection statistics table is updated, the F value corresponding to that position is recalculated. The gap frequency distribution records the corresponding missed detection frequency value F according to the pathogen category, which can be directly used for the next step of comparison with the overlapping drug list.

[0027] Overlapping drug use clauses in the baseline rules are categorized according to their coverage relationships to form an overlapping drug use list. Each drug use clause in the baseline rules records the drug name, route of administration, and the target pathogen category. The coverage relationship refers to the correspondence between the clause and the pathogen category. One clause can correspond to multiple pathogen categories, and multiple clauses can also correspond to the same pathogen category. The route of administration is distinguished by specific methods such as intravenous or oral administration. When the same drug corresponds to more than one route of administration, it is registered as a different clause according to the route. All clauses marked as overlapping in the baseline rules are categorized into two classes based on the number of pathogen categories associated with them in the mapping table. The classification of a clause is determined by its coverage relationship: the more categories associated with the same drug, the wider the range of pathogens covered by its selection pressure. Wider coverage means that when using the drug empirically, it is easier to generate selection pressure on undetected pathogens. Therefore, clauses associated with three or more categories are classified into the broad-spectrum category, and those associated with two categories are classified into the narrow-spectrum category. The three-category threshold is the median number of associated categories in the mapping table for overlapping drug clauses. The lower limit of two or more categories is used to determine whether a clause is overlapping. The latter further distinguishes between broad-spectrum and narrow-spectrum clauses within overlapping clauses. The two have different stages of action. The overlapping drug list is indexed separately for the broad-spectrum and narrow-spectrum categories. The clauses in the index are arranged in order according to their original position in the mapping table. Each clause fully records its drug name, route of administration, and all associated pathogen categories. Fields are not omitted due to different categories or different numbers of associated categories. The number of entries in the overlapping medication list varies depending on the subset used in the baseline rules. Detailed entries are not registered repeatedly, and the same entry is only classified into one category.

[0028] The gap frequency distribution is compared with the overlapping drug list to determine the induced risk points. In the gap frequency distribution, each pathogen category corresponds to a missed detection frequency value between 0 and 1. The overlapping drug list provides pathogen categories associated with each entry in two categories: broad-spectrum and narrow-spectrum. The pathogen category sets covered by the two inputs are usually not completely overlapping. The comparison object in this step is the intersection of the two sets; categories or entry frequencies outside the intersection are not included in the judgment. The comparison starts with whether the value in the gap frequency distribution reaches 1 / 3: categories with a value of at least 1 / 3 are selected as candidates. The 1 / 3 ratio is the maximum allowable missed detection tolerance within a preset period, calibrated by the laboratory based on the historical test data of the past 3 years, reviewed annually, and updated according to the review results. Values ​​falling on the boundary are also treated as candidates. The number of candidate categories is usually far less than the total number of pathogen categories in the baseline rules. A missed detection frequency reaching 1 / 3 is relatively rare in clinical practice; the missed detection frequency of most pathogen categories is significantly lower than this threshold, and only a very small number actually enter the candidate range. Candidate categories are checked one by one against the broad-spectrum category I of the overlapping drug list. Narrow-spectrum category I is not included in this determination. When a candidate category appears in the pathogen category list associated with a certain clause of the broad-spectrum category I, the candidate category and the corresponding clause are recorded as an induced risk point. The induced risk point consists of three parts: the candidate category name, the matched clause identifier, and the missed detection frequency value of the candidate category. When a candidate category appears in the list associated with more than one broad-spectrum clause, a corresponding number of induced risk points are generated according to the number of matched clauses. Each risk point is independent and not merged with others, and its three parts correspond to the clause it is matched with.

[0029] Drug resistance induction records were obtained by performing drug resistance induction association analysis using induction risk points. When the matching clause carried by the induction risk point corresponds to a drug currently being used by the patient, the drug name is based on the drug name associated with the drug administration exposure entry in the antimicrobial interference record. The cumulative number of days of drug use within the current treatment cycle is taken and included in the drug resistance induction record. The cumulative number of days is calculated using the formula D=(T_e-T_s) / 24, where T_s is the time of the first drug administration record within the current cycle, T_e is the current time, both are valued in hours, 24 is the number of hours in a day, and D is the cumulative number of days of drug use, rounded to one decimal place, with the unit of measurement being days, consistent with the cycle unit indicated in the product information leaflet, for easy direct comparison. A higher frequency of missed detections indicates that the candidate category may have accumulated and spread within the monitoring blind zone, thus reducing the permissible exposure time accordingly. The exposure time is calculated by multiplying the period indicated in the drug's package insert by 1 minus F. If the cumulative number of days D exceeds this product, it is marked as high risk; otherwise, it is marked as general risk. For drugs whose package inserts do not indicate the period, the average period for similar drugs published by the pharmaceutical administration department is used instead. When the same drug is discontinued and then restarted within the same period, the cumulative number of days of use is the sum of all usage intervals, calculated segment by segment using the above formula and then added together. The calculation is not reset due to discontinuation. When multiple induced risk points correspond to the same drug, the risk level of each risk point is calculated independently based on the missed detection frequency value of its respective candidate category. Records for both high-risk and general-risk levels may exist simultaneously under the same drug name. Drug resistance induction records include the corresponding drug name, cumulative number of days of use, and risk level for each induced risk point.

[0030] Step S103: The initial dose is obtained by matching the baseline rule with the body weight parameter. Compensatory manifestations at the lower limit of normal are identified from the liver and kidney function indicators to form a reduction coefficient. The initial dose is calculated and the target dose is determined using the reduction coefficient.

[0031] In some embodiments, the initial drug dose is obtained by matching the baseline rules with the weight parameters, including: determining drug administration constraints based on the baseline rules regarding the route of administration and frequency restrictions; identifying falsely high weight readings during the edema period using the weight parameters to obtain volume offset; matching the drug administration constraints with the volume offset to a preset dose table to form a standard dose value; and matching the standard dose value with preset drug administration parameters to obtain the initial drug dose.

[0032] Based on baseline rules, administration routes and frequency limits are used to determine dosing constraints. Essentially, each dosing constraint is a unique medication rule card for each drug. Each entry contains four fixed values: drug name, administration route, frequency limit, and number of daily doses. The absence of any one of these values ​​renders the entry incomplete. The administration routes are already recorded in the baseline rules and are directly read from the corresponding fields in the rules, requiring no further conversion. Frequency limits are retrieved from a pre-defined pharmacokinetic parameter table based on the drug name associated with the rule. This table records both half-life and frequency limits for the same drug name. The frequency limit is expressed as the maximum number of daily doses, measured in seconds. This upper limit is determined by the drug's elimination rate; if administration exceeds elimination, the drug accumulates in the body with each dose. The pre-defined pharmacokinetic parameter table covers all drug names in the commonly used antimicrobial drug catalog, and its scope is consistent with the medication rule database used for annual pathogen spectrum monitoring, both sharing the same drug naming conventions. If a baseline rule contains more than one medication provision, the number of entries in the dosing constraint section is equal to the number of provisions. Each provision corresponds to one entry. The daily dosing frequency is set to the frequency limit of the same entry by default when the entry is created, i.e., starting with the maximum dosage. It can be adjusted downwards or maintained based on actual implementation. Once the baseline rule is redefined, the dosing constraint section is regenerated, overwriting the previously existing entries. The daily dosing frequency is simultaneously set to the corresponding frequency limit of the new entry by default. The dosing constraint section is arranged in the order of the provisions according to their original positions in the baseline rule, facilitating future reference and verification of the source by provision.

[0033] Volumetric deviation is obtained by identifying artificially inflated weight readings during the edema phase using weight parameters. The weight reading at admission is marked as the baseline weight, which is the reading taken upon admission. The volumetric deviation is calculated using the formula O = (W_c - W_b) / W_b, where W_c is the morning weight reading in kilograms, W_b is the patient's baseline weight in kilograms, and O is the volumetric deviation, dimensionless and rounded to two decimal places. A positive value indicates a reading that is artificially inflated relative to the baseline, while a negative value indicates a reading that is lower than the baseline. Negative values ​​are treated as zero for subsequent correction. The denominator is the baseline weight, not the previous day's weight. Daily comparisons easily break down the cumulative slow fluid retention over several days into several seemingly normal small fluctuations. The baseline weight remains constant, reflecting the cumulative change relative to the admission time. When three consecutive weight measurements show an increase, and the cumulative increase exceeds 5% of the baseline weight, the measurement is considered to fall into the edema phase. The 5% threshold is based on historical statistical results of the normal short-term weight fluctuation range for hospitalized patients. For measurements taken during the edema phase, the volumetric offset O is calculated using the above formula and retained. The previous two volumetric offsets, which were temporarily recorded as zero due to not meeting the three consecutive measurement criteria, are retrospectively corrected and recalculated using the same formula. For measurements taken outside the edema phase, the volumetric offset O is set to zero, and this change is considered a true weight change. If this change occurs in the same direction three consecutive times, the baseline weight is reset to prevent true weight gain from being repeatedly misdiagnosed as fluid retention. Volumetric offsets are calculated and retained sequentially on the day of measurement. A single patient may have more than one volumetric offset value.

[0034] Standard dose values ​​are generated by matching a preset dose table with the dosing constraints and volume offset. The dosing constraints provide two indexes—drug name and route of administration—to determine a row in the preset dose table, while the volume offset provides a weight correction basis. Together, they form the foundation for this lookup and conversion. The preset dose table registers the dosage coefficient per kilogram of body weight according to the combination of drug name and route of administration, with the coefficient unit being milligrams per kilogram. The indexing method is consistent with the registration method of the medication instructions. The dosage coefficient is determined based on the median value of the recommended interval in the drug instructions, without using the upper or lower limit of the interval. The preset dose table registers the dosage coefficients for adults and children for the same drug, with the two branches distinguished by age boundaries. The patient's age is taken from the medical record information. In this step, the corresponding branch is selected based on the boundary that the patient's actual age falls into: the coefficient of the adult branch for adult patients and the coefficient of the pediatric branch for pediatric patients. Volume offset is used to correct for weight values: when the volume offset is zero, the current weight reading is used directly in the calculation; when the volume offset is greater than zero, the formula W_a=W_c / (1+O) is used to subtract the falsely high weight, where W_a is the estimated true weight after eliminating fluid retention, in kilograms. Since the measured reading is equal to the baseline weight multiplied by one plus 0, division restores it to the baseline level. The standard dose value is the product of the weight reading after eliminating the falsely high weight and the dose coefficient obtained from the preset dose table, in milligrams, accurate to one decimal place. When there are more than one dosing constraint item, the standard dose value is calculated separately for each item, with each item sharing the same weight value. The dose coefficient varies depending on the drug and route of administration.

[0035] The initial dosing dose is obtained by matching the standard dose value with preset dosing parameters. The standard dose value is a theoretical value continuously calculated based on body weight. The preset dosing parameters register the actual dosage forms and specifications that can be prepared for the drug, the upper and lower limits of the allowable single dose, and the upper limit of the daily cumulative dose. The difference between these two, and the gap between the theoretical continuous value and the actual dosing specifications, is what this step aims to bridge. The bridging method is to round down to the nearest unit based on the specification. When the standard dose value falls within the upper and lower limit range of the single dose specified by the preset dosing parameters, it is rounded down to the nearest unit based on the smallest adjustable unit of the drug dosage form and specification. The rounded value is the initial dosing dose. If the rounding happens to fall exactly at the midpoint between two adjustable units, it is rounded up. When the standard dose value is lower than the lower limit, the initial dosing dose is the lower limit value; when the standard dose value is higher than the upper limit, the initial dosing dose is the upper limit value. The minimum adjustable unit for dosage form is determined by the actual packaging of the drug. Intravenous preparations are typically continuously adjustable in milligrams, while oral preparations use integer multiples of the single tablet dose as the minimum adjustable unit. For continuously adjustable intravenous preparations, rounding is not involved; the standard dose value, after verification against upper and lower limits, is directly used as the initial dose. When a dosing constraint corresponds to more than one standard dose value, the initial dose is matched against preset dosing parameters item by item according to the standard dose value. The number of items is equal to the number of standard dose value items. The drug dosage form and upper / lower limits corresponding to each item are taken from the drug name registered in the preset dosing parameters for that item. The daily cumulative dose, calculated by multiplying the initial dose by the frequency limit, is also subject to upper limit verification. If the limit is exceeded, the single dose is recalculated according to the upper limit, and the recalculated standard dose value is not recalculated.

[0036] In some embodiments, identifying compensatory behaviors at the lower limit of normal from the liver and kidney function indicators to form a reduction factor includes: identifying dilutional normal readings based on the liver and kidney function indicators to form a list of spurious indicators; performing reserve overestimation screening on the list of spurious indicators to obtain a baseline creatinine screening table; determining real risk items based on the baseline creatinine screening table; and performing graded reduction matching based on the real risk items to form a reduction factor.

[0037] This step identifies a list of spurious indicators based on dilutional normal readings using liver and kidney function indicators. Factors such as fluid retention and testing errors can cause a serum creatinine reading to deviate from the patient's recent stable level, appearing normal but actually distorted. This step does not distinguish the specific cause, but only identifies the abnormal patterns in the readings themselves. Serum creatinine readings from the liver and kidney function indicators are arranged chronologically within the current treatment cycle according to sampling time. A reading that shows a sudden drop exceeding a predetermined percentage compared to the arithmetic mean of its three immediately preceding readings is included in the spurious indicator list. Such isolated sudden drops are suspected because they do not conform to the gradual pattern expected of genuine kidney function improvement. True kidney function improvement is usually accompanied by a gradual decline in readings over multiple periods, rather than a single abrupt jump. The percentage decrease is calculated using the formula P = |Cr_n - Cr_avg| / Cr_avg, where Cr_n is the current serum creatinine reading, Cr_avg is the arithmetic mean of the previous three consecutive readings (both in micromoles per liter), and P is the percentage decrease, dimensionless and rounded to two decimal places; a larger percentage indicates a more abrupt drop. Alanine aminotransferase (ALT) and aspartate aminotransferase (AST) are intracellular enzymes with relatively stable readings that do not fluctuate significantly with changes in blood volume; therefore, they are not included in this step. The estimated glomerular filtration rate (GFR) is derived from serum creatinine. Once serum creatinine is included in the list of spurious indicators, the estimated GFR value derived from it is also included at the same time. Both values ​​share the same entry and are not recorded separately. If there are fewer than three historical readings, this step will not determine a sudden drop in that reading until three readings have been accumulated to avoid misjudgment due to insufficient sample size. The list of illusory indicators records the included serum creatinine readings, the corresponding percentage decrease, and the sampling time, and is archived in chronological order of sampling time.

[0038] A baseline creatinine screening table was created by identifying false positives in the list of spurious indicators. Each record in the list of spurious indicators carries the serum creatinine reading, the percentage decrease, and the sampling time. The false positive screening method uses this data to retrospectively check the next serum creatinine reading immediately following the initial drop to determine if the decline continues. True renal function improvement typically occurs multiple times. If the next reading remains at the low level after the drop or decreases further, the reserve is considered normal, meaning this decline reflects genuine improvement and the patient's clearance capacity is indeed improving. If the next reading significantly increases compared to the initial drop, it indicates that the drop was an isolated anomaly, and the patient's ability to produce creatinine has not truly improved, indicating a false positive reserve. The previously identified abnormal reading, even if seemingly normal, may still mask true renal function impairment. The threshold for a rebound is set at an increase exceeding half of the original decrease percentage. A threshold that is too low may include normal small fluctuations as a rebound, while a threshold that is too high may miss genuine isolated anomalies and mistake false improvement for genuine improvement. If no subsequent reading is available for comparison, this record will not be judged until the next serum creatinine reading is available. The judgment result will be used in the same table as the case of excessive reserves. The baseline creatinine screening table records the corresponding serum creatinine reading, the percentage decrease, and the reserve judgment result for each entry in the original sampling time sequence of the false indicator list. The number of records is equal to the number of entries in the false indicator list. The baseline creatinine screening table itself does not cover other readings outside the false indicator list. Readings not included in the false indicator list are treated as true readings and do not participate in the screening judgment in this step. The original readings are directly used in subsequent steps.

[0039] The identification of true risk items is based on comparison with the baseline creatinine screening table. Comparisons are performed between adjacent records. The reserve assessment results in the baseline creatinine screening table, arranged in chronological order of sampling time, are compared one by one with the immediately preceding assessment result. True risk items are generated by consecutive instances of inflated reserves. The minimum number of consecutive triggers is fixed at two, and the sampling times of the two records do not need to be equal. This number is determined based on historical observations of the natural fluctuations in creatinine production capacity over a short period. The observation data is calibrated every two years, and the calibration results are used to update this number. A true risk item consists of two adjacent records in the baseline creatinine screening table that are both assessed as having inflated reserves. Each record carries its corresponding serum creatinine reading, the percentage decrease, and the sampling time. The time interval between the two samplings is also included. If the interval is no more than 72 hours, the two records belong to the same reserve decrease event and constitute one true risk item. If the interval exceeds 72 hours, they are treated as two unrelated single assessments, and no true risk item is generated. The consecutive count restarts from the next record. When three or more consecutive entries in the baseline creatinine screening table are determined to be elevated reserves, two adjacent records form a group, and records can be shared between groups. True risk items are generated in groups of two adjacent entries, with the number of generated entries being one less than the number of records used for the assessment. Adjacent groups share the middle record. True risk items are ordered according to the order of their triggering time, and the number of entries increases with the continuous accumulation of the baseline creatinine screening table. A single patient may generate more than one true risk item within the same treatment cycle.

[0040] A reduction coefficient is generated based on the grading and matching of the actual risk items. The two records associated with each actual risk item, along with their corresponding serum creatinine readings and the calculated reduction percentages, are used as the grading basis. The reduction percentages for both records have already been calculated in the spurious indicator list stage and are reused in this step without recalculation. The average percentage is rounded to two decimal places and is dimensionless. If no actual risk items have been generated during this treatment cycle, the reduction coefficient is set to the default value of 1.0, indicating that current evidence does not support a reduction in the initial dosage. A reduction percentage of less than 15% is classified as mild, with a reduction coefficient of 0.9; a reduction percentage exceeding 15% but not exceeding 35% is classified as moderate, with a reduction coefficient of 0.7; and a reduction percentage exceeding 35% is classified as severe, with a reduction coefficient of 0.5. The three grade limits and corresponding values ​​are based on historical observations of cases with impaired creatinine production. This statistical data is revised every two years, and the revised values ​​are simultaneously applied to the grading and matching in this step. When a patient has more than one real risk factor during the same treatment cycle, the reduction coefficient is determined by the most severe one among all real risk factors, not the average of the grading results. If the alanine aminotransferase (ALT) and aspartate aminotransferase (AST) readings of liver and kidney function indicators obtained from the same blood sample are both higher than the upper limit of the reference range, the reduction coefficient is lowered by one level from the existing level. This is because the overall clearance capacity is lower when both liver and kidney clearance pathways are damaged simultaneously. If the patient is already in the severe level, the reduction coefficient will not be lowered further and will remain at 0.5. The reduction coefficient is recalculated for each new real risk factor, thus the value range is fixed between 0.5 and 1.0, and no values ​​outside this range will appear.

[0041] The initial dose is calculated using a reduction factor to determine the target dose. The initial dose is the rounded value already calculated in the previous sub-step; it is not recalculated in this step. The target dose is calculated using the formula S_t = S_i × K_r, where S_i is the initial dose, K_r is the reduction factor, and S_t, from this step onwards, specifically refers to the final target dose after rounding. The theoretical value S_i × K_r obtained by multiplication is first rounded to the nearest integer according to the dosage form and specification of the drug to which the initial dose belongs. The rounding rule is the same as the rule used when matching the initial dose with preset dosing parameters. Only the rounded value is counted as S_t, with the unit consistent with S_i, which is milligrams. K_r is a dimensionless proportionality coefficient, with a fixed value range between 0.5 and 1.0. The dimension after multiplication remains milligrams, consistent with S_i. When the reduction factor is set to the default value of 1.0, the theoretical value S_i×K_r is numerically equal to S_i. This step recalculates using the above formula instead of directly using the initial dose, ensuring that the two values ​​are updated independently and that neither is omitted from the latest version. The rounded target dose is the final value entered into the medical record system. When entering the target dose, the theoretical value S_i×K_r before rounding is also retained to distinguish the source of deviation when checking blood drug concentrations later. This theoretical value is no longer assigned a separate symbol but is only retained in the medical record as a formula. When the dosing constraint corresponds to more than one initial dose, the target dose is calculated item by item according to the initial dose, and all items share the same reduction factor. This is because the reduction factor reflects the overall renal compensation status of the same patient and is unrelated to the specific antibiotic used. If either the initial dose or the reduction factor changes, the target dose must be recalculated. This is the final output of this step.

[0042] Step S104: Obtain blood drug concentration data, compare the blood drug concentration data with the expected concentration range calculated based on the target dosage to establish an execution deviation record, identify the peak-valley fluctuation narrowing feature based on the execution deviation record to obtain an accumulation tendency mark, and eliminate interference based on the accumulation tendency mark to generate a dosage insufficiency record.

[0043] Specifically, blood drug concentration data is obtained. Blood drug concentration data consists of two types of samples: trough concentration and peak concentration. Trough concentration is obtained from the blood sample immediately preceding the current administration, at which point the drug has been eliminated and reached its lowest level after a dosing interval. Peak concentration is obtained from the blood sample taken after the current administration, based on the time to peak concentration plus the time to peak concentration, at which point absorption and distribution are just completed, but elimination has not yet fully occurred. The time to peak concentration is obtained from a pre-defined pharmacokinetic parameter table by drug name. This table, in addition to half-life and dosing frequency restrictions, also records the time to peak concentration and apparent volume of distribution for each drug. The time to peak concentration is calculated from the end of the administration period as zero. Blood drug concentration readings are in milligrams per liter, rounded to two decimal places. Trough and peak concentrations belong to two separate testing batches, and results for each batch are issued separately; the same blood sample is not used for cross-batch analysis. Trough and peak concentrations are each collected only once within the same dosing cycle. The two blood samples share the same cycle number, which is sequentially numbered from the beginning of the treatment cycle according to the frequency restrictions determined by the dosing constraints. The rate of increase in the number is directly linked to the frequency restrictions. The time difference between blood collection and drug administration is recorded. The time difference corresponding to the trough concentration should be an integer multiple of the dosing interval relative to the first dosing time, and the time difference corresponding to the peak concentration should be close to the time of peak concentration. If the reading exceeds the allowable window, it is still retained but marked with a deviation indicator. This indicator does not affect the reading itself. The allowable window is 10% of the time of peak concentration or the dosing interval. Blood drug concentration data are accumulated and archived in the order of cycle number. The number of records for the same patient in this treatment cycle increases synchronously with the number of dosing cycles. Each dosing cycle contributes one record each for trough concentration and peak concentration, and there are no missing records due to blood collection delays.

[0044] A deviation record is established by comparing the blood drug concentration data with the expected concentration range calculated based on the target dosage. The blood drug concentration data are the measured trough and peak concentrations of the current cycle. Each time the target dosage is redefined, the expected concentration range is recalculated. The expected concentration range is calculated using the formulas C_p=S_t / V_d and C_v=C_p×e^(-k_e×τ), where S_t is the target dosage in milligrams, V_d is the apparent volume of distribution taken from the preset pharmacokinetic parameter table in liters, k_e=ln2 / t_h is the elimination rate constant in hours, t_h is the half-life of the drug, and τ is the dosing interval, both in hours. C_p and C_v are the expected peak and trough concentrations, respectively, both in milligrams per liter. These two values ​​represent the upper and lower limits of the expected concentration range and are used in conjunction with the blood drug concentration data of the current cycle. Peak and trough concentrations from two separate blood samples within the same dosing cycle are grouped together and subtracted from C_p and C_v respectively. The difference is divided by the expected value to obtain the relative deviation. Both are in milligrams per liter (mg / L), and the division is dimensionless. A positive relative deviation indicates higher than expected, and a negative deviation indicates lower than expected, rounded to two decimal places. The execution deviation record records the relative deviations of both peak and trough concentrations for each dosing cycle. The former is used to determine insufficient dosage in this step, while the latter is retained with the record for manual verification of whether the dosing interval has been adequately cleared. The expected peak and trough concentrations are also retained with the execution deviation record; whenever either the blood drug concentration data or the target dosage is updated, a new entry is added to the execution deviation record.

[0045] In some embodiments, identifying peak-valley volatility narrowing characteristics based on the execution deviation record to obtain an accumulation tendency marker includes: calculating peak-valley volatility for a continuous dosing cycle based on the execution deviation record to obtain a volatility sequence; statistically analyzing the volatility narrowing rate from the volatility sequence to determine a narrowing rate value; comparing the narrowing rate value with a preset steady-state standard to form a hysteresis level; and identifying an accumulation tendency based on the hysteresis level to obtain an accumulation tendency marker.

[0046] The peak-to-trough volatility is calculated based on the execution deviation record to obtain a volatility sequence for consecutive dosing cycles. A peak-to-trough blood sample is taken once for each dosing cycle. The relative deviations of peak and trough concentrations for the same dosing cycle in the execution deviation record are already calculated, but this step does not use them directly. Instead, the original blood sample readings from the blood drug concentration data are taken as the measured peak and trough concentrations. The original readings are then checked against the execution deviation record based on the relative deviations. If they do not match, the original readings prevail, and the check results are retained for future reference. The peak-to-trough volatility is calculated using the formula R=(M_p-M_v) / M_p, where M_p is the measured peak concentration, M_v is the measured trough concentration (both in milligrams per liter), and R is the peak-to-trough volatility, dimensionless and measured as a ratio rather than a difference, ranging from 0 to 1. A value close to 1 corresponds to a significant difference between peak and trough concentrations within the dosing cycle, while a value close to 0 corresponds to a trough concentration approaching the peak concentration. Figure 2 This invention visually demonstrates its technical advantage over single-point threshold judgment: While a low single fluctuation is easily affected by blood sampling errors, this invention uses three layers of verification: the downward trend of the entire fluctuation sequence, the narrowing rate after regression window smoothing, and the requirement of two consecutive triggers for it to take effect. Failure to meet any of these criteria will prevent the formation of an accumulation tendency marker, thus distinguishing true accumulation from random fluctuations. Figure 2 As shown, the volatility sequence 1 generally decreases as the dosing cycle progresses, with the initial stage before a significant decline constituting the early stable segment 2. When the measured peak concentration is zero or negative, the cycle is recorded as empty and marked with an abnormality indicator. The volatility sequence is arranged in order of the dosing cycle number, showing the peak and trough volatility of each cycle. For each new deviation record, a new value is added to the volatility sequence, and the sequence length is increased by one.

[0047] The narrowing rate value is determined by statistically analyzing the volatility narrowing rate from the volatility sequence. The volatility sequence uses values ​​corresponding to the most recent five consecutive dosing cycles for this statistical analysis. If the sequence length is less than five cycles, all existing values ​​are used for calculation. In this case, the calculated slope is marked as insufficient for identification. The five-cycle window length is taken from historical experience regarding the observation window length of typical accumulation processes. The narrowing rate value is the slope obtained by linear regression of the most recent five values ​​relative to the dosing cycle number. A negative slope indicates that volatility continues to decrease with the cycle, suggesting a narrowing trend; a positive slope indicates that volatility increases with the cycle, i.e., the peak-to-trough difference shows signs of widening; when the absolute value of the slope is less than 0.02 per cycle, it indicates that volatility remains basically stable, neither narrowing nor widening. This 0.02 per cycle value is the same as the boundary between no lag and mild lag in the subsequent lag level classification, and both are shared. The results are reviewed every two years, and the original value is replaced accordingly. The most recent five dosing cycles used in the statistical analysis constitute... Figure 2The regression window 3 shown contains a linear regression line 4, which reflects the rate of decline in that segment. The independent variable used in the regression is the drug administration cycle number, without normalization. The fitting utilizes all values ​​within the window, rather than just the first and last points, thus mitigating the occasional fluctuations of a single blood sample. The slope is expressed in volatility per drug administration cycle. The regression results also provide the standard error of the slope for future reference. The narrowing rate value is not fixed after a single calculation: for each new value added to the volatility sequence, the most recent five values ​​are used for re-regression, and the regression window shifts one position forward accordingly.

[0048] The hysteresis level is determined by comparing the narrowing rate value with a preset steady-state standard. Hysteresis refers to the slowing of drug clearance in the body and the gradual increase in trough concentration over time. The level is used to characterize the degree of slowing. The narrowing rate value is taken from the slope value obtained from the most recent linear regression calculation. The preset steady-state standard provides three cutoff slopes: a slope value not lower than -0.02 per cycle is classified as no hysteresis; a slope in the range of -0.02 to -0.05 is classified as mild hysteresis (excluding the left endpoint -0.02 and including the right endpoint -0.05); and a slope lower than -0.05 per cycle is classified as severe hysteresis. The three levels are consecutive, non-overlapping, and have no gaps. Figure 2 The cutoffs 9 (no hysteresis) and 10 (mild hysteresis) are compared side-by-side with the narrowing rate regression line 4, using reference lines. The slope falls between these two reference lines, and the relative position indicates the grading result. The specific values ​​of the slopes for the three grading levels are based on historical retrospective statistics of the narrowing process in accumulated cases. This statistical data is reviewed every two years, and the reviewed slope values ​​immediately replace the original values. If the narrowing rate value carries an insufficient sample indicator, or if its slope standard error exceeds the acceptable range of the fit, the resulting hysteresis grade is additionally marked with "Reference Grade," indicating that the grade is based on less than five complete cycles of data and its reliability is lower than usual. Once the narrowing rate value is recalculated, the hysteresis grade is updated accordingly, and the difference between the two gradings is not specifically recorded. The final hysteresis grade is taken from one of the three levels, and only the most recent result is retained; historical changes are not archived separately.

[0049] Accumulation tendency is identified based on lag grade to obtain an accumulation tendency marker. Accumulation tendency indicates a persistent slowing of clearance, distinct from occasional fluctuations in a single cycle. A patient is considered to have accumulation tendency when two consecutive lag grades fall within the range of mild or severe lag, with no interruption of lag between the two lags. An accumulation tendency marker is then generated, and the specific combination of the two grades is recorded in the marker. The determination requires two consecutive lags, not two cumulative lags within a single treatment cycle. A single instance of lag may be an illusion due to a shift in the blood collection time within a particular cycle; two consecutive lags are necessary to demonstrate a genuine and continuous slowing of clearance. If there is an instance of no lag in between, the consecutive count restarts, requiring two more consecutive lags to generate a new marker. If either of the two lags carries the reference grade marker, the accumulation tendency marker will simultaneously carry the reference marker; both lags are not required to carry it. When two consecutive lag grades are severe lag, the accumulation tendency marker additionally carries an expedited marker. This marker is independent of the reference marker and can appear independently. Figure 2 The first hysteresis level trigger point 5 and the second hysteresis level trigger point 6 appear consecutively, together determining the effective interval 7 of the accumulation tendency marker. The end of the interval corresponds to the fluctuation narrowing endpoint 8. Dosing cycles outside the interval are not included in this judgment. The cancellation condition after the accumulation tendency marker is generated is two consecutive periods without hysteresis, which is not met even if it is only met once. The effective period is the complete time range covered by the marker. The accumulation tendency marker records two items according to the time of generation: the combination of the two hysteresis levels and the expedited indicator. When cancellation occurs, the cancellation time is recorded simultaneously. After cancellation, the accumulation tendency marker is not deleted but marked as invalid and retained in its original position.

[0050] In some embodiments, eliminating interference to generate insufficient dose records based on the accumulation tendency marker includes: obtaining a hysteresis distribution table by statistically analyzing the dosing interval compression period based on the accumulation tendency marker; identifying hysteresis interference items within the excessive dosing period based on the hysteresis distribution table; performing interference removal on the hysteresis interference items to obtain the null exposure item; and eliminating residual interference in the null exposure item to generate insufficient dose records.

[0051] A lag distribution table is obtained based on the dosing interval compression period calculated using accumulation tendency markers. Early or excessively frequent dosing can lead to earlier blood sampling times for trough concentrations, causing trough concentration readings to appear higher due to insufficient clearance time, thus seemingly narrowing the peak-trough difference. Therefore, the interval compression during the effective period of the accumulation tendency marker is first investigated. For each dosing cycle involved in the effective period of the accumulation tendency marker, the nominal interval calculated from the frequency limit determined by the dosing constraint is subtracted from the actual dosing interval value for that cycle. When the marker carries an expedited indicator, the statistical range is extended to two dosing cycles before the marker was generated. If the difference is positive, that cycle is included in the compression period. The difference is the compression amount, measured in hours. A shortened interval results in the actual dosage per unit time being higher than the nominal level; the compression period is the aforementioned excessively frequent dosing period. The nominal interval determined by the dosing constraint is calculated based on the maximum daily dosing frequency, measured in hours. The actual dosing interval is taken from the time difference between two consecutive confirmed dosing records in the medical order execution system. Periods where dosing was not performed are not included in the statistics; only periods where both dosing was performed are calculated. The interval span may cover more than one original period number; in this case, the compression amount is uniformly recorded under the period name where the interval ends. The hysteresis distribution table records the corresponding compression amount values ​​according to the dosing period number. Periods within the table are all taken from the period when the accumulation tendency marker is in effect; periods outside the table are not included in this step's statistics. Periods with zero or negative compression amounts are also recorded in this table instead of being omitted, to present the complete trajectory of interval fluctuations, so that when extracting offset interference segments later, it is clear which interval interrupts the continuous interval. Once the effective status of the accumulation tendency marker changes, the hysteresis distribution table is updated synchronously; dosing periods added after the marker is removed are not included.

[0052] For example, identifying hysteresis interference items within a period of excessively frequent drug administration based on the hysteresis distribution table includes: extracting drug administration offset periods based on the hysteresis distribution table to determine offset interference segments; statistically analyzing discrete abnormal drug administration times based on the offset interference segments to establish a time-point exposure list; comparing the time-point exposure list with the offset interference segments to generate suspected interference items; and using the suspected interference items to perform hysteresis interference identification to determine hysteresis interference items.

[0053] The dosing deviation period is extracted based on the hysteresis distribution table to determine the deviation interference segment. The deviation interference segment is an interval of continuous compression. In the hysteresis distribution table, if a dosing cycle with a compression value greater than zero appears three or more times consecutively by cycle number, this continuous interval is extracted as a deviation interference segment. The starting number of three times is distinguished based on two types of cases in the dosing records: random sporadic early dosing and systematic overdosing. This distinction is revised every two years, and the revision results determine whether the starting number needs to be adjusted accordingly. The deviation interference segment records the start and end cycle numbers and the total interval duration. The total interval duration is the sum of the compression values ​​of all cycles within the interval, with a fixed unit of measurement in hours. The start and end numbers are used for subsequent verification of whether the dosing cycle falls within the interval. Verification can be done by directly comparing the cycle numbers, without having to check specific dates one by one, thus improving comparison efficiency. Situations in the hysteresis distribution table where the compression value is greater than zero but less than three times consecutively, or interrupted by cycles with zero or negative compression values, are not extracted as deviation interference segments. Extraction is only established when the consecutive occurrences reach the threshold of three times. Within the same treatment cycle, the hysteresis distribution table can contain more than one consecutive interval that meets the conditions. There is no requirement for a fixed interval length between intervals; only that the conditions are met continuously within each interval. Furthermore, the compression amplitudes of the intervals are not required to be similar; one interval may have a larger compression amplitude than another, and both scenarios are valid independently. The extracted offset interference segments will not be overwritten due to subsequent additions to the hysteresis distribution table. As long as the interval itself still meets the continuous compression condition, its number and range are retained unchanged without reallocation.

[0054] An exposure list is established based on the statistically significant discrepancies in drug administration times within the offset interference segment. Overall, early administration within the interval is considered a systematic offset, while significant deviations at individual moments within the interval have other causes and must be identified separately. First, the difference between the actual drug administration time and its nominal time (the theoretical drug administration time) is calculated for each instance within the offset interference segment. Then, the mean and standard deviation of all differences within the interval are calculated. Drug administration times where the difference deviates from the mean by more than twice the standard deviation are identified as discrepancies. Most deviations fall within this range; those exceeding this range are considered early or late deviations from the norm. The "twice" factor follows the common criteria for judging drug administration time fluctuations, and adjustments are made every two years based on the review results. The nominal time is calculated sequentially by adding an integer multiple of the nominal interval to the start time of the administration within the interval, accurate to the minute. The difference is calculated by subtracting the nominal time from the actual administration time; earlier is negative, and later is positive. The time-point exposure list records three things for each discrete abnormal time point: the offset interference segment number to which it belongs, the actual administration time of that time point, and the corresponding nominal time. These are accumulated in chronological order of occurrence. Each time point belongs to only one offset interference segment. If more than one discrete abnormal time point appears within the same offset interference segment, each abnormal time point becomes a separate entry in the list, with the number of entries equal to the number of abnormal time points in the interval. When all differences within the interval fall within 2 standard deviations, the number of abnormal time points in this statistical analysis for that interval is zero, and the list indicates that the interval has been verified and will not be counted again. The list content is updated accordingly. Once an offset interference segment is re-extracted, the time-point exposure list is re-analyzed, and the old statistical results are no longer retained separately.

[0055] The time-point exposure list is compared with the offset interference segments to generate suspected interference items. For each discrete abnormal time point in the time-point exposure list carrying an offset interference segment number, the total duration of the offset interference segment interval corresponding to that number is checked. If the difference between the actual and nominal dosing times corresponding to the abnormal time point accounts for 20% or more of the total interval duration, and the difference and the total interval duration are both in hours and dimensionless when divided, the abnormal time point and its corresponding offset interference segment are recorded as a single suspected interference item. These types of abnormal time points are commonly seen in clinical settings where the established dosing rhythm is disrupted, such as nursing handovers or bed adjustments. Most dosing within the same period is generally performed according to plan, but one instance is significantly advanced or delayed due to unforeseen circumstances. Once the proportion meets the standard, it is singled out and recorded as a suspected case, but not directly considered a confirmed systemic delay. Abnormal time points with a proportion below 20% are not generated as suspected interference items but are only recorded in the time-point exposure list for comparison reference; the comparison results are not written back to the list. When more than one anomalous time point meets the proportional condition within the same offset interference segment, the same number of suspected interference items are generated based on the number of anomalous time points. Each suspected interference item shares the offset interference segment number but carries its own anomalous time point information and its calculated proportional value. Offset interference segments marked in the time point exposure list as having been checked and for which no anomalous time points were found in this step are also checked once in this step, and the check result is zero suspected interference items. A suspected interference item consists of three parts: the offset interference segment number, the triggered anomalous time point, and the corresponding proportional value. When any item in the time point exposure list or the offset interference segment is updated, the suspected interference items are regenerated, and the number range always corresponds to the number range of the offset interference segment.

[0056] Using suspected interference items, a delayed interference screening process is performed to identify delayed interference items. Screening distinguishes between interference caused by a genuine abnormality in dosing timing and suspected items that are merely random fluctuations. The former is upgraded upon reaching a threshold, while the latter remains suspected. When a suspected interference item carries a proportion of 40% or higher, it is directly identified as a delayed interference item. The proportion is taken from the previous sub-step and is not recalculated. The 40% threshold is twice the 20% baseline proportion. Suspected interference items with proportions between 20% and 40% have slight deviations, making it difficult to determine if a single instance is a coincidence. This situation is common when scheduling, consumable replacements, or other factors continuously affect dosing several times within the same period, with one instance showing a particularly significant deviation, while the others, though minor, are a consequence of the same batch event. Therefore, it is necessary to check whether there are other suspected interference items within the same deviation interference segment that meet the 40% threshold. If so, it indicates a shared systemic delay, and both should be identified as delayed interference items. If not, the suspected status remains unchanged, and the process is not upgraded. When upgrading to a hysteresis interference item, the original proportion value and abnormal time point information are imported as a whole, without recalculating the proportion. A hysteresis interference item consists of three parts: the offset interference segment number, the abnormal time point, and the proportion value. The upgrade path is divided into two categories: direct compliance and shared upgrade. Both categories maintain the same recording format, differing only in the judgment path used in the screening process. Each time a suspected interference item is regenerated, the hysteresis interference item is simultaneously screened. New results replace old results. If a previously confirmed hysteresis interference item still meets the conditions in the new round of suspected interference items, its original number remains unchanged, and the existing numbering order is not disrupted by the re-screening.

[0057] Interference removal is performed on hysteresis interference items to obtain the exposure items to be verified. Removal means first eliminating cycles that can be explained by time point anomalies. Only the remaining cycles need to be verified to determine whether they are indeed accumulating. The offset interference interval range associated with the hysteresis interference item is used to verify each dosing cycle involved during the effective period of the accumulation tendency marker: when the dosing cycle falls within the interval range, the observed peak-to-trough narrowing phenomenon is considered to be fully explained by the dosing time point anomaly, and the cycle is removed from the set of cycles involved in the accumulation tendency marker. The removal record also retains the corresponding hysteresis interference item number, so that each removal has a corresponding judgment that can be checked. The cycles remaining after removal are those dosing cycles that cannot be explained by factors such as nursing shifts or consumable replacements, and whose peak-to-trough difference is indeed narrowing. These are the cycles that really need to be verified to determine whether they are drug accumulation. The remaining periods that do not fall within the associated interval of any hysteresis interference item constitute the exposure items to be verified. The exposure items to be verified are recorded by period number, one record per period, recording the execution deviation record entries corresponding to each remaining period, and uniformly including the current value of the hysteresis level at the time of formation. All remaining periods share the same hysteresis level result. The execution deviation record entries are each taken from the values ​​previously calculated for their corresponding periods. This step only filters the period range and does not recalculate the specific magnitudes of these two values. When all periods are removed, the exposure items to be verified are recorded as zero. If either the hysteresis interference item or the accumulation tendency marker is updated, the exposure items to be verified are re-removed. The old results before the overwrite are not archived separately. The exposure items to be verified and the execution deviation records share the same dosing cycle numbering system.

[0058] Residual interference in the exposure items to be tested is excluded to generate insufficient dose records. Both accumulation and insufficient dose manifest as a narrowing of the peak-to-trough difference, the difference being that the former results in an increased trough concentration, while the latter results in a lower peak concentration. This step is based on the latter for identification. For each cycle in the exposure items to be tested, the execution deviation record entry is checked back for the relative peak concentration deviation value of that cycle. If the relative peak concentration deviation is negative, that is, the measured peak concentration is lower than the expected peak concentration, it is determined that the narrowing of the peak-to-trough difference in that cycle is not due to drug accumulation, but rather to the residual interference factor that the single dose failed to reach the expected peak level. If the cycle meets both the conditions of a negative relative peak concentration deviation and being included in the exposure items to be tested, it is determined to be an insufficient dose cycle. The treatment is opposite depending on the cause: accumulation requires dose reduction, while insufficient dose requires dose increase. Cycles with a peak concentration relative deviation of zero or a positive value are not included in the insufficient dose record, but are retained in the interpretation of drug accumulation and are not marked again. Insufficient dosage records are numbered by period, with one record per period. For each period with insufficient dosage, the record includes four items: the drug name, the relative deviation of peak concentration, the hysteresis grade, and the actual target dosage used in that period. This information, combined, describes the specific extent of the dose deficiency, the accompanying hysteresis, and the actual dosing basis at the time, providing clinical guidance for determining whether a dose increase is necessary. The magnitude of the dose increase is referenced to the absolute value of the relative deviation of peak concentration; a larger absolute value indicates a more significant dose deficiency. If no period meets the criteria in the pending nuclear exposure category, the insufficient dosage record is recorded as zero, indicating that no evidence of insufficient dosage was found. Once the pending nuclear exposure category is removed, the insufficient dosage record is regenerated.

[0059] Step S105: Use the insufficient dosage record and drug resistance induction record to establish an initial adaptation plan through frequency compensation. Based on the initial adaptation plan, identify the negative conversion situation earlier than the preset assessment time point to generate relapse risk points. Use the feedback of the relapse risk points to adjust the initial adaptation plan and output an individualized dosing plan.

[0060] Specifically, frequency compensation is used to establish an initial adaptation plan based on underdosing records and resistance induction records. The initial adaptation plan is essentially a drug-by-drug dosing frequency adjustment sheet: how many times a day was previously given, how many times a day is now given, and the reason for the change—all three must be clearly stated. Frequency compensation refers to increasing the daily dosing frequency without changing the individual dose, thereby increasing the duration of concentration attainment. For each underdosing period in the underdosing record, the drug name corresponding to that period is checked against the resistance induction record. If not included or included but classified as a general risk, frequency compensation is applied to that drug: within the frequency limit, the daily dosing frequency is increased by one; if it exceeds the limit, the original frequency is maintained. After the increase, the cumulative daily dose must not exceed the upper limit registered in the preset dosing parameters. When the drug is included in the resistance induction record and is classified as high risk, it indicates that the cumulative exposure is close to the threshold for inducing resistance. Increasing the frequency would only increase selection pressure; therefore, it is marked as pending evaluation, and the cumulative number of days of medication is recorded. The pending evaluation status and frequency compensation are mutually exclusive; only one labeling will appear under the same drug name. For drugs undergoing frequency compensation, the compensated frequency is recorded. Drugs marked as pending evaluation or not appearing in the underdosing record retain the previously determined daily dosing frequency. The processing type is divided into three mutually exclusive categories: frequency compensation, pending evaluation, and retaining the original value. Drugs not appearing in the underdosing record are also recorded as having one record instead of being omitted; the processing type is recorded as retaining the original value. The initial adaptation protocol uses underdosing records and resistance induction records as the only two inputs. Any change from either source invalidates the entire protocol and reconstructs it based on the current data. The reconstructed protocol covers the same set of drug entries and frequency upper limit constraints, without narrowing the judgment range due to a single change.

[0061] In some embodiments, identifying negative conversion cases earlier than a preset assessment time point based on the initial adaptation scheme to generate relapse risk points includes: establishing an adjustment time point table by sorting out the step-down time points according to the initial adaptation scheme; identifying negative conversion time points in the adjustment time point table to determine early negative conversion items; identifying residual pathogen signals below a preset detection threshold for the early negative conversion items to determine residual false negative items; and performing graded risk assessment based on the residual false negative items to generate relapse risk points.

[0062] Based on the initial adaptation protocol, a timetable for adjusting the de-escalation step is established. De-escalation refers to reducing the types or intensity of medications after sufficient evidence has been obtained. Implementing this too early may leave uneliminated pathogens; therefore, it should only be implemented when the etiological results are sufficiently supportive. This is the purpose of the timetable analysis. For each drug in the initial adaptation protocol, regardless of whether it is marked as frequency compensation or pending evaluation, the treatment start time for the corresponding pathogen category is taken from the time the infection diagnosis was established as recorded in the start and end of the current treatment cycle. Adding 72 hours determines a preset evaluation time point. The 72 hours is taken from the upper limit of the usual time required for blood culture from sampling to obtaining the final identification result. If the corresponding pathogen detection sub-item uses nucleic acid testing instead of blood culture, the preset evaluation time point is changed to the treatment start time for that pathogen category plus 24 hours. The time required for nucleic acid testing to obtain results is significantly shorter than that for blood culture; the 24-hour time limit is also set according to the statistical settings of the routine nucleic acid testing result generation cycle. The adjustment time point table records two items by drug name: the preset evaluation time point and the detection method used. If different drugs in the same initial adaptation protocol use different detection methods, their corresponding evaluation time points can be different without affecting each other's values. Drugs marked as pending evaluation in the initial adaptation protocol are also included in the adjustment time point table, and the method for determining their evaluation time points is exactly the same as that for drugs marked as frequency compensation. Once the initial adaptation protocol is re-established, the adjustment time point table is immediately reviewed, with each drug entry corresponding to the initial adaptation protocol by drug name, without changing the judgment criteria due to simplified terminology.

[0063] The adjustment time point table is used to identify the negative conversion time point and determine the items for early negative conversion. This step checks each drug item listed in the adjustment time point table one by one, and takes the etiological detection sub-item corresponding to the pathogen category of each drug from the infection characteristic data. The negative result determines the negative conversion time point. The adjustment time point table provides the verification range and comparison benchmark. Negative conversion means that the etiological test changes from positive to negative and no pathogen is detected. When the etiological detection sub-item corresponding to a certain drug's pathogen category reports two consecutive negative results, and the interval between the two sampling times is not less than 72 hours, it can be determined that the negative conversion of this pathogen category is established. The negative conversion time point is the sampling time corresponding to the first negative result of these two consecutive results. The 72-hour interval requirement is the same value used for the re-examination cycle of inflammatory indicators when judging false negative results. The time point of negative conversion is compared with the preset assessment time point corresponding to the drug in the adjustment time point table. If the time point of negative conversion is earlier than the preset assessment time point, it indicates that the etiological evidence was insufficient when the negative conversion conclusion was established. In this case, there is a risk of relapse when the step-down method is downgraded, and it is judged as early negative conversion, and an early negative conversion item is generated. If the time point of negative conversion is not earlier than the preset assessment time point, it is treated as normal negative conversion, and no early negative conversion item is generated this time. The assessment time point of the drug remains unchanged and is not adjusted due to this verification. It still uses the previously determined value and waits for the next verification to make a judgment. The early negative conversion item consists of four parts: drug name, pathogen category, actual negative conversion time point, and corresponding preset assessment time point. The latter two time points are retained together to verify the extent of the negative conversion relative to the original arrangement. The early negative conversion item is driven by both the adjustment time point table and infection characteristic data. A refresh on either side triggers a full recalculation and uses the current time point snapshot. Drugs that were hit in the previous round but not in this round are cleared as a whole with the snapshot, and no historical residual entries are left in the results.

[0064] For early negative conversion items, residual pathogen signals below the preset detection threshold are identified to determine residual false negatives. If the pathogen category corresponding to the early negative conversion item is detected using nucleic acid testing, the values ​​recorded in the adjustment time point table are used. The amplification curve cycle count recorded at the time of negative conversion for that pathogen category is reviewed. If the difference between the cycle count and the preset cycle threshold does not exceed 3 cycles, residual pathogen signals are considered to exist in this test. The number of cycles reaching the threshold increases as the initial nucleic acid content in the sample decreases. The smaller the difference, the closer it is to the detection limit. The allowable range of 3 cycles is taken from historical validation data on repeatability errors of nucleic acid testing and is recalibrated every 2 years. If blood culture is used, the actual incubation time of the culture at the time of negative conversion is reviewed. If 80% of the standard positive reporting window is reached but no positive result is reported, residual signals are also considered to exist. Delayed positive reporting is more common when the bacterial count in the sample is low. The 80% percentage is taken from historical statistics of delayed positive reporting in culture and is re-verified every 2 years. Although the criteria for determining residual pathogen signals differ for the two types of detection methods, the judgment logic remains consistent, both using the common characteristic of the current result being close to the positive boundary as the basis for judgment. Residual false negatives carry the original drug name, pathogen category, and actual negative conversion time of the early negative conversion item. The detection method determines its interpretation path, and the specific signal value and detection method used in this judgment are added. If no residual signal is found, it is treated as a true negative conversion, and no residual false negative item is generated in this instance. The original preset evaluation time point in the adjustment time point table for this drug is executed as scheduled without any postponement. Once the re-identification of early negative conversion items occurs, residual false negative items are simultaneously re-identified.

[0065] Recurrence risk points are generated based on residual false negatives. The closer the signal is to the positive boundary, the greater the possibility that residual pathogens in the specimen have not been cleared. The grading quantifies the possibility of recurrence. For nucleic acid testing, residual false negatives are classified as high risk if the difference between the number of cycles and the preset cycle threshold is no more than one cycle, and medium risk if the difference is more than one cycle but no more than three cycles. Both thresholds are further subdivided within the original three-cycle tolerance range for residual false negatives, based on the degree to which the signal is close to the positive boundary. For blood culture, residual false negatives are classified as high risk if the actual incubation time reaches 90% or more of the standard positive reporting window, and medium risk if it reaches 80% but less than 90%. The two thresholds are also further subdivided within the original 80% tolerance range for residual false negatives. When more than one residual false negative exists within the same treatment cycle, the highest risk assessment result is taken as the risk level, not the average level. The residual signals of different drugs are independent; taking the most unfavorable result ensures no risk is overlooked. The relapse risk point consists of three parts: the original drug name, pathogen category, and actual time of negative conversion for the residual false negative. The assessed risk level determines the subsequent treatment approach and is added to the assessed risk level. Only two risk levels are used: high risk or medium risk; there is no low risk level, because the very fact that a residual false negative exists already indicates a certain degree of relapse, and there is no need for further subdivision. Each time a residual false negative is re-identified, the relapse risk point is re-assessed, and all drugs involved are reviewed again in this round.

[0066] The initial fitting protocol is adjusted using feedback from relapse risk points to generate individualized dosing regimens. For each drug involved in a relapse risk point, the corresponding dosing frequency is reviewed in the initial fitting protocol, and the original preset assessment time point is checked in the adjustment time point table. Residual signals indicate that the pathogen may not have been completely eliminated. In this case, downgrading or discontinuing the medication will leave room for relapse. For high-risk cases, the downgrading operation for that drug at the preset assessment time point is cancelled, and the treatment endpoint is extended by 3 days from that assessment time point. The treatment endpoint is the discontinuation time point. For medium-risk cases, the downgrading operation at the preset assessment time point is also cancelled, but the extension amount for the treatment endpoint is changed to 1 day, a smaller extension than in the high-risk case. Both extension amounts are fixed numbers of days and are not calculated continuously based on specific signal values. When the same drug corresponds to more than one relapse risk point, the treatment endpoint is calculated based on the extension amount corresponding to the highest risk level, without cumulative extensions. For drugs not involved in any relapse risk points, the dosing frequency and de-escalation arrangements are carried over to the individualized dosing regimen as before, and the preset assessment time point is executed as scheduled. The individualized dosing regimen records the final dosing frequency, treatment endpoint, and whether the course is extended, as three items. Any update to the end product is immediately executed as a complete re-output, without backflow to the aforementioned steps, and the old version is simultaneously invalidated. For example, in a certain anti-infection cycle: cefoperazone / sulbactam sodium, after frequency compensation, is 4 times daily, corresponding to a high risk of residual Gram-negative bacillus nucleic acid signals. The de-escalation is cancelled and extended by 3 days, resulting in 4 times daily, 3 days added to the endpoint, and extension. Vancomycin is marked as pending assessment and does not involve any risk points. The original 2 times daily and assessment time point are carried over as before, resulting in 2 times daily, as scheduled, and extension not. Voriconazole corresponds to a medium risk of residual fungal culture signals. The de-escalation is cancelled and extended by 1 day, resulting in 2 times daily, 1 day added to the endpoint, and extension.

[0067] In order to implement the above-described method embodiments, a personalized antimicrobial drug administration method for infected patients is proposed to achieve the corresponding functions and technical effects. Figure 3 This invention provides a structural block diagram of a personalized antimicrobial drug administration aid device 300 for infected patients, comprising: Interference identification unit 301 is used to acquire infection characteristic data, liver and kidney function indicators and weight parameters, and to determine interference factors based on the false negative results of the infection characteristic data to form an antibacterial interference record. The induction analysis unit 302 is used to determine the baseline rule based on the antibacterial interference record matching the preset medication rule, use the baseline rule to screen for missed pathogens to obtain pathogen gap markers, and perform drug resistance induction correlation analysis on the pathogen gap markers and the overlapping medication clauses in the baseline rule to obtain drug resistance induction records. The dose determination unit 303 is used to obtain an initial drug dose by matching the dose with the benchmark rule and the weight parameter, identify compensatory manifestations at the lower limit of normal from the liver and kidney function indicators to form a reduction coefficient, and use the reduction coefficient to calculate the target drug dose based on the initial drug dose. The deviation identification unit 304 is used to acquire blood drug concentration data, compare the blood drug concentration data with the expected concentration range calculated based on the target dosage to establish an execution deviation record, identify the peak-valley fluctuation narrowing feature based on the execution deviation record to obtain an accumulation tendency mark, and eliminate interference based on the accumulation tendency mark to generate a dosage insufficiency record. The scheme output unit 305 is used to establish an initial adaptation scheme by frequency compensation using the dose insufficiency record and the drug resistance induction record, identify negative conversion cases earlier than the preset assessment time point based on the initial adaptation scheme to generate relapse risk points, and use the relapse risk points to adjust the initial adaptation scheme and output an individualized dosing scheme.

[0068] The aforementioned personalized antimicrobial drug administration aid device 300 for infected patients can implement a personalized antimicrobial drug administration aid method for infected patients as described in the above method embodiments. The options in the above method embodiments are also applicable to this embodiment and will not be detailed here. The remaining content of this application embodiment can be referred to the content of the above method embodiments, and will not be repeated in this embodiment.

[0069] The purpose of the above embodiments is to reproduce and derive the technical solution of the present invention by way of example, and to fully describe the technical solution, purpose and effect of the present invention. The purpose is to enable the public to have a more thorough and comprehensive understanding of the disclosure of the present invention, and not to limit the scope of protection of the present invention.

Claims

1. A method for individualized administration of antibacterial drugs to infected patients, characterized in that, include: Acquire infection characteristic data, liver and kidney function indicators, and weight parameters; based on the false negative results of the infection characteristic data, identify interfering factors and form antibacterial interference records. Based on the antibacterial interference record, a baseline rule is determined by matching the preset medication rule. The baseline rule is used to screen for missed pathogens and obtain pathogen gap markers. Drug resistance induction correlation analysis is performed on the pathogen gap markers and the overlapping medication clauses in the baseline rule to obtain drug resistance induction records. The initial dosing dose is obtained by matching the baseline rule with the body weight parameter. Compensatory manifestations at the lower limit of normal are identified from the liver and kidney function indicators to form a reduction factor. The target dosing dose is determined by calculating the initial dosing dose using the reduction factor. Obtain blood drug concentration data, compare the blood drug concentration data with the expected concentration range calculated based on the target dosage to establish an execution deviation record, identify the peak-to-valley fluctuation narrowing feature based on the execution deviation record to obtain an accumulation tendency marker, and eliminate interference based on the accumulation tendency marker to generate a dosage insufficiency record; The initial adaptation plan is established by frequency compensation using the insufficient dosage record and the drug resistance induction record. Based on the initial adaptation plan, negative conversion cases earlier than the preset assessment time point are identified to generate relapse risk points. The relapse risk points are used to adjust the initial adaptation plan and output an individualized dosing plan.

2. The method according to claim 1, characterized in that, Drug resistance induction records are obtained by performing a drug resistance induction association analysis on the pathogen gap markers and overlapping drug use clauses in the baseline rules, including: The frequency distribution of the missing detections was obtained by statistically analyzing the missed detection frequency based on the pathogen gap markers. The overlapping drug use list is generated by classifying the overlapping drug use provisions in the aforementioned benchmark rules according to their coverage relationships. The frequency distribution of the gaps is compared with the list of overlapping medications to determine the induced risk points; Drug resistance induction records were obtained by performing drug resistance induction association analysis using the aforementioned induced risk points.

3. The method according to claim 1, characterized in that, The initial dosing dose is obtained by dose matching using the baseline rule and the body weight parameter, including: Based on the aforementioned benchmark rules, the routes of administration and frequency restrictions are analyzed to determine the dosing constraints; The volume deviation is obtained by using the aforementioned weight parameters to identify artificially inflated weight readings during the edema period. A standard dose value is formed by matching the dosing constraints with the volume offset using a preset dose table. The initial dosing dose is obtained by matching the standard dose value with preset dosing parameters.

4. The method according to claim 1, characterized in that, The reduction coefficient is formed by identifying compensatory behaviors at the lower limit of normal from the aforementioned liver and kidney function indicators, including: Based on the aforementioned liver and kidney function indicators, a list of false indicators is generated by identifying dilution-normal readings. The baseline creatinine screening table is obtained by identifying excessively high reserves in the aforementioned list of spurious indicators. The actual potential risks were determined based on the baseline creatinine screening table comparison. The reduction coefficient is generated by performing a graded reduction matching based on the actual hidden danger items.

5. The method according to claim 1, characterized in that, Based on the execution deviation record, the accumulation tendency marker is obtained by identifying the narrowing characteristics of peak and valley fluctuations, including: Based on the execution deviation record, the peak-to-trough volatility of the continuous dosing cycle is calculated to obtain a volatility sequence; The narrowing rate value is determined by statistically analyzing the volatility narrowing rate from the volatility sequence. The hysteresis level is determined by comparing the narrowing rate value with a preset steady-state standard; Accumulation tendency is identified based on the hysteresis level to obtain an accumulation tendency marker.

6. The method according to claim 1, characterized in that, Based on the accumulation tendency markers, interference-induced insufficient dose records are excluded, including: Based on the accumulation tendency markers, a hysteresis distribution table was obtained by statistically compressing the dosing interval. Based on the hysteresis distribution table, hysteresis interference items were identified during periods of excessive drug administration. The hysteresis interference items are subjected to interference removal to obtain the nuclear exposure items; Exclude residual interference in the nuclear exposure items to generate insufficient dose records.

7. The method according to claim 1, characterized in that, Based on the initial adaptation scheme, negative conversion cases earlier than the preset assessment time point are identified to generate relapse risk points, including: Based on the initial adaptation plan, a table of adjustment timing points was created to identify the timing points for step reduction. The adjustment time point table is used to identify the turning point and determine the advance turning point item; For the early negative conversion items, residual false negative items are identified by identifying residual pathogen signals below the preset detection threshold; Based on the residual false negatives, risk assessment is performed to generate recurrence risk points.

8. The method according to claim 2, characterized in that, Based on the pathogen gap markers, the missed detection frequency distribution is obtained by statistically analyzing the missed detection frequency, including: The number of missed bacterial species was obtained by retrospectively analyzing late-positive samples based on the pathogen gap markers. The batches of samples that tested negative consecutively among the missed bacterial species were tracked to obtain a series of negative results; A serial release statistics table is established based on the serial release length of the negative serial release statistics; The frequency distribution of missed detections is obtained by using the aforementioned consecutive firing statistics table to collect the missed detection frequency.

9. The method according to claim 6, characterized in that, Based on the hysteresis distribution table, hysteresis interference items are identified during periods of excessive dosing, including: Based on the hysteresis distribution table, the drug administration offset time period is extracted to determine the offset interference segment; A time-point exposure list is established based on the statistical discrepancies in drug administration time points according to the aforementioned offset interference segments; The list of exposed times is compared with the offset interference segments to generate suspected interference items; The suspected interference items are used to perform hysteresis interference identification to determine the hysteresis interference items.

10. A device for personalized administration of antimicrobial drugs to infected patients, characterized in that, include: An interference identification unit is used to acquire infection characteristic data, liver and kidney function indicators and weight parameters, and to identify interference factors based on the false negative results of the infection characteristic data to form an antibacterial interference record. The induction analysis unit is used to determine the baseline rule based on the antibacterial interference record matching the preset medication rule, use the baseline rule to screen for missed pathogens to obtain pathogen gap markers, and perform drug resistance induction correlation analysis on the pathogen gap markers and the overlapping medication clauses in the baseline rule to obtain drug resistance induction records. The dosage determination unit is used to obtain an initial drug dose by matching the baseline rule with the body weight parameter, identify compensatory manifestations at the lower limit of normal from the liver and kidney function indicators to form a reduction coefficient, and use the reduction coefficient to calculate the target drug dose based on the initial drug dose. The deviation identification unit is used to acquire blood drug concentration data, compare the blood drug concentration data with the expected concentration range calculated based on the target dosage to establish an execution deviation record, identify the peak-valley fluctuation narrowing feature based on the execution deviation record to obtain an accumulation tendency mark, and eliminate interference based on the accumulation tendency mark to generate a dosage insufficiency record. The protocol output unit is used to establish an initial adaptation protocol by using the insufficient dosage record and the drug resistance induction record for frequency compensation, identify negative conversion cases earlier than the preset assessment time point based on the initial adaptation protocol to generate relapse risk points, and use the relapse risk points to adjust the initial adaptation protocol and output an individualized dosing protocol.

Images