A stem cell preparation quality reliable management method and system based on quality rule object benchmark and execution boundary control

Abstract

The application discloses a stem cell preparation quality reliable management method and system based on quality rule object benchmark and execution boundary control, comprising the following steps: constructing a quality rule object set for stem cell preparation quality evaluation, acquiring multi-source quality data of a stem cell preparation batch, generating a standardized quality fact package based on the multi-source quality data, creating a one-time quality evaluation execution instance for the standardized quality fact package, establishing a binding relationship between the one-time quality evaluation execution instance and metadata, controlling a current quality operation based on the binding relationship and permission identification through a controlled execution boundary, executing stem cell preparation quality evaluation based on the quality rule object set and the standardized quality fact package, obtaining evaluation and decision conclusions, generating dynamic isolation records, and executing at least one safety disposal action, encapsulating management output results associated with the quality rule object set into standardized quality events, and forming a traceable evidence chain based on the standardized quality events.

Description

Technical Field

[0001] This application relates to the field of computer technology, specifically to a reliable quality control method for stem cell preparation based on quality rule object benchmarks and execution boundary control, a reliable quality control device for stem cell preparation based on quality rule object benchmarks and execution boundary control, a computer device, and a computer-readable storage medium. Background Technology

[0002] As stem cell preparation processes become increasingly automated and digitalized, quality control systems are incorporating edge data acquisition, multimodal detection, model inference, graphical model prediction, automated decision routing, and electronic batch record archiving. Research has revealed that stem cell preparation quality control methods generally suffer from several shortcomings, including fragmented rule benchmarks, inconsistent input facts, lack of controlled boundaries for evaluation instances, coarse-grained access control, susceptibility to silent loss of event feedback, lack of multi-source consensus verification and dynamic isolation mechanisms, and uncontrolled rule updates. These shortcomings lead to frequent problems such as rule drift, model misuse, factual conflicts, unauthorized operations, broken evidence chains, and the accidental release of high-risk batches. Consequently, it is difficult to establish a reliable, consistent, and auditable quality assessment loop across the entire chain, and a lack of a reliable end-to-end control mechanism centered on unified rule benchmarks and controlled execution boundaries. Summary of the Invention

[0003] This application provides a method, apparatus, device, and computer-readable storage medium for reliable quality control of stem cell preparation based on quality rule object benchmarks and execution boundary control. It can establish a full-link reliable control mechanism with unified rule benchmarks and controlled execution boundaries as the core, from rule construction, fact input, instance binding, operation verification, dynamic isolation to evidence chain archiving.

[0004] On the one hand, embodiments of this application provide a reliable quality control method for stem cell preparation based on quality rule object benchmarks and execution boundary control, including: Construct a quality rule object set for quality assessment of stem cell preparation. The quality rule object set includes at least one of rule version references, model version references, threshold version references, permission policies, risk policies, archiving field requirements, and quality assessment rules. Acquire multi-source quality data for stem cell preparation batches and generate standardized quality fact packages based on the multi-source quality data. The standardized quality fact packages are bound to batch identifiers, rule version references, model version references, and threshold version references. Create a one-time quality assessment execution instance for the standardized quality fact package, and establish the binding relationship between the one-time quality assessment execution instance and the batch identifier, fact package identifier, quality rule object version, model version, threshold version, and permission identifier; Before performing at least one quality operation among the following: a one-time quality assessment execution instance request to execute model inference call, automatic release, partial overwrite, archive write, or self-healing reconstruction, the current quality operation is controlled by the controlled execution boundary based on the binding relationship and permission identifier; When the current quality operation meets the control conditions of the controlled execution boundary, a quality assessment of stem cell preparation is performed based on the quality rule object set and the standardized quality fact package to obtain the quality assessment results, risk level or batch disposal decision candidates. When the current quality operation, quality assessment results, risk level, event feedback status, or batch disposal decision candidate does not meet the preset trust conditions, a dynamic isolation record is generated, and at least one security action is performed, including blocking, freezing, permission revocation, manual review, downgrade protection, or self-healing verification. At least one of the following is encapsulated as a standardized quality event: quality assessment results, boundary control results, safety handling results, and archived evidence. A traceable chain of evidence is then formed based on the standardized quality event.

[0005] On one hand, embodiments of this application provide a stem cell preparation quality reliable management device based on quality rule object benchmarks and execution boundary control. This stem cell preparation quality reliable management device includes: The building unit is used to build a set of quality rule objects for quality assessment of stem cell preparation. The set of quality rule objects includes at least one of rule version references, model version references, threshold version references, permission policies, risk policies, archiving field requirements, and quality assessment rules. The acquisition unit is used to acquire multi-source quality data of stem cell preparation batches and generate a standardized quality fact package based on the multi-source quality data. The standardized quality fact package is bound to batch identifier, rule version reference, model version reference and threshold version reference. The processing unit is used to create a one-time quality assessment execution instance for a standardized quality fact package and establish the binding relationship between the one-time quality assessment execution instance and the batch identifier, fact package identifier, quality rule object version, model version, threshold version, and permission identifier. And to control the current quality operation based on binding relationships and permission identifiers through controlled execution boundaries before at least one quality operation in a one-time quality assessment execution instance request to execute model inference invocation, automatic release, partial overwrite, archive write or self-healing reconstruction; And when the current quality operation meets the control conditions of the controlled execution boundary, perform stem cell preparation quality assessment based on the quality rule object set and standardized quality fact package to obtain quality assessment results, risk level or batch disposal decision candidates; And when the current quality operation, quality assessment result, risk level, event feedback status or batch disposal decision candidate does not meet the preset trust conditions, generate a dynamic isolation record and execute at least one of the following security actions: blocking, freezing, permission revocation, manual review, downgrade protection or self-healing verification. And for encapsulating at least one of quality assessment results, boundary control results, safety disposal results, and archived evidence into a standardized quality event, and forming a traceable chain of evidence based on the standardized quality event.

[0006] In one implementation, the standardized quality fact package includes a fact package summary value, a fact package signature value, a data confidence level, a batch identifier, a rule version reference, a model version reference, and a threshold version reference; the standardized quality fact package also includes at least one of the following: fact package identifier, stem cell type, preparation process stage, environmental gating data, equipment calibration status, process step integrity record, microscopic image summary features, flow cytometry summary features, qPCR summary features, sequencing summary features, culture time sequence summary features, batch ontology vector, and required archiving fields; the processing unit is further configured to: Before the standardized quality fact package enters the quality assessment process, a pre-access consistency check is performed. The pre-access consistency check includes at least two of the following: fact package integrity check, fact package summary check, batch identifier consistency check, rule version compatibility check, model version compatibility check, threshold version compatibility check, archived field integrity check, and data confidence check. If the pre-access consistency check fails, the standardized quality fact package is blocked from entering the quality assessment process, and an access failure event or dynamic isolation record is generated.

[0007] In one implementation, the controlled execution boundary performs runtime consistency checks based on normalized quality operation tuples; the normalized quality operation tuples include at least a batch identifier, a standardized quality fact package identifier, a one-time quality assessment execution instance identifier, a rule version reference, a model version reference, a threshold version reference, a permission bitmask, a lifecycle state, an operation type, and a controlled execution boundary identifier; the processing unit is further configured to: Consistency comparison is performed based on normalized quality operation tuples and binding relationships; If the normalized quality operation tuple is inconsistent with the binding relationship, or if the permission bit corresponding to the current operation is not set in the permission bit mask, the current quality operation is blocked.

[0008] In one implementation, the permission identifier includes automatic access permission; the permission identifier also includes a permission bitmask, wherein different permission bits in the permission bitmask correspond to at least one of environment gating calculation permission, batch ontology evaluation permission, model inference call permission, decision routing permission, partial overwrite request permission, archive write permission, and self-healing reconstruction trigger permission; the processing unit is further configured to: When a one-time quality assessment execution instance requests to execute the current operation, a bitwise matching verification is performed based on the permission bitmask and the current operation type. If the permission bit corresponding to the current operation is not set, block the current operation; Controlled execution boundaries are implemented through at least one of the following: API gateway, service broker, container security policy, system call interception, middleware interception, sandbox access control, workflow state machine, message queue topic isolation, or trusted execution environment.

[0009] In one implementation, the quality rule object set includes a global quality rule baseline and local overwrite rules; The global quality rule baseline is used to define the lower limit of quality requirements in the quality assessment of stem cell preparation; Local overwrite rules are used to make controlled adjustments to certain thresholds, weights, feature structures, or model compatibility relationships based on stem cell type, preparation process, equipment conditions, or laboratory node, without exceeding the global quality rule baseline; the processing unit is also used for: Before the local overwrite rule takes effect, an overwrite consistency check is performed. The overwrite consistency check includes checking whether the local overwrite field is an allowed overwrite field, checking whether the local overwrite threshold is lower than the lower limit of the quality requirement, checking whether the version of the rule object after local overwrite is compatible with the model version bound to the one-time quality assessment execution instance, checking whether the feature structure after local overwrite is compatible with the archive traceability rule object, and checking whether the local overwrite request has the corresponding permission bit.

[0010] In one implementation, the global quality rule baseline in the quality rule object set is saved as an immutable rule snapshot; the processing unit is further configured to: When a quality rule object changes, a new version of the quality rule object is generated in a non-overwrite form, and a quality rule benchmark evidence chain is formed based on the rule object summary value, the previous version summary value, the timestamp, the signature, and the version chain relationship. When a quality rule object is updated or reloaded, a candidate rule object or candidate rule version is generated, and signature verification and hash verification are performed; as well as at least one of the following: model compatibility verification, threshold compatibility verification, permission bit compatibility verification, fact package structure compatibility verification, risk policy compatibility verification, consensus policy compatibility verification, and archived field compatibility verification. When the verification passes, the pointer to the current valid rule version is switched atomically. When the verification fails, the currently valid quality rule object remains unchanged, and a standardized quality event for rule reload failure is generated.

[0011] In one implementation, the quality assessment includes at least two of environmental gating assessment, batch ontology assessment, and model reasoning assessment. Among them, the edge feature processing module, model inference module and isolated verification track only output standardized quality facts, candidate anomaly markers, risk labels or candidate disposal suggestions; The formal batch disposal decision is output by the decision-driven layer within the controlled execution boundary after passing runtime consistency verification and multi-source quality adjudication consistency verification. Multi-source quality adjudication consistency verification includes performing consistency verification on at least two of the following before automatic release or archiving of output: rule scoring results, batch ontology scoring results, model inference results, environmental gating results, historical similar batch results, data confidence results, and dynamic isolation records.

[0012] In one implementation, the standardized quality event includes at least one of EnvelopeID, BatchID, FactPackID, TaskID, TraceID, EvidenceID, RuleVersionRef, ModelVersionRef, ThresholdVersionRef, InstanceID, OperationType, EventType, EventPayload, EventState, Timestamp, PrevEnvelopeHash, EnvelopeHash, and Signature; the return status of the standardized quality event includes pending, sending, sent, and failed_permanent; the processing unit is further configured to: When a standardized quality event fails to be returned and the maximum number of retries is reached, the return status of the standardized quality event is switched to failed_permanent. When a failed_permanent status event exists that is associated with a target batch, target standardized quality fact package, or target one-time quality assessment execution instance, block automatic release, archiving completion confirmation, or rule version switching operations that depend on the standardized quality event.

[0013] Accordingly, this application provides a computer device comprising: Memory, which stores computer programs; The processor is used to load computer programs to implement the above-mentioned reliable quality control method for stem cell preparation based on quality rule object benchmarks and execution boundary control.

[0014] Accordingly, this application provides a computer-readable storage medium storing a computer program adapted to be loaded by a processor and executed by the aforementioned method for reliable quality control of stem cell preparation based on quality rule object benchmarks and execution boundary control.

[0015] In this embodiment, a quality rule object set for stem cell preparation quality assessment is constructed, multi-source quality data of stem cell preparation batches are acquired, and a standardized quality fact package is generated based on the multi-source quality data. The standardized quality fact package is bound to batch identifiers, rule version references, model version references, and threshold version references. A one-time quality assessment execution instance is created for the standardized quality fact package, and a binding relationship is established between the one-time quality assessment execution instance and the batch identifier, fact package identifier, quality rule object version, model version, threshold version, and permission identifier. Before the one-time quality assessment execution instance requests to execute at least one quality operation among model inference call, automatic release, partial overwrite, archive write, or self-healing reconstruction, the binding is based on the controlled execution boundary. Relationships and permission identifiers control the current quality operation. When the current quality operation meets the control conditions of the controlled execution boundary, a stem cell preparation quality assessment is performed based on the quality rule object set and standardized quality fact package, yielding the quality assessment result, risk level, or batch disposal decision candidate. When the current quality operation, quality assessment result, risk level, event feedback status, or batch disposal decision candidate does not meet the preset trust conditions, a dynamic isolation record is generated, and at least one of the following safety actions is executed: blocking, freezing, permission revocation, manual review, downgrade protection, or self-healing verification. At least one of the quality assessment result, boundary control result, safety disposal result, and archived evidence is encapsulated into a standardized quality event, and a traceable evidence chain is formed based on the standardized quality event. It is evident that by using unified rule benchmarks, standardized fact packages, controlled execution instances, refined permission verification, dynamic isolation, and a trustworthy self-healing mechanism, problems such as rule drift, fact conflicts, unauthorized operations, silent event loss, and accidental release of high-risk batches are effectively solved, achieving a trustworthy, consistent, and auditable closed-loop system for stem cell preparation quality assessment across the entire chain. With unified rule benchmarks and controlled execution boundaries as the core, a full-chain trusted management and control mechanism is established, encompassing rule construction, fact input, instance binding, operation verification, dynamic isolation, and evidence chain archiving. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 A flowchart illustrating a reliable quality control method for stem cell preparation based on quality rule object benchmarks and execution boundary control, provided for an embodiment of this application; Figure 2 A controlled execution boundary architecture diagram provided for embodiments of this application; Figure 3 A schematic diagram illustrating the working principle of an embodiment of this application; Figure 4 An architecture diagram of a stem cell preparation quality trust management system based on quality rule object benchmarks and execution boundary control is provided for embodiments of this application; Figure 5 A schematic diagram of a stem cell preparation quality reliable management device based on quality rule object benchmark and execution boundary control provided in this application embodiment; Figure 6 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0018] It should be noted in advance that, in order to enable those skilled in the art to better understand the technical solutions proposed in the embodiments of this application, the embodiments of this application will be described clearly and completely in conjunction with one or more accompanying drawings. Furthermore, the various drawings shown in the embodiments of this application are merely illustrative examples; for example, the execution order of each step in the drawings can be adaptively adjusted according to the actual application scenario. In addition, in the embodiments of this application, the block diagrams shown in the various drawings are merely functional entities and do not necessarily correspond to physically independent entities. That is, these functional entities can be implemented in software, or in one or more hardware modules or integrated circuits, or in different network and / or processor devices and / or microcontroller devices.

[0019] In the embodiments of this application, the terms "module" or "unit" refer to a computer program or part of a computer program that has a predetermined function and works with other related parts to achieve a predetermined goal, and can be implemented wholly or partially using software, hardware (such as processing circuitry or memory), or a combination thereof. Similarly, a processor (or multiple processors or memory) can be used to implement one or more modules or units. Furthermore, each module or unit can be part of an overall module or unit that includes the functionality of that module or unit.

[0020] The block diagrams shown in the accompanying drawings are merely functional entities and do not necessarily correspond to physically independent entities. That is, these functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor devices and / or microcontroller devices.

[0021] The flowcharts shown in the accompanying drawings are merely illustrative and do not necessarily include all content and operations / steps, nor do they necessarily have to be performed in the described order. For example, some operations / steps can be broken down, while others can be combined or partially combined; therefore, the actual execution order may change depending on the specific circumstances.

[0022] It should be noted that "multiple" in this article refers to two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone. The character " / " generally indicates that the preceding and following related objects have an "or" relationship.

[0023] Terms and definitions used in this application: Quality rule object: refers to a structured rule unit used to constrain the quality assessment behavior of stem cell preparation. It includes at least rule identifier, rule type, version information, threshold set, model reference, permission scope, permission bitmask, verification strategy, archive structure and signature information.

[0024] Standardized quality fact package: refers to an input benchmark object that uniformly encapsulates the batch identifier, preparation process data, quality testing data, marginal summary features, rule version references, model version references, threshold version references, and archive fields of the stem cell batch to be evaluated.

[0025] Normalized quality operation tuple: refers to a structured validation tuple used to describe a critical quality assessment operation, which includes at least batch identifier, fact package identifier, instance identifier, rule version reference, model version reference, threshold version reference, permission bitmask, lifecycle state, operation type, and controlled execution boundary identifier.

[0026] Standardized quality events (envelope): refers to a structured event encapsulation object used to carry quality facts, quality assessment results, runtime verification results, anomaly self-healing events, dynamic isolation records, rule candidate object verification results, and archived evidence.

[0027] TaskID: Refers to the task identifier used to associate a single stem cell batch quality assessment task.

[0028] TraceID: Refers to the trace identifier used throughout the entire process of pre-access consistency verification, instance registration and binding, controlled execution boundary verification, three-layer quality assessment, risk classification, multi-source adjudication, self-healing reconstruction, rule candidate object verification, and archiving operations.

[0029] EvidenceID: Refers to an evidence identifier used to associate tamper-proof archived evidence, supporting audits, dispute evidence presentation, and quality review.

[0030] Dynamic isolation records refer to isolation control records generated when runtime consistency checks fail, the state is unverifiable, operations go out of bounds, consensus is insufficient, or the envelope enters the failed_permanent state. These records are used to enable subsequent control points to inherit blocking, freezing, manual review, or self-healing strategies.

[0031] Candidate anomaly flags: These are non-final risk warnings generated by the edge feature processing module, model inference module, or isolated verification track based on quality facts, and do not constitute formal batch handling decisions.

[0032] Formal batch handling decision: refers to the final quality handling result output by the decision-driven layer within the controlled execution boundary after completing runtime consistency verification, multi-source quality adjudication consistency verification, permission bit verification, dynamic isolation record verification, and archived field integrity verification.

[0033] The "failed_permanent" status indicates a permanent failure state for a standardized quality event envelope after reaching the maximum number of retries without completing the return or archiving confirmation. Envelopes in this status must not be used as the basis for automatic release, archiving completion confirmation, or rule version switching.

[0034] Please see Figure 1 , Figure 1This document presents a flowchart illustrating a method for reliable quality control of stem cell preparation based on quality rule object benchmarks and execution boundary control, as provided in an embodiment of this application. This method can be applied to computer devices, such as smartphones (Android phones, iOS phones, HarmonyOS phones, etc.), tablets, desktop computers, portable personal computers, mobile internet devices (MIDs), smart interactive devices, smart home appliances, in-vehicle terminals, wearable devices, and other terminal devices. It can also be applied to servers or cloud server clusters. This method aims to reduce state contamination and invalid generation, and to generate traceable structured data objects. Figure 1 As shown, the method may include the following steps S101 to S107.

[0035] Step S101: Construct a set of quality rule objects for quality assessment of stem cell preparation.

[0036] In one implementation, the computer device pre-constructs a set of quality rule objects for stem cell preparation quality assessment, serving as a unified control benchmark for the entire quality management system. This set of rule objects may include at least one of the following: rule version references, model version references, threshold version references, permission policies, risk policies, archiving field requirements, and quality assessment rules. For example, the quality rule object set can be subdivided into multiple categories such as environmental gating rule objects, batch ontology assessment rule objects, and decision routing rule objects, with each category defining specific thresholds, weights, model references, and permission requirements.

[0037] In another implementation, the quality rule object set includes a global quality rule baseline and local overwrite rules. The global quality rule baseline is used to define the lower limit of quality requirements in stem cell preparation quality assessment. Local overwrite rules are used to make controlled adjustments to certain thresholds, weights, feature structures, or model compatibility relationships based on stem cell type, preparation process, equipment conditions, or laboratory node, without exceeding the global quality rule baseline. The computer device can also perform overwrite consistency verification before the local overwrite rules take effect; wherein, the overwrite consistency verification includes verifying whether the locally overwritten field is an allowed overwrite field, verifying whether the locally overwritten threshold is lower than the lower limit of quality requirements, verifying whether the version of the rule object after local overwriting is compatible with the model version bound to the one-time quality assessment execution instance, verifying whether the feature structure after local overwriting is compatible with the archived traceability rule object, and verifying whether the local overwrite request has at least one of the following:

[0038] In another implementation, the global quality rule baseline in the quality rule object set is stored as an immutable rule snapshot. When a quality rule object changes, the computer device generates a new version of the quality rule object in a non-overwrite manner, and forms a quality rule baseline evidence chain based on the rule object digest value, the previous version digest value, the timestamp, the signature, and the version chain relationship. Furthermore, a hash chain can be used to ensure version immutability. When a quality rule object is updated or reloaded, the computer device generates candidate rule objects or candidate rule versions and performs signature verification, hash verification, and at least one of the following: model compatibility verification, threshold compatibility verification, permission bit compatibility verification, fact package structure compatibility verification, risk policy compatibility verification, consensus policy compatibility verification, and archived field compatibility verification. When the verification passes, the computer device atomically switches the pointer to the currently valid rule version. When the verification fails, the computer device keeps the currently valid quality rule object unchanged and generates a standardized quality event indicating a rule reload failure.

[0039] In one embodiment, the set of quality rule objects is shown in Table 1: Table 1

[0040] As shown in Table 1, the quality rule object set comprehensively covers all key aspects of the stem cell preparation quality assessment process. Among them, the environmental gating rule object is used to define the culture environment thresholds and non-assessable conditions; the batch ontology assessment rule object clarifies D / P / I / G. n The scoring weights and sub-interception thresholds for core indicators such as / S / C are defined; the decision-making and diversion rule object sets the decision boundaries for automatic release, re-inspection, manual review, and scrapping; the edge feature processing rule object standardizes the summary feature structure, desensitization strategy, and minimum return rules; the model inference rule object manages the version, input and output structure, and calling permissions of the graph model; the self-healing processing rule object formulates isolation, permission revocation, and reconstruction strategies for anomaly types; the risk classification rule object determines the risk level based on the calculation factor and selects the corresponding assessment path and release restrictions; the consensus adjudication rule object coordinates the sources of multi-source adjudication, consensus conditions, and downgrade protection measures. The dynamic isolation rule object records the failure subject and failure type, and specifies the isolation strategy and inheritance rules; the event chain backflow rule object realizes event tracking and backflow through chained summaries, pre-order references, and backflow indexes. Event chain backflow can be implemented using pre-order hash references, timestamp signatures, evidence identifier indexes, or tamper-proof storage structures; the rule evolution rule object constrains the generation conditions of candidate rules, historical replay verification, and gray-scale verification processes; the archive traceability rule object specifies the required archive fields, hash verification, timestamp signatures, and storage strategies, thereby providing a structured, versioned, and traceable unified rule benchmark for the reliable control of stem cell preparation quality.

[0041] It is evident that by centrally managing and objectifying the rules scattered across various modules, the problem of rule drift caused by scattered rule configurations can be effectively avoided, ensuring that all evaluation stages follow the same set of rule benchmarks.

[0042] S102. Obtain multi-source quality data for stem cell preparation batches and generate a standardized quality fact package based on the multi-source quality data.

[0043] In one implementation, upon receiving a batch of stem cell preparation to be evaluated, a computer device acquires multi-source quality data for that batch (including but not limited to environmental gating data, microscopic image summarization features, flow cytometry summarization features, qPCR summarization features, culture time-series data, etc.), and generates a standardized quality fact package based on this data. This standardized quality fact package is bound to a batch identifier, rule version reference, model version reference, and threshold version reference.

[0044] For example, suppose the computer device receives the following batch of stem cell preparations to be evaluated: BatchID: Batch-iPSC-042CellType: iPSCProcessStage:DirectedDifferentiation-D7RuleVersion: RuleSet-iPSC-v2.1.0ModelVersion: GNN-iPSC-v4.0ThresholdVersion: th-iPSC-v2.1.0EnvironmentGateData: EnvVector-042FeatureSchema: iPSC-FEATURE-v3.2OntologyVector: [D,P,I,Gn,S,C]DataConfidence:0.93ArchiveRequiredFields: Complete The above content indicates that the stem cell preparation batch to be evaluated is Batch-iPSC-042, the currently valid rule version is RuleSet-iPSC-v2.1.0, and the model version is GNN-iPSC-v4.0.

[0045] Furthermore, the computer equipment encapsulates this information, along with the specific detection data, into a structured fact package QFP-iPSC-042. This fact package contains key fields such as fact package identifier, batch identifier, environment gating data vector, and batch ontology vector, and generates a fact package digest and signature through hash calculation. Specifically, it can be represented as: FactPackID: QFP-iPSC-042FactHash: Hash_QFP_iPSC_042FactSignature:Sig_QFP_iPSC_042 In another implementation, the standardized quality fact package includes a fact package summary value, a fact package signature value, data confidence level, batch identifier, rule version reference, model version reference, and threshold version reference. The standardized quality fact package also includes at least one of the following: fact package identifier, stem cell type, preparation process stage, environmental gating data, equipment calibration status, process step integrity record, microscopic image summary features, flow cytometry summary features, qPCR summary features, sequencing summary features, culture time sequence summary features, batch ontology vector, and required archiving fields. Before the standardized quality fact package enters the quality assessment process, the computer device performs a pre-access consistency check; this pre-access consistency check includes at least two of the following: fact package integrity check, fact package summary check, batch identifier consistency check, rule version compatibility check, model version compatibility check, threshold version compatibility check, archived field integrity check, and data confidence level check. When the pre-access consistency check fails, the computer device blocks the standardized quality fact package from entering the quality assessment process and generates an access failure event or dynamic isolation record.

[0046] As can be seen from the above, the standardized quality fact package, as the sole input benchmark for all subsequent evaluation stages, can ensure that different modules such as environmental gating, model inference, and manual review are all based on the exact same data and version configuration, thereby eliminating factual conflicts caused by inconsistent inputs.

[0047] S103. Create a one-time quality assessment execution instance for the standardized quality fact package, and establish the binding relationship between the one-time quality assessment execution instance and the batch identifier, fact package identifier, quality rule object version, model version, threshold version, and permission identifier.

[0048] In one implementation, the computer device creates a one-time quality assessment execution instance. This instance has a unique instance identifier, a lifecycle control handle, and a controlled execution boundary identifier. Furthermore, the computer device establishes bindings between this instance and batch identifiers, fact package identifiers, quality rule object versions, model versions, threshold versions, and permission identifiers.

[0049] For example, suppose a one-time quality assessment execution instance created by a computer device is as follows: nstanceID: EvalInst-MSC-001 Instance description digest signature value: Sig_EvalInst_MSC_001 Permission identifier: Perm_MSC_Eval_Release_Archive Permission bitmask: 01011111 Lifecycle management handle: LifeHandle_MSC_001 Controlled execution boundary identifier: Boundary_MSC_001 This indicates that for the stem cell preparation batch Batch-MSC-001, the computer device creates an instance EvalInst-MSC-001 with the instance description summary signature value Sig_EvalInst_MSC_001, the permission identifier Perm_MSC_Eval_Release_Archive, the permission bitmask 01011111, the lifecycle management handle LifeHandle_MSC_001, and the controlled execution boundary identifier Boundary_MSC_001.

[0050] Specifically, a one-time quality assessment execution instance can be a container instance, a virtual machine instance, a sandbox process, a Serverless function instance, a one-time task node in a workflow engine, or an isolated execution context in an edge gateway.

[0051] Furthermore, the computer device binds it to the fact package QFP-MSC-001, the rule version RuleSet-MSC-v1.3.0, the model GNN-MSC-v3.2, and a permission bitmask (such as 01011111) representing its permission scope. Specifically, this can be represented as: Batch-MSC-001 QFP-MSC-001 EvalInst-MSC-001 Sig_EvalInst_MSC_001 PermissionMask: 01011111 LifeHandle_MSC_001 Boundary_MSC_001 RuleSet-MSC-v1.3.0 GNN-MSC-v3.2 ThresholdSet-th-v1.3.0 In another implementation, the permission identifier includes automatic access permission; the permission identifier also includes a permission bitmask, where different permission bits in the permission bitmask correspond to at least one of the following: environment gating computing permission, batch ontology evaluation permission, model inference call permission, decision routing permission, partial overwrite request permission, archive write permission, and self-healing reconstruction trigger permission. When a one-time quality assessment execution instance requests to execute the current operation, the computer device performs bit-by-bit matching verification between the permission bitmask and the current operation type. When the permission bit corresponding to the current operation is not set, the computer device blocks the current operation.

[0052] It should be noted that "one-time use" and "binding" are the key features of the stem cell preparation quality trust management method provided in this application, which is based on quality rule object benchmarks and execution boundary control. The lifecycle of this instance is limited to the evaluation task of the current batch; it is terminated or archived after the task is completed and cannot be reused in other batches. Furthermore, because all its operations are strongly associated with its bound rules, models, and permissions, batch context pollution and permission abuse can be effectively prevented.

[0053] S104. Before any quality operation, such as a one-time quality assessment execution instance request execution model inference call, automatic release, partial overwrite, archive write, or self-healing reconstruction, the current quality operation is controlled by the controlled execution boundary based on the binding relationship and permission identifier.

[0054] In one implementation, before the computer device performs critical quality operations such as one-time quality assessment execution instance request execution model inference calls, automatic release, partial overwrite, archive write, or self-healing reconstruction, a controlled execution boundary intervenes and controls the current quality operation based on the binding relationship and permission identifier established in step S103. The controlled execution boundary can be implemented through at least one of API gateway, service mesh, permission token verification, sandbox access control, workflow state machine, message queue topic isolation, system call interception, trusted execution environment, or rule object signature verification.

[0055] Regarding automatic release, assuming the permission bit of a one-time quality assessment execution instance is 00001111, it indicates that the one-time quality assessment execution instance has environment gating calculation permission, batch ontology assessment permission, model inference call permission, and decision routing permission. However, the automatic release permission bit is 00010000. When the one-time quality assessment execution instance requests automatic release, the controlled execution boundary detection permission bit mismatch directly blocks the current operation and generates an out-of-bounds blocking event, such as "EventType: PermissionMaskDeniedOperation: AutoReleaseRequiredBit:00010000ActualMask: 00001111Result: Blocked".

[0056] In another implementation, the controlled execution boundary performs runtime consistency checks based on normalized quality operation tuples. These normalized quality operation tuples include at least a batch identifier, a standardized quality fact package identifier, a one-time quality assessment execution instance identifier, a rule version reference, a model version reference, a threshold version reference, a permission bitmask, a lifecycle state, an operation type, and a controlled execution boundary identifier. The computer device performs a consistency comparison between the normalized quality operation tuples and the binding relationships. When the normalized quality operation tuples and the binding relationships are inconsistent, or when the permission bit corresponding to the current operation is not set in the permission bitmask, the computer device blocks the current quality operation.

[0057] In one embodiment, to achieve fine-grained runtime verification, this application introduces a normalized quality operation tuple. This normalized quality operation tuple is a structured object; for example, it can be represented as: QOT = <BatchID, FactPackID, InstanceID, RuleVersionRef,ModelVersionRef, ThresholdVersionRef, PermissionMask, LifecycleState,OperationType, BoundaryID> .

[0058] When an instance requests to perform an operation (such as model inference), the computer device generates a normalized quality operation tuple for that operation and performs a consistency comparison with the binding relationships stored in S103.

[0059] For example, suppose an instance is bound to the model version GNN-MSC-v3.2, but its model inference request references GNN-MSC-v2.0, then the tuple comparison will fail. Alternatively, an instance's permission bitmask is 00001111 (lacking auto-allow permission), but its request to perform an auto-allow operation will also fail permission bit verification. In either case, the controlled execution boundary will block the current operation and generate an out-of-bounds blocking event. Conversely, if all verifications pass, the operation is allowed to continue. Controlled execution boundaries can be implemented through various methods such as API gateways, service meshes, and sandboxes.

[0060] As can be seen, by setting up a verification checkpoint on the operation execution path, it can be ensured that every critical operation is performed under the correct context, the correct version, and the correct authorization, thereby effectively intercepting risks such as model misuse, rule mismatch, and permission overreach.

[0061] S105. When the current quality operation meets the control conditions of the controlled execution boundary, perform a stem cell preparation quality assessment based on the quality rule object set and the standardized quality fact package to obtain the quality assessment results, risk level, or batch disposal decision candidates.

[0062] In one implementation, the computer device will only perform a stem cell preparation quality assessment based on the quality rule object set and standardized quality fact package after the current quality operation has passed the control conditions of the controlled execution boundary in S104 (i.e., the runtime consistency check has passed). The assessment process may include multiple layers such as an environment gating layer, a batch ontology layer, and a decision-driven layer.

[0063] Figure 2 This is a controlled execution boundary architecture diagram provided for an embodiment of this application. For example... Figure 2 As shown, the controlled execution boundary framework includes an environment gating layer, a batch ontology layer, and a decision-driven layer. The environment gating layer can calculate a gating feasibility score G (e.g., G=ω1E1+ω2E2+ω3E3+ω4E4). If G is higher than the threshold defined in the rules (e.g., 0.72), it passes. The batch ontology layer constructs and evaluates an ontology vector B (e.g., B=[D,P,I,Gn,S,C]) containing multiple quality indicators (e.g., cell viability D, purity P, etc.), and calculates the total ontology score Bs (e.g., Bs=αD+βP+γI+δGn+εS+ζC). If the total ontology score is higher than the threshold and each item meets the pre-threshold, the ontology evaluation passes. If at least one item is lower than the pre-threshold (e.g., P is lower than the undifferentiated marker residual safety threshold), then even if the total score is high, the item pre-interception is triggered. The decision-driven layer further calculates a credible treatment score A based on the evaluation results (e.g., A=μ1(1 R)+μ2(1 V)+μ3(1 K)+μ4Q). If A reaches the automatic release candidate threshold, it indicates that it has entered the automatic release candidate state. Risk labels (e.g., MediumRisk), candidate anomaly flags, etc., are used as "candidate results" or "non-final decisions" and are not directly used for the final batch release. Formal handling decisions must be output by the decision-driven layer within the controlled execution boundary after passing the following checks: rule version consistency check, fact package consistency check, permission bit check, lifecycle status check, risk level path check, multi-source adjudication consistency check, and archived field integrity check.

[0064] Taking risk level path verification as an example, suppose a certain batch of stem cell preparation has the following risk factors: high environmental fluctuations, model confidence Q=0.71, undifferentiated marker residue close to the threshold, a previous scrapping record among similar historical batches, a previous self-healing reconstruction of this batch, and a previous envelope backhaul failure record. Based on these risk factors, the computer equipment determines the batch risk level to be L3. In this case, even if the credible disposal score A reaches the automatic release candidate threshold, the computer equipment still selects the three-layer quality assessment + multi-source review + manual review gating path and temporarily does not grant automatic release permission.

[0065] Taking multi-source adjudication consistency verification as an example, suppose the multi-source adjudication results before the release of a certain batch of stem cell preparation are as follows:

[0066] If the computer determines that the multi-source decision has not reached a consensus (not all decision conditions are met), it will enter a downgrade protection mode, such as prohibiting automatic release, revoking the automatic release permission bit, freezing the release lock, triggering manual review, requiring supplementary functional tests, generating dynamic isolation records, and recording consensus deficiency events.

[0067] In another implementation, the quality assessment includes at least two of environmental gating assessment, batch ontology assessment, and model inference assessment; wherein the edge feature processing module, model inference module, and isolation verification track only output standardized quality facts, candidate anomaly flags, risk labels, or candidate disposal suggestions. Formal batch disposal decisions are output by the decision-driven layer within the controlled execution boundary after passing runtime consistency checks and multi-source quality adjudication consistency checks; wherein the multi-source quality adjudication consistency check includes performing consistency checks on at least two of the following before automatic release or archiving confirmation: rule scoring results, batch ontology scoring results, model inference results, environmental gating results, historical similar batch results, data confidence results, and dynamic isolation records. In one implementation, when making an automatic release decision, the quality assessment simultaneously includes environmental gating assessment, batch ontology assessment, and model inference assessment; if any one of these three assessments fails, automatic release is prohibited, and the system enters a degraded protection mode.

[0068] It is evident that stratified assessment allows for multi-dimensional and progressive analysis of stem cell quality. Defining all intermediate results as "candidates" preserves control over the final decision-making process for subsequent multi-source adjudication and safety verification, preventing misjudgments at a single stage from directly leading to erroneous releases.

[0069] S106. When the current quality operation, quality assessment result, risk level, event feedback status, or batch disposal decision candidate does not meet the preset trust conditions, generate a dynamic isolation record and execute at least one of the following security actions: blocking, freezing, permission revocation, manual review, downgrade protection, or self-healing verification.

[0070] In one implementation, a security handling mechanism is triggered when the computer device detects any situation that does not meet preset trust conditions. These situations that do not meet preset trust conditions may include, but are not limited to: the operation in S104 being blocked, the quality assessment result or risk level generated in S105 being too high, the failure to transmit critical events (such as standardized quality events), or the batch handling decision candidates failing to reach a consensus in multi-source verification.

[0071] In one embodiment, when a standardized quality event fails to be retried and the maximum number of retries is reached, the computer device switches the retried status of the standardized quality event to failed_permanent. When a failed_permanent status event is associated with a target batch, a target standardized quality fact package, or a target one-off quality assessment execution instance, the computer device blocks automatic release, archiving completion confirmation, or rule version switching operations that depend on the standardized quality event.

[0072] Taking standardized quality events as an example, the event's feedback status can include pending, sending, sent, and failed_permanent as the final failure state. After three consecutive failed feedback attempts: the first attempt: pending → sending → failed → pending; the second attempt: pending → sending → failed → pending; the third attempt: pending → sending → failed_permanent. The third final failure (failed_permanent) will trigger a safety handling mechanism.

[0073] If a security handling mechanism is triggered, the computer device generates a dynamic isolation record. This record details the failure subject (e.g., BatchID, InstanceID), failure type, number of failures, associated rule version, and final handling status (e.g., AutoReleaseBlocked). This record is then written back to the quality rule object set or isolation policy index. The dynamic isolation record can be stored in the rule object set, isolation policy index, audit log system, or quality management system.

[0074] In one implementation, the event state can be obtained through a lifecycle state machine. The lifecycle states of a one-time quality assessment execution instance include: pending registration → bound → running → pending archiving → archived → terminated. When an exception occurs, the state transitions are: running → frozen → cleanup → isolation verification → rebuilt → running → pending archiving → archived → terminated. When rule candidate object verification occurs, the state transitions are: candidate generation → historical playback verification → grayscale consistency verification → authorization confirmation → atomic switch or verification failure archiving. The state of the standardized quality event envelope includes: failed_permanent, where failed_permanent is the final failure state, indicating that the event has reached the maximum number of retries but has not yet completed the return or archiving confirmation. The system must not use it as the basis for automatic release, archiving completion confirmation, or rule version switching.

[0075] In addition, the computer device will perform at least one security action, which includes, but is not limited to: blocking the current operation, freezing the instance's subsequent operation permissions, revoking its automatic release permissions, transferring the task to manual review, initiating a downgrade protection mode (such as disabling automatic release and requiring supplementary testing), or triggering a dual-track self-healing verification process.

[0076] For example, when a standardized quality event envelope fails to send back data consecutively up to the maximum number of retries (e.g., 3 times), its status changes to failed_permanent. The computer device then generates a dynamic isolation record and blocks all automatic release or archiving confirmation operations that depend on that envelope. As another example, when multi-source adjudication fails to reach a consensus, the computer device enters a degraded protection mode, disabling automatic release and requiring manual intervention.

[0077] Furthermore, dynamic isolation records allow the handling strategy for a single failure to be inherited by subsequent evaluation instances. For example, when the same batch requests automatic release again, the boundary can be frozen directly based on the record, achieving adaptive security protection.

[0078] S107. Encapsulate at least one of the following: quality assessment results, boundary control results, safety disposal results, and archived evidence into a standardized quality event, and form a traceable chain of evidence based on the standardized quality event.

[0079] In one implementation, the computer device encapsulates key operations and results into standardized quality events (envelopes). Each standardized quality event is a structured data object that may include at least one of the following: EnvelopeID, BatchID, FactPackID, TaskID, TraceID, EvidenceID, RuleVersionRef, ModelVersionRef, ThresholdVersionRef, InstanceID, OperationType, EventType, EventPayload, EventState, Timestamp, PrevEnvelopeHash, EnvelopeHash, and Signature. Specifically, TaskID is used to associate a complete quality assessment task, TraceID is used to chain operations from access verification to archiving, and EvidenceID is used to associate the final, immutable archiving evidence. These event envelopes can be organized and reflowed according to a chained digest structure (i.e., each envelope records the hash value of the previous envelope, PrevEnvelopeHash). Each event record includes: current event summary, previous event summary reference, rule object version summary, standardized quality fact package summary, instance identifier reference, time-ordered index, EvidenceID, and TraceID.

[0080] Standardized quality events can be transmitted and archived through message queues, event buses, audit log systems, electronic batch recording systems, or quality management systems.

[0081] By encapsulating standardized events, fragmented information (such as quality assessment results, boundary control results, security handling results, and archived evidence) scattered across different modules and time points can be integrated into a unified, traceable data object. The chained summary structure and the failed_permanent final state together ensure the integrity and loss prevention of the evidence chain, enabling the complete process of any quality assessment to be accurately audited and fully reproduced afterward.

[0082] Figure 3 This is a schematic diagram illustrating the working principle of an embodiment of this application. Figure 3As shown, firstly, the computer equipment constructs a unified set of quality rule objects as the benchmark for the entire evaluation system. For each batch to be evaluated, its multi-source quality data is encapsulated into a standardized quality fact package, and a pre-access consistency check is performed before entering the evaluation process to ensure the integrity and version compatibility of the fact package. After the check passes, a one-time quality evaluation execution instance is created for the fact package, and a binding relationship is established between the instance and the fact package, rule version, model version, etc., i.e., instance-fact package-rule binding. Before each subsequent critical operation, the computer equipment generates a normalized quality operation tuple, and performs runtime consistency check on the current operation by combining permission bitmask verification and controlled execution boundary. After the check passes, the three-layer quality evaluation (environment gating layer, batch ontology layer, and decision-driven layer) is entered, and a risk level path is selected according to the risk level of the batch to determine the differentiated evaluation intensity. After completing the preliminary evaluation, the computer equipment performs a multi-source adjudication consistency check, integrating the adjudication results from multiple sources. If the check fails or an abnormal condition is triggered, corresponding security actions are executed, including generating dynamic isolation records, triggering a failure retry state machine, and initiating dual-track self-healing verification. All key events and results generated throughout the process are encapsulated as standardized quality events and fed back to the rule object set in summary order through a chained event feedback mechanism for subsequent verification of quality rule candidate objects. Finally, the computer device integrates the entire chain of information, including rules, fact packages, events, and operations, to form a rule-fact package-event-operation evidence chain archive, thereby achieving a complete and reliable closed loop from rule benchmark to final archive.

[0083] This application embodiment constructs a fully trusted closed loop through the synergistic effect of steps S101 to S107, encompassing rule definition, data input, instance isolation, operation verification, hierarchical evaluation, dynamic isolation, and evidence archiving. This closed loop not only solves single-point problems such as rule drift, factual conflict, and unauthorized operation in the prior art, but also improves the overall credibility, security, and auditability of stem cell preparation quality control.

[0084] The foregoing has described the method and flow of this application in detail. To further illustrate the robustness of this solution in dealing with complex situations, the technical solution of this application is supplemented below with several specific application scenarios.

[0085] Scenario 1: Rule Version Changes and Controlled Evolution When quality rules need to be updated, the computer device does not directly overwrite existing rules. Instead, it first generates a candidate rule object (e.g., RuleSet-MSC-v1.4.0-candidate). This candidate rule must undergo a series of compatibility checks in an isolated verification environment, including model compatibility, threshold compatibility, and permission bit compatibility. If the checks pass, the computer device generates a new quality access baseline and, at an atomic switch point, points the valid rule version pointer to the new rule; the old rule is then marked as "audit-only." If the checks find that the model referenced by the candidate rule has not yet completed validity registration, the computer device will refuse to switch and generate a rule reload failure event. This mechanism ensures the security and controllability of rule updates, avoiding large-scale quality control failures caused by incorrect rule implementation.

[0086] Scenario 2: Anomaly Recovery and Dual-Track Self-Healing Verification Suppose an evaluation instance, EvalInst-MSC-015, incorrectly references an older rule version during runtime. The computer device immediately blocks its output, revokes relevant permissions, and transitions its instance state to "frozen." Simultaneously, the computer device generates a dynamic isolation record. To recover this batch of quality evaluations, the computer device initiates a dual-track self-healing verification process: first, a read-only verification instance is created in the isolation verification track to snapshot and clean up residual objects (such as caches and temporary tokens) from the aberrant instance; after successful verification, a completely new instance is created in the production evaluation track, and the evaluation is re-executed. This mechanism of separating the production evaluation track from the isolation verification track ensures that the anomaly recovery process does not contaminate production data, achieving reliable self-healing reconstruction.

[0087] Scenario 3: Quality Preservation Backhaul in Edge Computing Scenarios When collecting data at edge nodes, to reduce bandwidth and storage pressure, computer devices do not upload large raw files (such as 96MB microscopic images). Instead, they process objects according to edge feature rules, extracting key summary features such as cell density, fusion degree, and proportion of necrotic areas, with a total size of only a few KB. These summary features are packaged into standardized quality fact packages and uploaded to the central platform for evaluation. This approach significantly reduces data transmission volume while retaining the core information required for quality assessment, achieving minimal backhaul for "quality preservation."

[0088] Scenario 4: End-to-End Audit and Dispute Evidence When a dispute arises regarding the final quality conclusion of a batch, the evidence chain of this application can be used for admissibility. Auditors can retrieve the complete ArchiveIndex from the archiving computer device based on the batch ID. This index allows for the chaining of complete TaskIDs, TraceIDs, and EvidenceIDs, tracing back to the envelope chain of all standardized quality events corresponding to that batch, the original standardized quality fact package, the chain of rule versions used (including immutable rule snapshots), model versions, permission mask verification records, dynamic isolation records, and the multi-source decision consistency verification results at the time of the final decision. This comprehensive archiving capability provides objective evidence for the quality conclusion.

[0089] As can be seen from the above, the stem cell preparation quality trust management method based on quality rule object benchmarks and execution boundary control provided in this application can effectively improve the trustworthiness, consistency, traceability, abnormal recovery safety and evidence chain integrity of stem cell batch quality assessment results. It is applicable to MSC, iPSC, ESC, immune cell induced differentiation products, organoid cell preparation and other cell preparation scenarios that require strict quality gating.

[0090] Figure 4 This is an architecture diagram of a stem cell preparation quality trust management system based on quality rule object benchmarks and execution boundary control, provided as an embodiment of this application. Figure 4As shown, firstly, multi-source quality data for stem cell preparation provided by external data sources enters the stem cell preparation quality trust management system based on quality rule object benchmarks and execution boundary control. The quality rule object management module, acting as the rule foundation, is responsible for maintaining metadata such as rule versions, permissions, policies, and evaluation rules, providing a unified rule benchmark for the entire evaluation process. The standardized quality fact package generation module receives the multi-source quality data and binds it with batch IDs, rule versions, model versions, and threshold versions to form a standardized quality fact package. Subsequently, the execution instance management module creates a one-time evaluation instance for this fact package and completes the binding of all dimensions of identifiers (such as batch, fact package, rule, model, threshold, permission, etc.). Next, the controlled execution boundary module controls various quality operations (such as model inference, automatic release, etc.) based on the permission identifiers and the established binding relationships. If the verification passes, the system proceeds to the quality assessment and decision-making module, which outputs the assessment results, risk level, and proposed actions. If the current state does not meet the trust conditions (e.g., unauthorized operation, event feedback failure), the system switches to the dynamic isolation and security handling module, which performs security actions such as blocking, freezing, and manual review, and generates a dynamic isolation record. Finally, both the normal assessment results from the quality assessment and decision-making module and the handling results output by the dynamic isolation and security handling module are integrated into the event feedback and evidence archiving module. This module is responsible for encapsulating various results into standardized quality events and building a traceable chain of evidence, thus forming a complete and trustworthy closed loop from data access, rule constraints, instance isolation, boundary verification, assessment and decision-making, anomaly handling to evidence archiving.

[0091] The methods of the embodiments of this application have been described in detail above. In order to facilitate better implementation of the above solutions of the embodiments of this application, the apparatus of the embodiments of this application is provided below.

[0092] Please see Figure 5 , Figure 5 This application provides a schematic diagram of the structure of a stem cell preparation quality reliable management device based on quality rule object benchmarks and execution boundary control, as an embodiment of the present application. Figure 5 The stem cell preparation quality trust management device shown, based on quality rule object benchmarks and execution boundary control, can be mounted on a computer device. Figure 5 The stem cell preparation quality trust management device shown, based on quality rule object benchmarks and execution boundary control, can be used to perform the above-mentioned tasks. Figure 1 Some or all of the functionality described in the method embodiments. Please refer to [link / reference]. Figure 5 The stem cell preparation quality trust management device based on quality rule object benchmarks and execution boundary control includes: Processing unit 501 is used to construct a quality rule object set for quality assessment of stem cell preparation. The quality rule object set includes at least one of rule version references, model version references, threshold version references, permission policies, risk policies, archiving field requirements, and quality assessment rules. The acquisition unit 502 is used to acquire multi-source quality data of stem cell preparation batches and generate a standardized quality fact package based on the multi-source quality data. The standardized quality fact package is bound to batch identifier, rule version reference, model version reference and threshold version reference. The processing unit 501 is also used to create a one-time quality assessment execution instance for the standardized quality fact package, and to establish a binding relationship between the one-time quality assessment execution instance and the batch identifier, fact package identifier, quality rule object version, model version, threshold version and permission identifier; And to control the current quality operation based on binding relationships and permission identifiers through controlled execution boundaries before at least one quality operation in a one-time quality assessment execution instance request to execute model inference invocation, automatic release, partial overwrite, archive write or self-healing reconstruction; And when the current quality operation meets the control conditions of the controlled execution boundary, perform stem cell preparation quality assessment based on the quality rule object set and standardized quality fact package to obtain quality assessment results, risk level or batch disposal decision candidates; And when the current quality operation, quality assessment result, risk level, event feedback status or batch disposal decision candidate does not meet the preset trust conditions, generate a dynamic isolation record and execute at least one of the following security actions: blocking, freezing, permission revocation, manual review, downgrade protection or self-healing verification. And for encapsulating at least one of quality assessment results, boundary control results, safety disposal results, and archived evidence into a standardized quality event, and forming a traceable chain of evidence based on the standardized quality event.

[0093] In one implementation, the standardized quality fact package includes a fact package summary value, a fact package signature value, a data confidence level, a batch identifier, a rule version reference, a model version reference, and a threshold version reference; the standardized quality fact package also includes at least one of the following: fact package identifier, stem cell type, preparation process stage, environmental gating data, equipment calibration status, process step integrity record, microscopic image summary features, flow cytometry summary features, qPCR summary features, sequencing summary features, culture time sequence summary features, batch ontology vector, and archived required fields; the processing unit 501 is further configured to: Before the standardized quality fact package enters the quality assessment process, a pre-access consistency check is performed. The pre-access consistency check includes at least two of the following: fact package integrity check, fact package summary check, batch identifier consistency check, rule version compatibility check, model version compatibility check, threshold version compatibility check, archived field integrity check, and data confidence check. If the pre-access consistency check fails, the standardized quality fact package is blocked from entering the quality assessment process, and an access failure event or dynamic isolation record is generated.

[0094] In one implementation, the controlled execution boundary performs runtime consistency checks based on normalized quality operation tuples; the normalized quality operation tuples include at least a batch identifier, a standardized quality fact package identifier, a one-time quality assessment execution instance identifier, a rule version reference, a model version reference, a threshold version reference, a permission bitmask, a lifecycle state, an operation type, and a controlled execution boundary identifier; the processing unit 501 is further configured to: Consistency comparison is performed based on normalized quality operation tuples and binding relationships; If the normalized quality operation tuple is inconsistent with the binding relationship, or if the permission bit corresponding to the current operation is not set in the permission bit mask, the current quality operation is blocked.

[0095] In one embodiment, the permission identifier includes automatic release permission; the permission identifier also includes a permission bitmask, wherein different permission bits in the permission bitmask correspond to at least one of environment gating calculation permission, batch ontology evaluation permission, model inference call permission, decision routing permission, partial overwrite request permission, archive write permission, and self-healing reconstruction trigger permission; the processing unit 501 is further configured to: When a one-time quality assessment execution instance requests to execute the current operation, a bitwise matching verification is performed based on the permission bitmask and the current operation type. If the permission bit corresponding to the current operation is not set, block the current operation; Controlled execution boundaries are implemented through at least one of the following: API gateway, service broker, container security policy, system call interception, middleware interception, sandbox access control, workflow state machine, message queue topic isolation, or trusted execution environment.

[0096] In one implementation, the quality rule object set includes a global quality rule baseline and local overwrite rules; The global quality rule baseline is used to define the lower limit of quality requirements in the quality assessment of stem cell preparation; Local overwrite rules are used to make controlled adjustments to certain thresholds, weights, feature structures, or model compatibility relationships based on stem cell type, preparation process, equipment conditions, or laboratory node, without exceeding the global quality rule baseline; processing unit 501 is also used for: Before the local overwrite rule takes effect, an overwrite consistency check is performed. The overwrite consistency check includes checking whether the local overwrite field is an allowed overwrite field, checking whether the local overwrite threshold is lower than the lower limit of the quality requirement, checking whether the version of the rule object after local overwrite is compatible with the model version bound to the one-time quality assessment execution instance, checking whether the feature structure after local overwrite is compatible with the archive traceability rule object, and checking whether the local overwrite request has the corresponding permission bit.

[0097] In one implementation, the global quality rule baseline in the quality rule object set is saved as an immutable rule snapshot; the processing unit 501 is further configured to: When a quality rule object changes, a new version of the quality rule object is generated in a non-overwrite form, and a quality rule benchmark evidence chain is formed based on the rule object summary value, the previous version summary value, the timestamp, the signature, and the version chain relationship. When a quality rule object is updated or reloaded, a candidate rule object or candidate rule version is generated, and signature verification and hash verification are performed; as well as at least one of the following: model compatibility verification, threshold compatibility verification, permission bit compatibility verification, fact package structure compatibility verification, risk policy compatibility verification, consensus policy compatibility verification, and archived field compatibility verification. When the verification passes, the pointer to the current valid rule version is switched atomically. When the verification fails, the currently valid quality rule object remains unchanged, and a standardized quality event for rule reload failure is generated.

[0098] In one implementation, the quality assessment includes at least two of environmental gating assessment, batch ontology assessment, and model reasoning assessment. Among them, the edge feature processing module, model inference module and isolated verification track only output standardized quality facts, candidate anomaly markers, risk labels or candidate disposal suggestions; The formal batch disposal decision is output by the decision-driven layer within the controlled execution boundary after passing runtime consistency verification and multi-source quality adjudication consistency verification. Multi-source quality adjudication consistency verification includes performing consistency verification on at least two of the following before automatic release or archiving of output: rule scoring results, batch ontology scoring results, model inference results, environmental gating results, historical similar batch results, data confidence results, and dynamic isolation records.

[0099] In one implementation, the standardized quality event includes at least one of EnvelopeID, BatchID, FactPackID, TaskID, TraceID, EvidenceID, RuleVersionRef, ModelVersionRef, ThresholdVersionRef, InstanceID, OperationType, EventType, EventPayload, EventState, Timestamp, PrevEnvelopeHash, EnvelopeHash, and Signature; the return status of the standardized quality event includes pending, sending, sent, and failed_permanent; the processing unit 501 is further configured to: When the standardized quality event fails to be returned and the maximum number of retries is reached, the return status of the standardized quality event is switched to failed_permanent. When a failed_permanent status event exists that is associated with a target batch, target standardized quality fact package, or target one-off quality assessment execution instance, block automatic release, archiving completion confirmation, or rule version switching operations that depend on the standardized quality event.

[0100] According to one embodiment of this application, Figure 1 The steps involved in the stem cell preparation quality reliability management method based on quality rule object benchmarks and execution boundary control shown can be derived from... Figure 5 The various units in the stem cell preparation quality trust management device shown are executed based on quality rule object benchmarks and execution boundary control. For example, Figure 1 Steps S101 and S103-S107 shown can be derived from... Figure 5 The processing unit 501 shown executes the step S102, which can be performed by... Figure 5 The acquisition unit 502 shown is executed. Figure 5The various units in the stem cell preparation quality trust management device based on quality rule object benchmarks and execution boundary control shown can be individually or entirely merged into one or more other units, or some of the units can be further divided into multiple functionally smaller units. This achieves the same operation without affecting the technical effects of the embodiments of this application. The above units are based on logical function division. In practical applications, the function of one unit can also be implemented by multiple units, or the function of multiple units can be implemented by one unit. In other embodiments of this application, the stem cell preparation quality trust management device based on quality rule object benchmarks and execution boundary control may also include other units. In practical applications, these functions can also be implemented with the assistance of other units, and can be implemented collaboratively by multiple units.

[0101] According to another embodiment of this application, a general-purpose computing device, such as a computer device including processing elements and storage elements such as a central processing unit (CPU), random access memory (RAM), and read-only memory (ROM), can perform operations such as... Figure 1 The computer program (including program code) involved in each step of the corresponding method shown, to construct such... Figure 5 The diagram illustrates a stem cell preparation quality assurance management device based on quality rule object benchmarks and execution boundary control, and a method for implementing stem cell preparation quality assurance management based on quality rule object benchmarks and execution boundary control according to embodiments of this application. The computer program can be recorded on, for example, a computer-readable recording medium, loaded onto the aforementioned computing device via the computer-readable recording medium, and executed therein.

[0102] Based on the same inventive concept, the principle and beneficial effects of the stem cell preparation quality reliable management device based on quality rule object benchmark and execution boundary control provided in the embodiments of this application are similar to the principle and beneficial effects of the stem cell preparation quality reliable management method based on quality rule object benchmark and execution boundary control in the embodiments of this application. Please refer to the implementation principle and beneficial effects of the method. For the sake of brevity, they will not be repeated here.

[0103] Please see Figure 6 , Figure 6 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. The computer device may be a terminal device. Figure 6As shown, the computer device includes at least a processor 601, a communication interface 602, and a memory 603. The processor 601, communication interface 602, and memory 603 can be connected via a bus or other means. The processor 601 (or Central Processing Unit, CPU) is the computing and control core of the computer device. It can parse various instructions within the computer device and process various data. For example, the CPU can parse power-on / off commands issued by an object to the computer device and control the computer device to perform power-on / off operations; it can also transmit various interactive data between internal structures of the computer device, and so on. The communication interface 602 may optionally include a standard wired interface or a wireless interface (such as Wi-Fi, mobile communication interface, etc.), and under the control of the processor 601, it can be used to send and receive data; the communication interface 602 can also be used for data transmission and interaction within the computer device. The memory 603 is the storage device in the computer device, used to store programs and data. It can be understood that the memory 603 here can include the computer device's built-in memory, or it can include extended memory supported by the computer device. The memory 603 provides storage space for storing the operating system of the computer device, which may include, but is not limited to, Android, iOS, Windows Phone, etc. This application does not limit this.

[0104] This application embodiment also provides a computer-readable storage medium (Memory), which is a memory device in a computer device used to store programs and data. It is understood that the computer-readable storage medium here can include both built-in storage media in the computer device and extended storage media supported by the computer device. The computer-readable storage medium provides storage space that stores the processing system of the computer device. Furthermore, the storage space also stores computer programs suitable for loading and execution by the processor 601. It should be noted that the computer-readable storage medium here can be high-speed RAM or non-volatile memory, such as at least one disk storage device; optionally, it can also be at least one computer-readable storage medium located remotely from the aforementioned processor.

[0105] In one embodiment, processor 601 performs the following operations by running a computer program stored in memory 603: Construct a quality rule object set for quality assessment of stem cell preparation. The quality rule object set includes at least one of rule version references, model version references, threshold version references, permission policies, risk policies, archiving field requirements, and quality assessment rules. Acquire multi-source quality data for stem cell preparation batches and generate standardized quality fact packages based on the multi-source quality data. The standardized quality fact packages are bound to batch identifiers, rule version references, model version references, and threshold version references. Create a one-time quality assessment execution instance for the standardized quality fact package, and establish the binding relationship between the one-time quality assessment execution instance and the batch identifier, fact package identifier, quality rule object version, model version, threshold version, and permission identifier; Before performing at least one quality operation among the following: a one-time quality assessment execution instance request to execute model inference call, automatic release, partial overwrite, archive write, or self-healing reconstruction, the current quality operation is controlled by the controlled execution boundary based on the binding relationship and permission identifier; When the current quality operation meets the control conditions of the controlled execution boundary, a quality assessment of stem cell preparation is performed based on the quality rule object set and the standardized quality fact package to obtain the quality assessment results, risk level or batch disposal decision candidates. When the current quality operation, quality assessment results, risk level, event feedback status, or batch disposal decision candidate does not meet the preset trust conditions, a dynamic isolation record is generated, and at least one security action is performed, including blocking, freezing, permission revocation, manual review, downgrade protection, or self-healing verification. At least one of the following is encapsulated as a standardized quality event: quality assessment results, boundary control results, safety handling results, and archived evidence. A traceable chain of evidence is then formed based on the standardized quality event.

[0106] As an optional embodiment, the standardized quality fact package includes a fact package summary value, a fact package signature value, a data confidence level, a batch identifier, a rule version reference, a model version reference, and a threshold version reference; the standardized quality fact package also includes at least one of the following: fact package identifier, stem cell type, preparation process stage, environmental gating data, equipment calibration status, process step integrity record, microscopic image summary features, flow cytometry summary features, qPCR summary features, sequencing summary features, culture time sequence summary features, batch ontology vector, and archived required fields; the processor 601 further performs the following operations by running the computer program in the memory 603: Before the standardized quality fact package enters the quality assessment process, a pre-access consistency check is performed. The pre-access consistency check includes at least two of the following: fact package integrity check, fact package summary check, batch identifier consistency check, rule version compatibility check, model version compatibility check, threshold version compatibility check, archived field integrity check, and data confidence check. If the pre-access consistency check fails, the standardized quality fact package is blocked from entering the quality assessment process, and an access failure event or dynamic isolation record is generated.

[0107] As an optional embodiment, the controlled execution boundary performs runtime consistency checks based on normalized quality operation tuples; the normalized quality operation tuples include at least a batch identifier, a standardized quality fact package identifier, a one-time quality assessment execution instance identifier, a rule version reference, a model version reference, a threshold version reference, a permission bitmask, a lifecycle state, an operation type, and a controlled execution boundary identifier; the processor 601 also performs the following operations by running a computer program in the memory 603: Consistency comparison is performed based on normalized quality operation tuples and binding relationships; If the normalized quality operation tuple is inconsistent with the binding relationship, or if the permission bit corresponding to the current operation is not set in the permission bit mask, the current quality operation is blocked.

[0108] As an optional embodiment, the permission identifier includes automatic release permission; the permission identifier also includes a permission bitmask, wherein different permission bits in the permission bitmask correspond to at least one of environment gating calculation permission, batch ontology evaluation permission, model inference call permission, decision routing permission, partial overwrite request permission, archive write permission, and self-healing reconstruction trigger permission; the processor 601, by running the computer program in the memory 603, also performs the following operations: When a one-time quality assessment execution instance requests to execute the current operation, a bitwise matching verification is performed based on the permission bitmask and the current operation type. If the permission bit corresponding to the current operation is not set, block the current operation; Controlled execution boundaries are implemented through at least one of the following: API gateway, service broker, container security policy, system call interception, middleware interception, sandbox access control, workflow state machine, message queue topic isolation, or trusted execution environment.

[0109] As an optional implementation, the quality rule object set includes a global quality rule baseline and local overwrite rules; The global quality rule baseline is used to define the lower limit of quality requirements in the quality assessment of stem cell preparation; Local overwrite rules are used to make controlled adjustments to certain thresholds, weights, feature structures, or model compatibility relationships based on stem cell type, preparation process, equipment conditions, or laboratory node, without exceeding the global quality rule baseline; processor 601 also performs the following operations by running the computer program in memory 603: Before the local overwrite rule takes effect, an overwrite consistency check is performed. The overwrite consistency check includes checking whether the local overwrite field is an allowed overwrite field, checking whether the local overwrite threshold is lower than the lower limit of the quality requirement, checking whether the version of the rule object after local overwrite is compatible with the model version bound to the one-time quality assessment execution instance, checking whether the feature structure after local overwrite is compatible with the archive traceability rule object, and checking whether the local overwrite request has the corresponding permission bit.

[0110] As an optional embodiment, the global quality rule baseline in the quality rule object set is saved as an immutable rule snapshot; the processor 601 also performs the following operations by running a computer program in the memory 603:

[0111] When a quality rule object changes, a new version of the quality rule object is generated in a non-overwrite form, and a quality rule benchmark evidence chain is formed based on the rule object summary value, the previous version summary value, the timestamp, the signature, and the version chain relationship. When a quality rule object is updated or reloaded, a candidate rule object or candidate rule version is generated, and signature verification and hash verification are performed; as well as at least one of the following: model compatibility verification, threshold compatibility verification, permission bit compatibility verification, fact package structure compatibility verification, risk policy compatibility verification, consensus policy compatibility verification, and archived field compatibility verification. When the verification passes, the pointer to the current valid rule version is switched atomically. When the verification fails, the currently valid quality rule object remains unchanged, and a standardized quality event for rule reload failure is generated.

[0112] As an optional embodiment, the quality assessment includes at least two of the following: environmental gating assessment, batch ontology assessment, and model reasoning assessment. Among them, the edge feature processing module, model inference module and isolated verification track only output standardized quality facts, candidate anomaly markers, risk labels or candidate disposal suggestions; The formal batch disposal decision is output by the decision-driven layer within the controlled execution boundary after passing runtime consistency verification and multi-source quality adjudication consistency verification. Multi-source quality adjudication consistency verification includes performing consistency verification on at least two of the following before automatic release or archiving of output: rule scoring results, batch ontology scoring results, model inference results, environmental gating results, historical similar batch results, data confidence results, and dynamic isolation records.

[0113] As an optional embodiment, the standardized quality event includes at least one of EnvelopeID, BatchID, FactPackID, TaskID, TraceID, EvidenceID, RuleVersionRef, ModelVersionRef, ThresholdVersionRef, InstanceID, OperationType, EventType, EventPayload, EventState, Timestamp, PrevEnvelopeHash, EnvelopeHash, and Signature; the return status of the standardized quality event includes pending, sending, sent, and failed_permanent; the processor 601 further performs the following operations by running the computer program in memory 603: When the standardized quality event fails to be returned and the maximum number of retries is reached, the return status of the standardized quality event is switched to failed_permanent. When a failed_permanent status event exists that is associated with a target batch, target standardized quality fact package, or target one-off quality assessment execution instance, block automatic release, archiving completion confirmation, or rule version switching operations that depend on the standardized quality event.

[0114] Based on the same inventive concept, the principle and beneficial effects of the computer device provided in the embodiments of this application in solving the problem are similar to the principle and beneficial effects of the stem cell preparation quality reliable management method based on quality rule object benchmark and execution boundary control in the method embodiments of this application. You can refer to the principle and beneficial effects of the method implementation. For the sake of brevity, it will not be repeated here.

[0115] This application also provides a computer-readable storage medium storing a computer program adapted to be loaded by a processor and executed by the above-described method embodiment of the stem cell preparation quality reliable management method based on quality rule object benchmark and execution boundary control.

[0116] This application also provides a computer program product or computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the aforementioned method for reliable quality control of stem cell preparation based on quality rule object benchmarks and execution boundary control.

[0117] The steps in the method of this application embodiment can be adjusted, combined, or deleted according to actual needs.

[0118] The modules in the device of this application embodiment can be merged, divided, and deleted according to actual needs.

[0119] In the embodiments of this application, the term "module" or "unit" refers to a computer program or part of a computer program with a predetermined function, which works together with other related parts to achieve a predetermined goal, and can be implemented wholly or partially using software, hardware (such as processing circuitry or memory), or a combination thereof. Similarly, a processor (or multiple processors or memory) can be used to implement one or more modules or units. Furthermore, each module or unit can be part of an overall module or unit that includes the functionality of that module or unit.

[0120] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, which may include: flash drive, read-only memory (ROM), random access memory (RAM), magnetic disk or optical disk, etc.

[0121] The above-disclosed embodiments are merely preferred embodiments of this application and should not be construed as limiting the scope of this application. Those skilled in the art will understand that all or part of the processes for implementing the above embodiments and equivalent variations made in accordance with the claims of this application are still within the scope of this application.

Claims

1. A reliable quality control method for stem cell preparation based on quality rule object benchmarks and execution boundary control, characterized in that, include: Construct a quality rule object set for quality assessment of stem cell preparation, wherein the quality rule object set includes at least one of rule version references, model version references, threshold version references, permission policies, risk policies, archiving field requirements, and quality assessment rules; Acquire multi-source quality data of stem cell preparation batches, and generate a standardized quality fact package based on the multi-source quality data. The standardized quality fact package is bound to batch identifier, rule version reference, model version reference, and threshold version reference. Create a one-time quality assessment execution instance for the standardized quality fact package, and establish the binding relationship between the one-time quality assessment execution instance and the batch identifier, fact package identifier, quality rule object version, model version, threshold version, and permission identifier; Before the one-time quality assessment execution instance requests execution model inference call, automatic release, partial overwrite, archive write, or self-healing reconstruction, the current quality operation is controlled by the controlled execution boundary based on the binding relationship and the permission identifier; When the current quality operation meets the control conditions of the controlled execution boundary, a stem cell preparation quality assessment is performed based on the quality rule object set and the standardized quality fact package to obtain the quality assessment results, risk level, or batch disposal decision candidates. When the current quality operation, quality assessment results, risk level, event feedback status, or batch disposal decision candidate does not meet the preset trust conditions, a dynamic isolation record is generated, and at least one security action is performed, including blocking, freezing, permission revocation, manual review, downgrade protection, or self-healing verification. At least one of the following is encapsulated as a standardized quality event: quality assessment results, boundary control results, safety handling results, and archived evidence. A traceable chain of evidence is then formed based on the standardized quality event.

2. The method as described in claim 1, characterized in that, The standardized quality fact package includes a fact package summary value, a fact package signature value, data confidence level, batch identifier, rule version reference, model version reference, and threshold version reference; the standardized quality fact package also includes at least one of the following: fact package identifier, stem cell type, preparation process stage, environmental gating data, equipment calibration status, process step integrity record, microscopic image summary features, flow cytometry summary features, qPCR summary features, sequencing summary features, culture time sequence summary features, batch ontology vector, and required archiving fields; the method further includes: Before the standardized quality fact package enters the quality assessment process, a pre-access consistency check is performed. The pre-access consistency check includes at least two of the following: fact package integrity check, fact package digest check, batch identifier consistency check, rule version compatibility check, model version compatibility check, threshold version compatibility check, archived field integrity check, and data confidence check. If the pre-access consistency check fails, the standardized quality fact package is blocked from entering the quality assessment process, and an access failure event or dynamic isolation record is generated.

3. The method as described in claim 1, characterized in that, The controlled execution boundary is based on runtime consistency verification performed using normalized quality operation tuples; the normalized quality operation tuples include at least a batch identifier, a standardized quality fact package identifier, a one-time quality assessment execution instance identifier, a rule version reference, a model version reference, a threshold version reference, a permission bitmask, a lifecycle state, an operation type, and a controlled execution boundary identifier; the method further includes: A consistency comparison is performed based on the normalized quality operation tuple and the binding relationship; When the normalized quality operation tuple is inconsistent with the binding relationship, or when the permission bit corresponding to the current operation is not set in the permission bit mask, the current quality operation is blocked.

4. The method as described in claim 1, characterized in that, The permission identifier includes automatic access permission; the permission identifier also includes a permission bitmask, wherein different permission bits in the permission bitmask correspond to at least one of the following: environment gating calculation permission, batch ontology evaluation permission, model inference call permission, decision routing permission, partial overwrite request permission, archive write permission, and self-healing reconstruction trigger permission; the method further includes: When the one-time quality assessment execution instance requests to execute the current operation, a bit-by-bit matching verification is performed based on the permission bitmask and the current operation type; If the permission bit corresponding to the current operation is not set, block the current operation; The controlled execution boundary is implemented through at least one of the following: API gateway, service proxy, container security policy, system call interception, middleware interception, sandbox access control, workflow state machine, message queue topic isolation, or trusted execution environment.

5. The method as described in claim 1, characterized in that, The set of quality rule objects includes a global quality rule baseline and local overwrite rules; The global quality rule baseline is used to define the lower limit of quality requirements in the quality assessment of stem cell preparation; The local overwrite rule is used to controllably adjust certain thresholds, weights, feature structures, or model compatibility relationships based on stem cell type, preparation process, equipment conditions, or laboratory node, without exceeding the global quality rule baseline; the method further includes: Before the partial overwrite rule takes effect, an overwrite consistency check is performed; the overwrite consistency check includes checking whether the partial overwrite field is an allowed overwrite field, checking whether the partial overwrite threshold is lower than the lower limit of the quality requirement, checking whether the version of the rule object after partial overwrite is compatible with the model version bound to the one-time quality assessment execution instance, checking whether the feature structure after partial overwrite is compatible with the archive traceability rule object, and checking whether the partial overwrite request has a corresponding permission bit.

6. The method as described in claim 1, characterized in that, The global quality rule baseline in the quality rule object set is saved as an immutable rule snapshot; the method further includes: When a quality rule object changes, a new version of the quality rule object is generated in a non-overwrite form, and a quality rule benchmark evidence chain is formed based on the rule object summary value, the previous version summary value, the timestamp, the signature, and the version chain relationship. When a quality rule object is updated or reloaded, a candidate rule object or candidate rule version is generated, and signature verification and hash verification are performed; as well as at least one of the following: model compatibility verification, threshold compatibility verification, permission bit compatibility verification, fact package structure compatibility verification, risk policy compatibility verification, consensus policy compatibility verification, and archived field compatibility verification. When the verification passes, the pointer to the current valid rule version is switched atomically. When the verification fails, the currently valid quality rule object remains unchanged, and a standardized quality event for rule reload failure is generated.

7. The method as described in claim 1, characterized in that, The quality assessment includes at least two of the following: environmental gating assessment, batch ontology assessment, and model reasoning assessment. Among them, the edge feature processing module, model inference module and isolated verification track only output standardized quality facts, candidate anomaly markers, risk labels or candidate disposal suggestions; The formal batch disposal decision is output by the decision-driven layer within the controlled execution boundary after passing runtime consistency verification and multi-source quality adjudication consistency verification. The multi-source quality adjudication consistency verification includes performing consistency verification on at least two of the following before automatic release or archiving of output: rule scoring results, batch ontology scoring results, model inference results, environmental gating results, historical similar batch results, data confidence results, and dynamic isolation records.

8. The method as described in claim 1, characterized in that, The standardized quality events include at least one of EnvelopeID, BatchID, FactPackID, TaskID, TraceID, EvidenceID, RuleVersionRef, ModelVersionRef, ThresholdVersionRef, InstanceID, OperationType, EventType, EventPayload, EventState, Timestamp, PrevEnvelopeHash, EnvelopeHash, and Signature; the return status of the standardized quality events includes pending, sent, sent, and failed_permanent; the method further includes: When the standardized quality event fails to be returned and the maximum number of retries is reached, the return status of the standardized quality event is switched to failed_permanent. When a failed_permanent status event is associated with a target batch, target standardized quality fact package, or target one-off quality assessment execution instance, block automatic release, archiving completion confirmation, or rule version switching operations that depend on the standardized quality event.

9. A reliable quality control system for stem cell preparation based on quality rule object benchmarks and execution boundary control, characterized in that, include: The quality rule object management module is used to construct and maintain the set of quality rule objects required for stem cell preparation quality assessment; the set of quality rule objects includes at least one of rule version references, model version references, threshold version references, permission policies, risk policies, archiving field requirements, and quality assessment rules; A standardized quality fact package generation module is used to acquire multi-source quality data of stem cell preparation batches and generate standardized quality fact packages that are bound to batch identifiers, rule version references, model version references, and threshold version references; the standardized quality fact packages are bound to batch identifiers, rule version references, model version references, and threshold version references. The execution instance management module is used to create a one-time quality assessment execution instance for the standardized quality fact package, and to establish the binding relationship between the one-time quality assessment execution instance and the batch identifier, fact package identifier, quality rule object version, model version, threshold version, and permission identifier. The controlled execution boundary module is used to control the current quality operation based on the binding relationship and the permission identifier through the controlled execution boundary before at least one quality operation in the one-time quality assessment execution instance request execution model inference call, automatic release, partial overwrite, archive write or self-healing reconstruction. The quality assessment and decision-making module is used to perform stem cell preparation quality assessment based on the quality rule object set and the standardized quality fact package when the current quality operation meets the control conditions of the controlled execution boundary, and obtain the quality assessment result, risk level or batch disposal decision candidate. The dynamic isolation and security handling module is used to generate a dynamic isolation record and execute at least one of the following security handling actions when the current quality operation, quality assessment result, risk level, event feedback status or batch handling decision candidate does not meet the preset trust conditions: blocking, freezing, permission revocation, manual review, downgrade protection or self-healing verification. The event feedback and evidence archiving module is used to encapsulate at least one of the following: quality assessment results, boundary control results, safety handling results, and archived evidence into a standardized quality event, and to form a traceable evidence chain based on the standardized quality event.

10. A computer device, characterized in that, include: A memory, wherein a computer program is stored; A processor is configured to load the computer program to implement the stem cell preparation quality reliable management method based on quality rule object benchmarks and execution boundary control as described in any one of claims 1-8.

Images