A method and system for evaluating results of a pharmaceutical analysis process

Abstract

The present application relates to the technical field of quality management, in particular to a drug analysis process result evaluation method and system, in the present application, the field order splicing is used to generate a digest value sequence, and an iterative hash merging is used to generate a root digest, to form an evidence root fingerprint, so that the evidence set triggers root digest difference and supports integrity verification when any field is changed, a signature value is generated for the root digest and written into a certificate number, a time stamp request is submitted and a time stamp token is written, so that the signature subject, signature time, evidence version, method change and rule range are one-to-one associated, the Dijkstra algorithm is used to calculate the shortest cost path matched with the standard sequence on the event graph, the missing count, additional count, sequence violation count and key bypass count are output, and the grades are mapped according to the threshold, the process consistency conclusion set is generated, and the correlation verification efficiency of the method version, rule version, evidence set, event sequence and disposal action is improved.

Description

Technical Field

[0001] This invention relates to the field of quality management technology, and in particular to a method and system for evaluating the results of a drug analysis process. Background Technology

[0002] The field of quality management technology aims to plan, execute, monitor, measure, and improve multiple activities required to achieve target quality for products or services, so that the output results meet preset quality indicators and acceptance criteria, and trigger handling processes and form traceable records when deviations occur. Target quality is defined by a set of quality indicators.

[0003] The purpose of a drug analysis process result evaluation method is to incorporate multiple results generated by the drug analysis process and their formation process into a unified judgment framework. By setting quality indicators, acceptance criteria and performing rule-based verification, a conclusion is given for each analysis result, and when the trigger threshold is exceeded, the deviation type and handling path are output. The intended effects include shortening the time for result release judgment, improving result traceability and supporting batch release and audit inspection.

[0004] Existing technologies integrate multiple results and processes generated during drug analysis into a unified judgment framework. Their operational model drives rule-based verification based on preset quality indicators and acceptance criteria, outputting deviation types and handling paths when thresholds are exceeded. However, they lack verifiable constraints on judgment criteria and evidence objects. Common practices primarily rely on result values ​​and pass / fail conclusions, with insufficient correlation between method / rule version changes and judgment records. This leads to inconsistencies in the same batch after method revisions or rule updates, increasing time consumption and introducing the risk of omissions. Existing technologies lack strong constraints on field order, numbering locks, and summary consistency. When integral parameters are adjusted or sample preparation records are supplemented, it is difficult to quickly determine whether unauthorized changes have occurred to the evidence set, causing disputes over result release and prolonging deviation investigation cycles. Existing technologies often use textual descriptions or checklists to address process compliance deviations. Missing events, additional events, sequential violations, and bypassing of key steps lack unified counting criteria and grading standards. Furthermore, traceability records are often based on time recordings and signatures, lacking a chain-like relationship between time, signing entities, evidence versions, and handling actions, increasing traceability costs and reducing response speed to external inspections. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of existing technologies and to propose a method and system for evaluating the results of drug analysis processes.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a method for evaluating the results of a drug analysis process, comprising the following steps: S1: Based on the drug analysis task identifier, read the chromatographic method version and check the rule version entries, take the values ​​of plate number, tailing factor, resolution, correlation coefficient, residual, and relative standard deviation, and compare them with the thresholds one by one to generate a set of criteria judgment elements; S2: Based on the criteria for determining the element group, select the original chromatographic data, integration parameter set, standard solution preparation record, and sample preparation record and lock the number, splice them in the field order to generate a summary value sequence and iteratively merge them into a root summary to generate the evidence root fingerprint; S3: Based on the evidence root fingerprint, generate a signature value for the root digest and write the certificate number, submit a timestamp request and write the timestamp token, write the previous version root digest and write the parent-child reference, write the method change control number and bind the rule version number, and generate a time anchor chain node. S4: Based on the time-anchored chain nodes, extract the weighing, liquid preparation, injection, integration confirmation, and result verification events and sort the timestamps. Using the Dixtro algorithm, compare the standard sequence to count missing, extra, sequence violation, and critical bypass, and select the level according to the threshold to generate a process consistency conclusion set. S5: Based on the process consistency conclusion set, reconstruct the root summary according to the sibling node path and compare it with the evidence root fingerprint, verify the signature value and verify the consistency of the timestamp token, read the criteria to determine the deviation type of the element group and select retesting, reprocessing or reintegration in combination with the level and write it into the range item to generate the disposal closed loop entry.

[0007] As a further aspect of the present invention, the criterion judgment element group includes a system applicability judgment item, a standard curve judgment item, and a repeated injection judgment item; the evidence root fingerprint includes a root summary value, an object summary sequence, and an object sequence index; the time anchoring chain node includes a parent reference summary value, a current root summary value, and a time anchor token; the process consistency conclusion set includes a missing event count, an extra event count, and a key control point bypass count; and the disposal closed-loop entry includes a disposal action type, a disposal scope item, and an evidence reference identifier.

[0008] As a further aspect of the present invention, the specific steps for generating the criterion determination element group are as follows: Based on the drug analysis task identifier, the chromatographic method version is read and the rule version entry is checked. The plate number, tailing factor, and resolution are taken and compared with the thresholds respectively. The correlation coefficient and residual are taken and compared with the thresholds respectively. The relative standard deviation is taken and compared with the thresholds. The conclusion, threshold number, and index number of each comparison item are recorded, and the threshold comparison conclusion item is generated. Based on the threshold comparison conclusion, the set of passing item numbers and the set of failing item numbers are summarized, the set of out-of-limit item numbers is extracted and written into the trigger identifier, the deviation type entry is matched and the handling path entry is matched by looking up the table according to the out-of-limit item number, the review requirement identifier is written and the review signature requirement field is written, and the criterion judgment element group is generated.

[0009] As a further aspect of the present invention, the specific steps for generating the evidence root fingerprint are as follows: Based on the aforementioned criteria, the element group is determined, and a hash algorithm is used to select the original chromatographic data, integration parameter set, standard solution preparation record, and sample preparation record and lock the number. The batch number, sample number, and injection sequence number are checked and written into the index table. The batch number, sample number, injection sequence number, method version number, operator identifier, and timestamp are concatenated in the field order to generate a summary value sequence and register the sequence number, thus generating an object summary sequence. Based on the object summary sequence, check the continuity of the numbers in the index table and write it into the missing number set. Pair the numbers in pairs and register the left and right numbers. When a single last number is encountered, copy the last summary value and register the copying identifier. Generate the sibling node path index according to the pairing relationship and register the level number. Generate the sibling node path index set. Based on the sibling node path index set, the paired digest values ​​are merged according to the level number, and the upper-level digest value is generated and written to the upper-level sequence number. The merging is repeated until only a single digest value is retained and the root digest identifier is written. The object number sequence is written, the digest value sequence version number is written, and the generation timestamp is written to generate the evidence root fingerprint.

[0010] As a further aspect of the present invention, the hash algorithm first reads the original chromatographic data, integration parameter set, standard solution preparation record, and sample preparation record and locks the number. It then verifies the batch number, sample number, and injection sequence number and writes them into an index table. The batch number, sample number, injection sequence number, method version number, operator identifier, and timestamp are concatenated into a byte string according to the field order. The field order is verified, and null value fields are written into placeholders. A hash calculation is performed on each byte string to obtain a fixed-length summary value. The summary value, object number, and sequence number are written into the summary index entry. The summary index entries are arranged in ascending order of sequence number, and the summary value sequence is output. When the same object has a duplicate number, a duplicate identifier is written, and the summary value corresponding to the latest timestamp is retained, forming an object summary sequence and outputting it.

[0011] As a further aspect of the present invention, the specific steps for generating the time-anchored chain node are as follows: Based on the evidence root fingerprint, generate a signature value for the root digest and write the certificate number, read the signer identifier and write it into the signature time field, submit a timestamp request and write the timestamp token, read the previous version root digest and write the parent-child reference, write the method change control number and write the rule version number and write the task identifier, and generate a time anchor chain field set. Based on the time anchor chain field set, verify the consistency between the parent and child reference fields and the previous version root digest archive value, verify the rule version number and the chromatography method version adaptation relationship entries, verify that the certificate number status is valid and verify that the validity period range covers the signature time field, verify the uniqueness of the timestamp token number and register the token issuing institution identifier, and generate a chain node verification mark.

[0012] As a further aspect of the present invention, the specific steps for generating the process consistency conclusion set are as follows: Based on the time anchoring chain nodes, the events of weighing, liquid preparation, injection, integration confirmation, and result verification are extracted and their timestamps are sorted. For each event, the event name, occurrence time, operator identifier, equipment identifier, method version number, and task identifier are written and an event list is generated. The event names are compared with the standard sequence item by item and the matching positions are recorded to generate an event alignment sequence. Based on the event alignment sequence, mark the standard events that do not appear and register the missing event name and the position number that should appear, mark the non-standard events and register the additional event name and the position number that appears, mark the standard events that cross the sequence number and reverse the order and register the sequence violation event pairs and accumulate the number of times, mark the points confirmation and result verification that do not appear and register the key bypass events and accumulate the number of times, and generate a process deviation count set; Based on the process deviation count set, the Dixtler algorithm is used to read the critical bypass threshold and compare it with the number of critical bypasses, and select the level field. The total threshold is read and the number of missing, extra, and sequential violations is summed and compared with the total threshold, and the level field is selected. The level field is written, the deviation identifier field is written, and the trigger item number set is written to generate the process consistency conclusion set.

[0013] As a further aspect of the present invention, the Dixtra algorithm first constructs an alignment state graph. The state node field is written with the standard sequence position number, event list position number, and cumulative cost value. The edge field is written with the movement type and cost value. The movement type is divided into synchronous movement, event-only movement, and standard-only movement. The cost value of synchronous movement is written as zero, the cost value of event-only movement is written as one, and the cost value of standard-only movement is written as one. For integral confirmation and result verification, a weighted cost value of five is written. The cumulative cost value of the starting node is written as zero, and the starting node is written to the candidate set. The node with the smallest cumulative cost value in the candidate set is taken out in a loop and written to the fixed set. The new cumulative cost value of the adjacent nodes is calculated and compared with the existing cumulative cost value of the adjacent nodes. The smaller value is replaced and written to the predecessor pointer. The algorithm stops when the candidate set is empty. The alignment path is obtained by backtracking according to the predecessor pointer of the termination node. The alignment path is converted into an event alignment sequence and the total cost value is output.

[0014] As a further aspect of the present invention, the specific steps for generating the closed-loop disposal entry are as follows: Based on the process consistency conclusion set, extract the sibling node path and the object summary value sequence, splice them layer by layer according to the path order and generate summary values ​​layer by layer until the root summary is reconstructed, read the root summary of the evidence root fingerprint and perform an equivalence comparison, record the inconsistency position index, record the object number and record the path index, and generate evidence consistency markers. Based on the evidence consistency marker, verify the matching relationship between the signature value and the certificate number and record the verification conclusion; verify the matching relationship between the timestamp token and the root digest fingerprint and record the verification conclusion; read the deviation type of the criterion judgment element group, read the level field and read the trigger identifier; select and write the retest, re-preparation or re-integration action item according to the deviation type and level execution conditions; write the affected sample number set and the injection sequence number set and write the review signature status; and generate the disposal closed-loop entry.

[0015] A drug analysis process result evaluation system, the system being used to execute the above-described drug analysis process result evaluation method, the system comprising: Threshold verification module: Based on the drug analysis task identifier, it verifies the chromatographic method version and rule version entries, compares the plate number, tailing factor, resolution, correlation coefficient, residual, relative standard deviation and threshold, and generates a set of criteria elements; Evidence splicing module: Based on the aforementioned criterion element group, it locks the original chromatographic data, integration parameter set, standard solution preparation record, and sample preparation record number, splices the batch number, sample number, injection sequence number, method version number, operator identifier, and timestamp, uses the secure hash algorithm 256 to generate a summary and merges the root summary, and generates the root fingerprint number; Anchor Link Module: Based on the root fingerprint number, write the certificate number, signature value, and timestamp token, register the previous version root digest and write the parent-child reference, register the method change control number and write the rule version number, and generate anchor chain entries; Process sequence module: Based on the anchor chain entries, extract weighing, liquid preparation, injection, integration confirmation, and result verification events, sort the timestamps and locate matching points by comparing with the standard sequence, use the Dixtra algorithm to select the alignment path with the minimum cost, and count missing, extra, sequence violation, and critical bypass according to this and determine the level, and generate a consistency level set; The closed-loop processing module reconstructs the root summary based on the consistency level set, compares it with the root fingerprint number according to the sibling node path, writes the signature verification conclusion and timestamp verification conclusion, reads the deviation type of the criterion element group and selects retest, re-preparation, and re-integration and writes it into the range item to generate a closed-loop entry.

[0016] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In this invention, the chromatographic method version is read based on the drug analysis task identifier and the rule version entry is checked. The judgment scope is limited to the method version and rule version and can be verified. Then, the plate number, tailing factor, resolution, correlation coefficient, residual, and relative standard deviation are compared with the threshold item by item. The output consists of a criterion judgment element group composed of parameter name, measured value, threshold, direction of exceeding limit, and judgment conclusion, so that the result conclusion has a field-level evidence chain and reduces the drift of the caliber. In this invention, a sequence of digest values ​​is generated by concatenating fields in order and then iteratively hashing and merging them to generate a root digest, forming a root fingerprint of evidence. This allows the evidence set to trigger a root digest difference and support integrity verification when any field changes. A signature value is generated for the root digest and a certificate number is written into it. A timestamp request is submitted and a timestamp token is written into it. At the same time, the previous version root digest, parent-child references, method change control number, and rule version number are written into it and bound to the rule version number, forming a time anchor chain node. This establishes a one-to-one association between the signing subject, signing time, evidence version, method change, and rule definition, reducing the space for post-event data entry and improving audit reproducibility. In this invention, the Dixtra algorithm is used to calculate the shortest cost path matching the standard sequence on the event graph, outputting missing counts, extra counts, sequence violation counts, and critical bypass counts, and mapping levels according to thresholds to generate a process consistency conclusion set. This transforms process deviations from textual descriptions into quantifiable counts and levels, forming review priorities and shortening the time spent on selecting disposal paths. The root summary is reconstructed according to the sibling node paths and compared with the evidence root fingerprint. The signature value and timestamp token consistency are verified. The source of differences is located in multiple version branches. At the same time, the deviation type of the criterion judgment element group is read and combined with the level to select retesting, re-preparation, re-integration and write it into the range item, forming a disposal closed loop entry, improving the efficiency of verifying the association of multiple objects such as method version, rule version, evidence set, event sequence and disposal action. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the workflow of the present invention; Figure 2 This is a system flowchart of the present invention. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Example

[0019] Please see Figure 1 This invention provides a technical solution: a method for evaluating the results of a drug analysis process, comprising the following steps: S1: Based on the drug analysis task identifier, read the chromatographic method version and check the rule version entries, take the values ​​of plate number, tailing factor, resolution, correlation coefficient, residual, and relative standard deviation, and compare them with the thresholds one by one to generate a set of criteria judgment elements; S2: Based on the criteria judgment element group, select the original chromatographic data, integration parameter set, standard solution preparation record, and sample preparation record and lock the number, splice them in the field order to generate a summary value sequence and iteratively merge them into a root summary to generate the evidence root fingerprint; S3: Based on the root fingerprint of evidence, generate a signature value for the root digest and write the certificate number, submit a timestamp request and write the timestamp token, write the previous version root digest and write the parent-child reference, write the method change control number and bind the rule version number, and generate a time anchor chain node. S4: Based on time-anchored chain nodes, extract the weighing, liquid preparation, injection, integration confirmation, and result verification events and sort the timestamps. Use the Dixtro algorithm to count missing, extra, sequence violations, and critical bypasses against the standard sequence and select the level according to the threshold to generate a process consistency conclusion set. S5: Based on the process consistency conclusion set, reconstruct the root summary according to the sibling node path and compare it with the evidence root fingerprint, verify the signature value and verify the consistency of the timestamp token, read the criteria to determine the deviation type of the element group and combine the level to select retesting, reprocessing or reintegration and write it into the range item, and generate the disposal closed-loop entry.

[0020] The criteria judgment element group includes system applicability judgment item, standard curve judgment item, and repeated injection judgment item; the evidence root fingerprint includes root summary value, object summary sequence, and object sequence index; the time anchor chain node includes parent reference summary value, current root summary value, and time anchor token; the process consistency conclusion set includes missing event count, extra event count, and key control point bypass count; and the disposal closure loop item includes disposal action type, disposal scope item, and evidence reference identifier.

[0021] The specific steps for generating the criterion-based feature group are as follows: Based on the drug analysis task identifier, the chromatographic method version is read and the rule version entry is checked. The plate number, tailing factor, and resolution are taken and compared with the thresholds respectively. The correlation coefficient and residual are taken and compared with the thresholds respectively. The relative standard deviation is taken and compared with the thresholds. The conclusion, threshold number, and index number of each comparison item are recorded, and the threshold comparison conclusion item is generated. Based on the threshold comparison conclusion, summarize the set of pass item numbers and the set of fail item numbers, extract the set of over-limit item numbers and write it into the trigger identifier, look up the table by the over-limit item number to match the deviation type item and the handling path item, write the review identifier and the review signature requirement field, and generate the criterion judgment element group. Based on the drug analysis task identifier, a decision table index is used to locate rule version entries and lock entry numbers. Least squares linear regression is used to calculate regression coefficients for the standard curve point sequence and write the correlation coefficient values ​​(retaining 4 decimal places) and residual sequence values ​​(retaining 4 decimal places). Plate number values ​​are read, retained as integers, and compared with the lower threshold limit (greater than or equal to). Tailing factor values ​​are read, retained as 2 decimal places, and compared with the upper threshold limit (less than or equal to). Resolution values ​​are read, retained as 2 decimal places, and compared with the lower threshold limit (greater than or equal to). Correlation coefficient values ​​are read and compared with the lower threshold limit (0.995) (greater than or equal to). Maximum residual value is read and compared with the upper threshold limit (5.0) (less than or equal to). Two-pass statistical calculations are used to calculate the mean and standard deviation of the repeated injection result sequence, converting the relative standard deviation to 2 decimal places and comparing it with the upper threshold limit (2.0) (less than or equal to). The conclusion, threshold number, and indicator number for each comparison item are recorded, generating a threshold comparison conclusion item. Based on the threshold comparison conclusion, the condition node is written using the Rete rule matching method, and the node number is written. The condition node is written with the index number, comparison conclusion, and threshold number. The connection node is written with the connection relationship as logical AND and logical NOT. The mutual exclusion rule pair is written using the priority mutual exclusion matrix method, and the priority value is written. The priority value is written as 100 for system applicability failure, 90 for standard curve failure, 80 for repeated injection failure, and 70 for blank interference exceeding the limit. The comparison conclusion is loaded as a fact item and the hit rule number sequence is updated. The set of pass item numbers is summarized and the set of fail item numbers is summarized. The set of exceeding the limit item numbers is extracted and the trigger flag value is 1. The deviation type item is matched with the handling path item by the exceeding limit item number. The review signature requirement field is written with two signatures and the role identifier is written as reviewer and approver. The criterion judgment element group is generated.

[0022] The specific steps for generating the root fingerprint of evidence are as follows: Based on the criteria for determining the element group, a hash algorithm is used to select the original chromatographic data, integration parameter set, standard solution preparation record, and sample preparation record and lock the number. The batch number, sample number, and injection sequence number are checked and written into the index table. The batch number, sample number, injection sequence number, method version number, operator identifier, and timestamp are concatenated in the field order to generate a summary value sequence and register the sequence number, thus generating an object summary sequence. Based on the object summary sequence, check the continuity of the numbers in the index table and write them into the missing number set. Pair them up by number and register the left and right numbers. When a single last number is encountered, copy the last summary value and register the copying identifier. Generate the sibling node path index according to the pairing relationship and register the level number. Generate the sibling node path index set. Based on the sibling node path index set, merge the paired digest values ​​by level number and generate the upper-level digest value and write the upper-level sequence number. Repeat the merging until only a single digest value is retained and write the root digest identifier. Write the object number sequence, write the digest value sequence version number and write the generation timestamp to generate the evidence root fingerprint. Based on the criteria for determining the element group, the secure hash algorithm 256 is used to select the original chromatographic data, integration parameter set, standard solution preparation record, and sample preparation record and lock the number. The batch number, sample number, and injection sequence number are checked and written into the index table. The batch number, sample number, injection sequence number, method version number, operator identifier, and timestamp are concatenated in a fixed order using the field sequence table. The field separator is written in 1F hexadecimal, the character encoding is written in UTF8, the timestamp is written in YYYYMMDDhhmmss format, and the null value field is written in NA. The concatenated byte string is input into the hash calculation and the summary value is output. The summary output encoding is written in hexadecimal lowercase, the sequence number start value is written as 1 and sorted and incremented according to the index table to generate the object summary sequence. Based on the object summary sequence, the Merkle tree algorithm is used to check the continuity of the numbers in the index table and write them into the missing number set. The pairing rules are used to pair the numbers 1 and 2 and register the left and right numbers and increment the pairing. The last individual number is copied to the last summary value and the copying identifier is registered and written to DUP. The level number is written to the starting value of 1 and incremented by level. The sibling node summary value is written according to the pairing relationship and the direction identifier is written to L or R and the same level number is written. The path index field is written with the level number, the same level number, and the direction identifier to generate the sibling node path index set. Based on the sibling node path index set, the Merkle tree algorithm and the secure hash algorithm 256 are used to merge paired digest values ​​according to the level number and write them to the upper level number. The concatenation order is written with the left digest value first and the right digest value last. The concatenation separator is written as 00 hexadecimal. The concatenated byte string is input for hash calculation and the upper level digest value is output and written. The merging is repeated until only a single digest value is retained and the root digest identifier is written. The object number sequence is written and arranged in ascending order. The digest value sequence version number is written with the rule version number plus a 4-digit serial number. The timestamp is generated and written in the format YYYYMMDDhhmmss to generate the evidence root fingerprint.

[0023] The hash algorithm first reads the raw chromatographic data, integration parameter set, standard solution preparation record, and sample preparation record, and locks the serial numbers. It then verifies the batch number, sample number, and injection sequence number and writes them into the index table. The batch number, sample number, injection sequence number, method version number, operator identifier, and timestamp are concatenated into a byte string according to the field order. The field order is verified, and null values ​​are checked and written into placeholders. A hash calculation is performed on each byte string to obtain a fixed-length summary value. The summary value, along with the object number and sequence number, is written into the summary index entry. The summary index entries are arranged in ascending order of sequence number, and the summary value sequence is output. When the same object has a duplicate number, a duplicate identifier is written, and the summary value corresponding to the latest timestamp is retained. This forms the object summary sequence and is then output.

[0024] The specific steps for generating time-anchored link nodes are as follows: Based on the root fingerprint of evidence, a signature value is generated from the root digest and written into the certificate number, the signer identifier is read and written into the signature time field, a timestamp request is submitted and a timestamp token is written, the previous version root digest is read and written into the parent-child reference, the method change control number is written and the rule version number is written, the task identifier is written, and a time anchor chain field set is generated. Based on the time-anchored chain field set, verify the consistency between the parent and child reference fields and the previous version root digest archive value, verify the rule version number and the chromatographic method version adaptation relationship entries, verify that the certificate number status is valid and verify that the validity period range covers the signature time field, verify the uniqueness of the timestamp token number and register the token issuing institution identifier, and generate chain node verification marks. Based on the root fingerprint, using RSA signature, the root digest byte string is read and the digest encoding is fixed as hexadecimal lowercase. The private key length is written to 2048 bits, the padding mode is written to PSS, the hash identifier is written to SHA256, the salt length is written to 32 bytes, the signature value is generated and written to the signature value field and the certificate number field, the signer identifier is read and written to the signature time field in the format YYYYMMDDhhmmss, the RFC3161 timestamp protocol is adopted, the request version is written to 1, the message digest algorithm identifier is written to SHA256, the nonce length is written to 8 bytes, the certificate request identifier is written to true, the timestamp request is submitted and the timestamp token is written and written to the token number field, the previous version root digest is read and written to the parent-child reference field and the reference type field is written to PARENT, the method change control number is written and the rule version number is written and the task identifier is written, and a time anchor chain field set is generated. Based on the time-anchored chain field set, the certificate chain verification rules are adopted. The parent-child reference field is read and compared byte by byte with the root digest archive value of the previous version and a consistency identifier is written. The rule version number is read and compared item by item with the chromatography method version adaptation relationship entry and an adaptation identifier is written. The certificate number is read and the certificate status field value VALID is verified. The certificate validity start time is verified to be earlier than the signing time and the certificate validity end time is verified to be later than the signing time. The certificate revocation status field value NOTREVOKED is verified. The RFC3161 token verification rules are adopted. The timestamp token is read and the token signer certificate serial number is verified. The message digest in the token is read and compared byte by byte with the root digest fingerprint. The token number is read and compared with the uniqueness field of the token registration table and a duplicate identifier is written. The token issuing institution identifier is registered and written into the institution code field. A chain node verification mark is generated.

[0025] The specific steps for generating a process consistency conclusion set are as follows: Based on time-anchored chain nodes, the events of weighing, liquid preparation, injection, integration confirmation, and result verification are extracted and their timestamps are sorted. For each event, the event name, occurrence time, operator identifier, equipment identifier, method version number, and task identifier are written and an event list is generated. The event names are compared with the standard sequence item by item and the matching position is recorded to generate an event alignment sequence. Based on the event alignment sequence, mark the standard events that do not appear and register the missing event name and the position number that should appear, mark the non-standard events and register the additional event name and the position number of the occurrence, mark the standard events across the sequence number in reverse order and register the sequence violation event pairs and accumulate the number of times, mark the integration confirmation and result review that do not appear and register the key bypass events and accumulate the number of times, and generate a process deviation count set; Based on the process deviation count set, the Dixtella algorithm is used to read the critical bypass threshold and compare it with the number of critical bypasses and select the level field. The total threshold is read and the number of missing, extra and sequential violations are summed and compared with the total threshold and the level field is selected. The level field is written and the deviation identifier field is written and the set of trigger item numbers is written to generate the process consistency conclusion set. Based on time-anchored chain nodes, the events of weighing, solution preparation, sample injection, integration confirmation, and result verification are extracted and their timestamps are sorted. For each event, the event name, occurrence time, operator identifier, equipment identifier, method version number, and task identifier are written, and an event list is generated. The rule for writing the event identifier code is event name followed by a colon followed by the method version number. The writing time format is YYYYMMDDhhmmss. The missing identifier value is MISS, the extra identifier value is EXTRA, the reverse order identifier value is ORDER, and the critical bypass identifier value is BYPASS. The event names are compared with the standard sequence item by item, and the matching position is recorded. The branch identifier value is XOR, the parallel identifier value is AND, and the loop identifier value is LOOP. The occurrence frequency is summarized by task identifier and written as a count, generating an event alignment sequence. Based on the event alignment sequence, and using the process alignment cost rule, the system marks the absence of standard events and registers the missing event name and its corresponding position number; marks non-standard events and registers the additional event name and its corresponding position number; marks standard events that cross sequence number reversal and registers the sequence violation event pairs and accumulates the number of occurrences; marks the absence of points confirmation and result verification and registers the key bypass events and accumulates the number of occurrences; writes a value of 0 to the value field for synchronous movement, a value of 1 to the value field for missing events, a value of 1 to the value field for additional events, a value of 1 to the value field for sequence violation, and a value of 5 to the value field for key bypass; and generates a process deviation count set. Based on the process deviation count set, the Dixtler algorithm is used to construct a state node field containing the standard position number, event position number, and cumulative cost value. An edge field is constructed containing the movement type and cost value. The cumulative cost value of the starting node is written as 0. The candidate set is written to the starting node. The node with the smallest cumulative cost value in the candidate set is retrieved in a loop and written to the established set. The new cumulative cost value is calculated for adjacent nodes and compared with the existing cumulative cost value of adjacent nodes. The smaller value is replaced and written to the predecessor pointer. The terminating node is written to the predecessor pointer chain, and the alignment path is obtained by backtracking and written to the total cost value. The critical bypass threshold is read and compared with the number of critical bypasses, and the level field is selected. The total threshold is read and the number of missing, extra, and sequence violations are summed and compared with the total threshold, and the level field is selected. The level field, deviation identifier field, and trigger item number set are written to generate a process consistency conclusion set.

[0026] Dixtra's algorithm first constructs an alignment state graph. The state node field contains the standard sequence position number, event list position number, and cumulative cost value. The edge field contains the movement type and cost value. Movement types are categorized as synchronous movement, event-only movement, and standard-only movement. Synchronous movement has a cost value of zero, event-only movement has a cost value of one, and standard-only movement has a cost value of one. For integral confirmation and result verification, a weighted cost value of five is written. The cumulative cost value of the starting node is written to zero, and the starting node is added to the candidate set. The algorithm iteratively retrieves the node with the smallest cumulative cost value from the candidate set and adds it to the established set. A new cumulative cost value is calculated for adjacent nodes and compared with their existing cumulative costs. The smaller value is replaced and written to the predecessor pointer. The algorithm stops when the candidate set is empty. The alignment path is obtained by backtracking according to the predecessor pointer of the terminating node. The alignment path is converted into an event alignment sequence, and the total cost value is output.

[0027] The specific steps for generating closed-loop disposal entries are as follows: Based on the process consistency conclusion set, extract the sibling node path and the object summary value sequence, concatenate them layer by layer according to the path order and generate summary values ​​layer by layer until the root summary is reconstructed, read the root summary of the evidence root fingerprint and perform equivalence comparison, record the inconsistency position index, record the object number and record the path index, and generate evidence consistency mark. Based on the evidence consistency mark, verify the matching relationship between the signature value and the certificate number and record the verification conclusion; verify the matching relationship between the timestamp token and the root digest fingerprint and record the verification conclusion; read the deviation type of the criterion judgment element group and read the level field and the trigger identifier; select and write the retest or re-preparation or re-integration action item according to the deviation type and level execution conditions; write the affected sample number set and the injection sequence number set and write the review signature status; and generate the disposal closed-loop entry. Based on the process consistency conclusion set, a secure hash algorithm 256 and a Merkel proof algorithm are used to extract sibling node paths and object digest value sequences. The path field is written with the level number, same-level sequence number, and direction identifier L or R. The digest value field is written with hexadecimal lowercase. The paths are concatenated layer by layer in order, and the concatenation order is written with the digest value first when the direction identifier L is used and the digest value last when the direction identifier R is used. The concatenation separator is written with 00 hexadecimal. The concatenated byte string is input for hash calculation and output with a 256-bit digest, which is written to the level digest field. The level number is incremented in a loop until the reconstructed root digest is generated. The evidence root fingerprint root digest is read and a byte-by-byte equality comparison is performed. The inconsistency position index is recorded and written with the index starting value 1 and incremented by byte. The object number and path index are recorded to generate evidence consistency markers. Based on the evidence consistency marker, the system adopts the RSA signature verification rule and the RFC3161 token verification rule. It reads the signature value field and encodes it as Base64, reads the certificate number field, reads the 2048-bit public key length, writes the hash identifier SHA256, writes the padding mode PSS, and writes the salt length of 32 bytes. It verifies the consistency between the signature value and the root digest and writes the verification conclusion field with the value PASS or FAIL. It reads the timestamp token, reads the message digest within the token, and compares it byte-by-byte with the root digest fingerprint, writing the comparison conclusion field with the value PASS or FAIL. It reads the deviation type of the criterion judgment element group, reads the level field, and reads the trigger identifier with the value 1 or 0. It writes the condition branch number according to the deviation type and level and writes the action item value of retest, re-preparation, or re-integration. It writes the affected sample number set and arranges it in ascending order by sample number. It writes the sample injection sequence number set and arranges it in ascending order by sequence number. It writes the review signature status value SIGNED or UNSIGNED, generating a closed-loop disposal entry.

[0028] Please see Figure 2 A drug analysis process result evaluation system, used to execute the above-mentioned drug analysis process result evaluation method, the system comprising: Threshold verification module: Based on the drug analysis task identifier, it verifies the chromatographic method version and rule version entries, compares the plate number, tailing factor, resolution, correlation coefficient, residual, relative standard deviation and threshold, and generates a set of criteria elements; Evidence assembly module: Based on the criterion element group, it locks the original chromatographic data, integration parameter set, standard solution preparation record, and sample preparation record number, and assembles the batch number, sample number, injection sequence number, method version number, operator identifier, and timestamp. It uses the secure hash algorithm 256 to generate a summary and merge the root summary to generate the root fingerprint number. Anchor Link Module: Based on the root fingerprint, write the certificate number, signature value, and timestamp token, register the previous version root digest and write the parent and child references, register the method change control number and write the rule version number, and generate anchor chain entries; Process sequence module: Based on anchor chain entries, it extracts weighing, liquid preparation, injection, integration confirmation, and result verification events, sorts the timestamps and locates matching points by comparing with the standard sequence, uses the Dixtra algorithm to select the alignment path with the minimum cost, and counts missing, extra, sequence violation, and critical bypass according to this and determines the level, generating a consistency level set; The closed-loop processing module reconstructs the root summary based on the consistency level set, reconstructs it according to the sibling node path and compares it with the root fingerprint number, writes the signature verification conclusion and timestamp verification conclusion, reads the deviation type of the criterion element group and selects retest, re-preparation, and re-integration and writes it into the range item to generate the closed-loop entry.

[0029] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A method for evaluating the results of a drug analysis process, characterized in that, Includes the following steps: S1: Based on the drug analysis task identifier, read the chromatographic method version and check the rule version entries, take the values ​​of plate number, tailing factor, resolution, correlation coefficient, residual, and relative standard deviation, and compare them with the thresholds one by one to generate a set of criteria judgment elements; S2: Based on the criteria for determining the element group, select the original chromatographic data, integration parameter set, standard solution preparation record, and sample preparation record and lock the number, splice them in the field order to generate a summary value sequence and iteratively merge them into a root summary to generate the evidence root fingerprint; S3: Based on the evidence root fingerprint, generate a signature value for the root digest and write the certificate number, submit a timestamp request and write the timestamp token, write the previous version root digest and write the parent-child reference, write the method change control number and bind the rule version number, and generate a time anchor chain node. S4: Based on the time-anchored chain nodes, extract the weighing, liquid preparation, injection, integration confirmation, and result verification events and sort the timestamps. Using the Dixtro algorithm, compare the standard sequence to count missing, extra, sequence violation, and critical bypass, and select the level according to the threshold to generate a process consistency conclusion set. S5: Based on the process consistency conclusion set, reconstruct the root summary according to the sibling node path and compare it with the evidence root fingerprint, verify the signature value and verify the consistency of the timestamp token, read the criteria to determine the deviation type of the element group and select retesting, reprocessing or reintegration in combination with the level and write it into the range item to generate the disposal closed loop entry.

2. The method for evaluating the results of drug analysis process according to claim 1, characterized in that, The criteria judgment element group includes system applicability judgment item, standard curve judgment item, and repeated injection judgment item; the evidence root fingerprint includes root summary value, object summary sequence, and object sequence index; the time anchor chain node includes parent reference summary value, current root summary value, and time anchor token; the process consistency conclusion set includes missing event count, extra event count, and key control point bypass count; and the disposal closed loop entry includes disposal action type, disposal scope item, and evidence reference identifier.

3. The method for evaluating the results of drug analysis process according to claim 1, characterized in that, The specific steps for generating the criterion-based decision element group are as follows: Based on the drug analysis task identifier, the chromatographic method version is read and the rule version entry is checked. The plate number, tailing factor, and resolution are taken and compared with the thresholds respectively. The correlation coefficient and residual are taken and compared with the thresholds respectively. The relative standard deviation is taken and compared with the thresholds. The conclusion, threshold number, and index number of each comparison item are recorded, and the threshold comparison conclusion item is generated. Based on the threshold comparison conclusion, the set of passing item numbers and the set of failing item numbers are summarized, the set of out-of-limit item numbers is extracted and written into the trigger identifier, the deviation type entry is matched and the handling path entry is matched by looking up the table according to the out-of-limit item number, the review requirement identifier is written and the review signature requirement field is written, and the criterion judgment element group is generated.

4. The method for evaluating the results of drug analysis process according to claim 1, characterized in that, The specific steps for generating the root fingerprint of the evidence are as follows: Based on the aforementioned criteria, the element group is determined, and a hash algorithm is used to select the original chromatographic data, integration parameter set, standard solution preparation record, and sample preparation record and lock the number. The batch number, sample number, and injection sequence number are checked and written into the index table. The batch number, sample number, injection sequence number, method version number, operator identifier, and timestamp are concatenated in the field order to generate a summary value sequence and register the sequence number, thus generating an object summary sequence. Based on the object summary sequence, check the continuity of the numbers in the index table and write it into the missing number set. Pair the numbers in pairs and register the left and right numbers. When a single last number is encountered, copy the last summary value and register the copying identifier. Generate the sibling node path index according to the pairing relationship and register the level number. Generate the sibling node path index set. Based on the sibling node path index set, the paired digest values ​​are merged according to the level number, and the upper-level digest value is generated and written to the upper-level sequence number. The merging is repeated until only a single digest value is retained and the root digest identifier is written. The object number sequence is written, the digest value sequence version number is written, and the generation timestamp is written to generate the evidence root fingerprint.

5. The method for evaluating the results of drug analysis process according to claim 1, characterized in that, The hash algorithm first reads the raw chromatographic data, integration parameter set, standard solution preparation record, and sample preparation record and locks the numbers. It then verifies the batch number, sample number, and injection sequence number and writes them into the index table. The batch number, sample number, injection sequence number, method version number, operator identifier, and timestamp are concatenated into a byte string according to the field order. The field order is verified, and null values ​​are checked and written into placeholders. A hash calculation is performed on each byte string to obtain a fixed-length summary value. The summary value, object number, and sequence number are written into the summary index entry. The summary index entries are arranged in ascending order of sequence number, and the summary value sequence is output. When the same object has a duplicate number, a duplicate identifier is written, and the summary value corresponding to the latest timestamp is retained. This forms the object summary sequence and is output.

6. The method for evaluating the results of drug analysis process according to claim 1, characterized in that, The specific steps for generating the time-anchored chain node are as follows: Based on the evidence root fingerprint, generate a signature value for the root digest and write the certificate number, read the signer identifier and write it into the signature time field, submit a timestamp request and write the timestamp token, read the previous version root digest and write the parent-child reference, write the method change control number and write the rule version number and write the task identifier, and generate a time anchor chain field set. Based on the time anchor chain field set, verify the consistency between the parent and child reference fields and the previous version root digest archive value, verify the rule version number and the chromatography method version adaptation relationship entries, verify that the certificate number status is valid and verify that the validity period range covers the signature time field, verify the uniqueness of the timestamp token number and register the token issuing institution identifier, and generate a chain node verification mark.

7. The method for evaluating the results of drug analysis process according to claim 1, characterized in that, The specific steps for generating the process consistency conclusion set are as follows: Based on the time anchoring chain nodes, the events of weighing, liquid preparation, injection, integration confirmation, and result verification are extracted and their timestamps are sorted. For each event, the event name, occurrence time, operator identifier, equipment identifier, method version number, and task identifier are written and an event list is generated. The event names are compared with the standard sequence item by item and the matching positions are recorded to generate an event alignment sequence. Based on the event alignment sequence, mark the standard events that do not appear and register the missing event name and the position number that should appear, mark the non-standard events and register the additional event name and the position number that appears, mark the standard events that cross the sequence number and reverse the order and register the sequence violation event pairs and accumulate the number of times, mark the points confirmation and result verification that do not appear and register the key bypass events and accumulate the number of times, and generate a process deviation count set; Based on the process deviation count set, the Dixtler algorithm is used to read the critical bypass threshold and compare it with the number of critical bypasses, and select the level field. The total threshold is read and the number of missing, extra, and sequential violations is summed and compared with the total threshold, and the level field is selected. The level field is written, the deviation identifier field is written, and the trigger item number set is written to generate the process consistency conclusion set.

8. The method for evaluating the results of a drug analysis process according to claim 1, characterized in that, The Dixtra algorithm first constructs an alignment state graph. The state node field contains the standard sequence position number, event list position number, and cumulative cost value. The edge field contains the movement type and cost value. Movement types are divided into synchronous movement, event-only movement, and standard-only movement. The cost value for synchronous movement is zero, for event-only movement it is one, and for standard-only movement it is one. For integral confirmation and result verification, a weighted cost value of five is written. The cumulative cost value of the starting node is written to zero, and the starting node is written to the candidate set. The algorithm iteratively extracts the node with the smallest cumulative cost value from the candidate set and writes it to the established set. A new cumulative cost value is calculated for adjacent nodes and compared with the existing cumulative cost values ​​of adjacent nodes. The smaller value is replaced and written to the predecessor pointer. The algorithm stops when the candidate set is empty. The alignment path is obtained by backtracking according to the predecessor pointer of the termination node. The alignment path is converted into an event alignment sequence, and the total cost value is output.

9. The method for evaluating the results of a drug analysis process according to claim 1, characterized in that, The specific steps for generating the aforementioned closed-loop disposal entry are as follows: Based on the process consistency conclusion set, extract the sibling node path and the object summary value sequence, splice them layer by layer according to the path order and generate summary values ​​layer by layer until the root summary is reconstructed, read the root summary of the evidence root fingerprint and perform an equivalence comparison, record the inconsistency position index, record the object number and record the path index, and generate evidence consistency markers. Based on the evidence consistency marker, verify the matching relationship between the signature value and the certificate number and record the verification conclusion; verify the matching relationship between the timestamp token and the root digest fingerprint and record the verification conclusion; read the deviation type of the criterion judgment element group, read the level field and read the trigger identifier; select and write the retest, re-preparation or re-integration action item according to the deviation type and level execution conditions; write the affected sample number set and the injection sequence number set and write the review signature status; and generate the disposal closed-loop entry.

10. A drug analysis process result evaluation system, characterized in that, The method for evaluating the results of a drug analysis process according to any one of claims 1-9, wherein the system comprises: Threshold verification module: Based on the drug analysis task identifier, it verifies the chromatographic method version and rule version entries, compares the plate number, tailing factor, resolution, correlation coefficient, residual, relative standard deviation and threshold, and generates a set of criteria elements; Evidence splicing module: Based on the aforementioned criterion element group, it locks the original chromatographic data, integration parameter set, standard solution preparation record, and sample preparation record number, splices the batch number, sample number, injection sequence number, method version number, operator identifier, and timestamp, uses the secure hash algorithm 256 to generate a summary and merges the root summary, and generates the root fingerprint number; Anchor Link Module: Based on the root fingerprint number, write the certificate number, signature value, and timestamp token, register the previous version root digest and write the parent-child reference, register the method change control number and write the rule version number, and generate anchor chain entries; Process sequence module: Based on the anchor chain entries, extract weighing, liquid preparation, injection, integration confirmation, and result verification events, sort the timestamps and locate matching points by comparing with the standard sequence, use the Dixtra algorithm to select the alignment path with the minimum cost, and count missing, extra, sequence violation, and critical bypass according to this and determine the level, and generate a consistency level set; The closed-loop processing module reconstructs the root summary based on the consistency level set, compares it with the root fingerprint number according to the sibling node path, writes the signature verification conclusion and timestamp verification conclusion, reads the deviation type of the criterion element group and selects retest, re-preparation, and re-integration and writes it into the range item to generate a closed-loop entry.

Images