A cloud platform-based medical instrument sales management method and system

Abstract

The application relates to the technical field of medical instrument sales management, and discloses a medical instrument sales management method and system based on a cloud platform, wherein the method comprises the following steps: obtaining applicable range texts before and after registration certificate change, performing sentence-level semantic difference, and generating an applicable range semantic difference result; performing semantic mapping on the applicable range semantic difference result through a medical instrument classification directory knowledge graph to generate a structured range reduction influence description; generating a comprehensive influence product list based on the structured range reduction influence description and specification and model change information; searching for uncompleted orders based on the comprehensive influence product list to generate a hierarchical treatment suggestion list; pushing a patient exclusion feature constraint set to a medical institution data exchange node, receiving a matching result, and generating a potential high-risk patient screening result; and calculating the clinical risk comprehensive score of each potential high-risk patient to generate a patient-level clinical risk assessment list.

Description

Technical Field

[0001] This invention relates to the field of medical device sales management technology, and more specifically, to a cloud-based medical device sales management method and system. Background Technology

[0002] The scope of application recorded on a medical device registration certificate is the core basis for the legal sale and use of the product. The cloud platform is used to manage core business data such as product registration certificate information, product master data, sales and delivery records, and UDI code tracking. When the scope of application of the registration certificate changes, the differences before and after the change are usually described in natural language, which may imply a substantial reduction in the scope of application.

[0003] The existing process relies on quality management personnel to manually compare the changes. Manual comparison depends on overall semantic understanding, which can easily overlook the scope reduction implied by subtle wording differences, resulting in products on sale that no longer meet the applicable scope after the changes not being identified in a timely manner.

[0004] Furthermore, for high-value implantable medical devices, the narrowing change in the scope of application means that some patients who had already received implantation before the change may fall into the excluded group after the change. Existing cloud platforms only track the flow of products to medical institutions and cannot automatically link registration certificate changes to specific patients who have already received implantation. This results in potentially high-risk patients who have already received implantation not being identified and tracked during the sales management process. Summary of the Invention

[0005] This invention provides a cloud-based medical device sales management method and system, which solves the technical problem in related technologies that the affected products and potentially high-risk implantation patients cannot be automatically identified after the scope of application of the registration certificate is changed.

[0006] This invention discloses a cloud-based medical device sales management method, comprising: performing sentence-level segmentation on the text of the applicable scope before and after the registration certificate change; performing the longest common subsequence algorithm on the sentence segment sequence before and after the change; identifying purely deleted sentence segments, modified sentence segments and restrictive words through position alignment and word-by-word comparison; and generating semantic difference results of the applicable scope. The semantic difference results of the applicable scope are semantically mapped in the medical device classification catalog knowledge graph to generate a scope reduction impact description that includes a set of affected product classification codes, a set of applicable condition constraints, and a set of patient exclusion feature constraints. Based on the cloud platform product master data associated with the scope reduction impact description, a comprehensive list of affected products is generated. High-value implantable products are selected from the comprehensive impact product list, and implantation traceability records are retrieved from the cloud platform to generate a full list of implanted patients. The patient exclusion feature constraint set is pushed to the medical institution's data exchange node, whereby each institution performs exclusion feature matching and returns the matching status to generate screening results for potentially high-risk patients.

[0007] Furthermore, the identification of purely deleted sentence segments, modified sentence segments, and restrictive qualifiers through position alignment and word-by-word comparison includes: marking sentence segments in the original sentence segment sequence that do not belong to the longest common clause segment set as deleted sentence segments, and marking sentence segments in the post-modified sentence segment sequence that do not belong to the longest common clause segment set as newly added sentence segments; aligning the deleted and newly added sentence segments according to their positions in the original sequence, pairing deleted sentence segments that can establish positional correspondence with newly added sentence segments, performing word-by-word comparison on each pair, identifying the newly added sentence segments in the pair as modified sentence segments, and retaining deleted sentence segments that cannot establish positional correspondence as purely deleted sentence segments; and extracting restrictive qualifiers from the words added in each modified sentence segment relative to the corresponding original sentence segment to obtain a set of restrictive qualifiers.

[0008] Furthermore, the restrictive qualifiers include condition-limiting words and scope-narrowing words; wherein, the condition-limiting words are new limiting phrases with added preconditions added in the modified sentence segment; and the scope-narrowing words are words added in the modified sentence segment to narrow the coverage when the generalized expression is replaced with a specialized expression.

[0009] Furthermore, the nodes of the medical device classification catalog knowledge graph include product classification code nodes, scope of application description nodes, applicable condition attribute nodes, and patient characteristic nodes, which are connected by semantic relationship edges. The step of performing semantic mapping on the semantic difference results of the scope of application in the medical device classification catalog knowledge graph includes: extracting medical device terms from each deleted and modified sentence segment, matching them with the corresponding product classification code nodes, and summarizing them to obtain an affected product classification code set; matching each restrictive term with the corresponding applicable condition attribute node to obtain an applicable condition constraint set; and based on each constraint in the applicable condition constraint set, reasoning along the association path from the applicable condition node to the patient characteristic node to obtain a patient exclusion characteristic constraint set, which includes exclusionary comorbidity codes, exclusionary age ranges, and exclusionary indication codes.

[0010] Further, the step of generating a comprehensive list of affected products based on the cloud platform product master data related to the scope reduction impact description includes: performing a set difference operation on the specification and model lists of the versions before and after the registration certificate change to identify the deleted specification and model set; associating and matching each specification and model in the deleted specification and model set with the SKU code in the cloud platform product master data according to the specification and model field to obtain a first affected SKU code set; associating and matching each code in the affected product category code set with the SKU code in the cloud platform product master data according to the product category code field to obtain a second affected SKU code set; performing a merge and deduplication operation on the first affected SKU code set and the second affected SKU code set to obtain a comprehensive affected SKU code set; wherein, the labels from the first affected SKU code set are specification deletion type, and the labels from the second affected SKU code set are scope reduction type.

[0011] Furthermore, it also includes: retrieving sales orders with an incomplete status based on the comprehensive impact product list; for each incomplete order, calculating the ratio of the affected product amount to the total order amount as the impact percentage, and querying available SKU codes of the same category code not in the comprehensive impact product list as substitute products; when the impact percentage is greater than a preset high impact threshold and there are no substitute products, marking the order as urgently suspended; when there are substitute products, generating product replacement suggestions containing the SKU codes of the substitute products and available inventory; when the impact percentage is not greater than the preset high impact threshold and there are no substitute products, marking the order as low impact pending observation; summarizing into a tiered handling suggestion list.

[0012] Further, the step of pushing the patient exclusion feature constraint set to the medical institution data exchange node includes: encapsulating the patient exclusion feature constraint set into structured query conditions, the structured query conditions including a list of excluded comorbidity codes, an excluded age range, and a list of excluded indication codes; grouping according to the medical institution codes of each record in the full implantation patient list, and for each involved medical institution, pushing the structured query conditions and the corresponding desensitized patient identifier list to the data exchange node of that medical institution; for each patient in the received desensitized patient identifier list, each medical institution data exchange node determines whether the patient has any comorbidity in the list of excluded comorbidity codes, whether the patient is within the excluded age range, or whether the patient has any indication in the list of excluded indication codes, and returns the matching status to the cloud platform.

[0013] Furthermore, it also includes: associating the screening results of the potential high-risk patients with the full list of implantation patients using desensitized patient identifiers to obtain the implantation surgery date and implantation site code for each potential high-risk patient; calculating the implantation duration based on the difference between the implantation surgery date and the current date; performing a lookup operation on each potential high-risk patient using a pre-configured three-dimensional scoring matrix, where the row dimension of the three-dimensional scoring matrix corresponds to the clinical severity level, the column dimension corresponds to the implantation duration interval, and the layer dimension corresponds to the anatomical risk level; the comprehensive score is determined by the following formula: ,in, This is the overall score. It is a three-dimensional rating matrix. This is the highest severity level value among all the exclusion features matched for this patient. The interval number to which the implantation duration falls after being divided into preset intervals. Assign a value to the anatomical risk level corresponding to the implantation site; based on the aforementioned comprehensive score. The intervals divide patients into three levels: emergency re-examination, intensive follow-up, and routine observation, generating a patient-level clinical risk assessment list.

[0014] Furthermore, the clinical severity level The determination method is as follows: when a patient matches multiple exclusion features simultaneously, the severity level corresponding to each matching exclusion feature is obtained, and the highest value is taken as the clinical severity level. Among them, the severity of excluded comorbidities is higher than that of excluded age ranges and excluded indications.

[0015] This invention discloses a cloud-based medical device sales management system, comprising: a semantic difference module, used to perform sentence-level segmentation on the applicable scope text before and after the registration certificate change, to perform the longest common subsequence algorithm on the sentence segment sequence before and after the change, to identify purely deleted sentence segments, modified sentence segments and restrictive words through position alignment and word-by-word comparison, and to generate semantic difference results of the applicable scope; The knowledge graph mapping module is used to perform semantic mapping on the semantic difference results of the applicable scope in the medical device classification catalog knowledge graph to generate a scope reduction impact description that includes a set of affected product classification codes, a set of applicable condition constraints, and a set of patient exclusion feature constraints. The product impact analysis module is used to generate a comprehensive list of impacted products based on the cloud platform product master data associated with the scope reduction impact description. The implantation traceability retrieval module is used to filter high-value implantable products from the comprehensive impact product list and retrieve implantation traceability records from the cloud platform to generate a full list of implanted patients. The patient screening module is used to push the patient exclusion feature constraint set to the medical institution's data exchange node, whereby each institution performs exclusion feature matching and returns the matching status to generate screening results for potentially high-risk patients.

[0016] The beneficial effects of this invention are as follows: This invention performs sentence-level segmentation and longest common subsequence comparison on the text of the registration certificate's scope of application. It decomposes natural language-level wording changes into purely deleted sentence segments, modified sentence segments, and restrictive qualifiers, presenting implicit scope reduction in a structured form. Compared to manual comparison methods, this approach can output subtle wording differences as calculable, structured results, helping to reduce the omission of implicit scope reduction and thus enabling more timely identification of the compliance status of products on the market.

[0017] This invention achieves patient-level exclusion feature screening by pushing a structured set of patient exclusion feature constraints to the data exchange nodes of various medical institutions. Each institution completes the matching within its internal system and only reports the matching status. This is done without the sales company directly accessing the patient's diagnostic data. This approach extends the tracking granularity of the cloud platform from the medical institution level to the specific patient level, enabling potentially high-risk implanted patients in non-recall registration certificate change scenarios to be discovered and tracked during the sales management process. Attached Figure Description

[0018] Figure 1 This is a flowchart of a cloud-based medical device sales management method provided in an embodiment of the present invention; Figure 2 This is a schematic diagram illustrating the classification and statistics of semantic difference results within the applicable scope provided in this embodiment of the invention; Figure 3 This is a schematic diagram illustrating the comprehensive impact of SKU inventory and in-transit order volume provided in an embodiment of the present invention; Figure 4 This is a schematic diagram illustrating the distribution of suggestions for tiered handling of incomplete orders provided in an embodiment of the present invention; Figure 5 This is a schematic diagram of the three-dimensional score distribution for implantation patient risk assessment provided in an embodiment of the present invention; Figure 6 This is a schematic diagram illustrating the distribution statistics of at-risk patients in various medical institutions provided in an embodiment of the present invention; Figure 7 This is a schematic diagram illustrating the output statistics of the entire data processing steps provided in this embodiment of the invention; Figure 8 This is a schematic diagram of the distribution of patient exclusion feature matching types provided in an embodiment of the present invention; Figure 9 This is a schematic diagram of the impact types of affected SKUs and the distribution of inventory risk provided in the embodiments of the present invention. Detailed Implementation

[0019] The subject matter described herein will now be discussed with reference to exemplary embodiments. It should be understood that these embodiments are discussed only to enable those skilled in the art to better understand and implement the subject matter described herein, and changes may be made to the function and arrangement of the elements discussed without departing from the scope of this specification. Various processes or components may be omitted, substituted, or added as needed in the examples. Furthermore, some features described in the examples may be combined in other examples.

[0020] At least one embodiment of the present invention discloses a medical device sales management method based on a cloud platform, see [link to relevant documentation]. Figure 1 This includes the following steps: Step 1: Obtain the text of the applicable scope before and after the registration certificate change, perform sentence-level semantic difference, and generate the semantic difference result of the applicable scope.

[0021] Obtain the full-text structured data of the target registration certificate before and after the change from the cloud platform, and extract the applicable scope field text from it. Perform sentence-level segmentation on the applicable scope text of the before and after the change, respectively, to obtain the sentence segment sequence before the change and the sentence segment sequence after the change.

[0022] The longest common subsequence algorithm is applied to both the original and modified sentence segment sequences. The input consists of two sentence segment sequences, and the output is the set of the longest common subsequences of the two sequences. Segments in the original sequence that do not belong to the set of the longest common subsequences are marked as deleted segments, and segments in the modified sequence that do not belong to the set of the longest common subsequences are marked as added segments. The deleted and added segments are aligned according to their positions in the original sequences. Deleted segments that can establish a positional correspondence are paired with added segments. A word-by-word comparison is performed on each pairing. Added segments in the pairings are identified as modified segments, while deleted segments that cannot establish a positional correspondence are retained as purely deleted segments.

[0023] For each modified sentence segment, newly added words relative to the corresponding unmodified sentence segment are extracted using restrictive qualifiers to obtain a set of restrictive qualifiers. These restrictive qualifiers include conditional qualifiers and scope-narrowing qualifiers. The set of purely deleted sentence segments, the set of modified sentence segments, and the set of restrictive qualifiers are combined to form a semantic difference result for applicable scope.

[0024] It should be noted that the aforementioned restrictive terms refer to words or phrases added in the revised sentences that narrow the scope of application. Specifically, conditional terms include expressions that add preconditions, such as restrictive phrases like "under specific conditions" or "after evaluation"; narrowing terms include words that replace generalized expressions with specialized expressions, such as the newly added "under specific conditions" in the revised "for treatment of XX parts" to "for treatment of XX parts under specific conditions".

[0025] Step 2: Semantically map the applicable scope semantic difference results through the medical device classification catalog knowledge graph to generate a structured scope reduction impact description.

[0026] Using the pure deletion segments, modified segments, and restrictive qualifiers from the semantic difference results of the applicable scope as query input, semantic mapping is performed in the medical device classification catalog knowledge graph.

[0027] It should be noted that the above-mentioned medical device classification catalog knowledge graph is a graph structure pre-generated based on the medical device classification catalog as the data source. Its nodes include product classification code nodes, scope of application description nodes, applicable condition attribute nodes, and patient characteristic nodes. The nodes are connected by semantic relationship edges such as "applicable to", "restricted to", and "excluded".

[0028] Step 2 may specifically include: Step 201: After extracting the medical device terms from each deleted and modified sentence segment, match the corresponding product classification code nodes in the medical device classification catalog knowledge graph, and summarize to obtain the set of affected product classification codes.

[0029] Step 202: Match each restrictive term with the corresponding applicable condition attribute node in the medical device classification catalog knowledge graph to obtain the set of applicable condition constraints.

[0030] Step 203: Based on the constraints in the applicable condition constraint set, reason along the association path from the applicable condition node to the patient feature node in the medical device classification catalog knowledge graph to obtain the patient exclusion feature constraint set. The patient exclusion feature constraint set includes exclusionary comorbidity codes, exclusionary age ranges, and exclusionary indication codes.

[0031] The affected product classification code set, applicable condition constraint set, and patient exclusion feature constraint set are combined into a structured scope reduction impact description.

[0032] Step 3: Generate a comprehensive list of affected products based on the structured description of scope reduction impact and specification change information.

[0033] Perform a set difference operation on the specification and model lists in the pre-change and post-change versions of the registration certificate, and identify the specification and model that exist in the specification and model list of the pre-change version but do not exist in the specification and model list of the post-change version as the deleted specification and model set.

[0034] The first set of affected SKU codes is obtained by associating and matching each specification in the set of deleted specifications with the SKU codes in the cloud platform product master data according to the specification field. The second set of affected SKU codes is obtained by associating and matching each code in the set of affected product category codes in the structured scope reduction impact description with the SKU codes in the cloud platform product master data according to the product category code field.

[0035] Perform a merge and deduplication operation on the first and second affected SKU code sets to obtain the comprehensive affected SKU code set. For each SKU code in the comprehensive affected SKU code set, query its current inventory and in-transit order quantity, and label the impact type according to its source. SKU codes from the first affected SKU code set are labeled as specification deletion type, and those from the second affected SKU code set are labeled as range narrowing type. Summarize the above information into a comprehensive affected product list.

[0036] Step 4: Based on the comprehensive impact product list, retrieve the incomplete orders and generate a tiered disposal suggestion list.

[0037] In the cloud platform sales order system, using the SKU codes in the comprehensive affected product list as query conditions, all sales orders containing the affected products and whose status is incomplete are retrieved.

[0038] For each incomplete order obtained from the retrieval, the following processing is performed: the ratio of the amount of the affected products in the order to the total order amount is calculated as the impact percentage; based on the product category code and specification parameters corresponding to the affected SKU code, available SKU codes with the same category code but not in the comprehensive affected product list are queried in the cloud platform product master data as alternative products.

[0039] When the impact percentage exceeds the preset high impact threshold and there is no alternative product, the order is marked as urgently paused. If an alternative product exists, a product replacement suggestion is generated, which includes the SKU code and available inventory of the alternative product. When the impact percentage is not greater than the preset high impact threshold and there is no alternative product, the order is marked as low impact and awaiting observation. All incomplete order handling information is summarized into a tiered handling suggestion list.

[0040] In this embodiment of the application, in order to conduct risk screening for patients who have already had high-value medical devices implanted, steps 5 to 7 are included in addition to step 4. The problem addressed by steps 5 to 7 is that after the scope of application is narrowed, some patients who had already had the devices implanted before the change may belong to the excluded applicable group, and the cloud platform cannot automatically associate the change with specific patients.

[0041] Step 5: Based on the high-value implantable products in the comprehensive impact product list, retrieve implantation traceability records and generate a full list of implanted patients.

[0042] From the comprehensive impact product list, SKU codes of high-value implantable medical devices are selected, and the corresponding product DI code sets are extracted. Using the product DI code sets as search criteria, all completed implantation UDI code records are retrieved from the cloud platform's implantation traceability records.

[0043] For each implantation traceability record obtained from the retrieval, the anonymized patient identifier, implantation surgery date, implantation site code, and medical institution code are extracted. All extracted results are then compiled into a complete list of implantation patients.

[0044] Step 6: Push the patient exclusion feature constraint set to the medical institution's data exchange node, receive the matching results, and generate screening results for potentially high-risk patients.

[0045] The structured scope reduction effect description encapsulates the set of patient exclusion feature constraints into structured query conditions. These structured query conditions include a list of excluded comorbidity codes, an excluded age range, and a list of excluded indication codes.

[0046] Patients are grouped according to the medical institution codes of each record in the full implantation patient list. For each involved medical institution, the structured query conditions and the corresponding de-identified patient ID list are pushed to the data exchange node of that medical institution through the cloud platform data exchange interface.

[0047] Each medical institution's data exchange node, within its internal system, performs an exclusion feature matching query for each patient in the received de-identified patient identifier list. This query determines whether the patient has any comorbidity in the exclusion comorbidity code list, is within the exclusion age range, or has any indication in the exclusion indication code list. Each institution then transmits the matching results back to the cloud platform.

[0048] The matching results returned by various medical institutions are aggregated, and the patient identifiers that match at least one exclusion feature and the specific exclusion features matched are extracted to generate screening results for potentially high-risk patients.

[0049] It should be noted that during the aforementioned push and feedback process, the cloud platform does not directly obtain the original patient diagnostic data. Each medical institution only sends back structured matching results, i.e., the matching status of exclusion features corresponding to each de-identified patient identifier. This method completes patient-level exclusion feature screening without the sales company possessing the patient's clinical data.

[0050] Step 7: Calculate the comprehensive clinical risk score for each potentially high-risk patient and generate a patient-level clinical risk assessment list.

[0051] The screening results of potentially high-risk patients were linked to the full list of implanted patients using desensitized patient identifiers to obtain the implantation surgery date and implantation site code for each potentially high-risk patient. The implantation duration was calculated based on the difference between the implantation surgery date and the current date.

[0052] Risk scoring is performed on each potentially high-risk patient based on the following three dimensions: the clinical severity level of the exclusion feature, the implantation duration, and the anatomical risk level corresponding to the implantation site coding. The clinical severity level of the exclusion feature is determined by the highest severity level among all the exclusion features matched for the patient. Specifically, when a patient matches multiple exclusion features simultaneously, the severity level corresponding to each matched exclusion feature is obtained separately, and the highest value is taken as the clinical severity level. The severity of the excluded comorbidities is higher than that of the excluded age range and excluded indications.

[0053] It should be noted that the risk scoring matrix described above is a pre-configured three-dimensional lookup table structure. Its row dimension corresponds to the clinical severity level, column dimension to the implantation duration interval, and layer dimension to the anatomical risk level. Each cell stores the comprehensive score value under the corresponding combination of conditions. The risk scoring matrix is ​​used to perform lookup operations on the level values ​​of each of the three dimensions to output the comprehensive score value. Specifically, let the clinical severity level be 1. , This represents the highest severity level among all the exclusion features matched for this patient; the index of the implantation duration interval is... , This indicates the interval number into which the implantation duration falls after being divided into preset intervals; the anatomical risk level is... , The anatomical risk level value corresponding to the implantation site code is indicated; the comprehensive score is then determined by the following formula: in, This is the overall score. For a pre-configured three-dimensional rating matrix, This indicates a clinical severity level of Implantation duration interval index is The autopsy risk level is The cell value stored under the combined conditions.

[0054] Based on overall score Patients are categorized into three levels: urgent follow-up, intensive follow-up, and routine observation. Each patient's comprehensive score, risk level, and corresponding tiered follow-up tasks are combined to form a patient-level clinical risk assessment checklist.

[0055] Step 8: Summarize the results of each processing link and generate a comprehensive response report for sales management of registration certificate changes.

[0056] The following data are summarized: the compliance status of each product in the comprehensive impact product list, the handling results of each order in the tiered handling recommendation list, and the patient-level clinical risk assessment list.

[0057] The percentage of high-risk patients in each batch was calculated, specifically the ratio of the number of patients in the patient-level clinical risk assessment list who require urgent re-examination or intensive follow-up to the total number of patients in the full implantation list. The distribution of high-risk patients across different medical institutions was also calculated, specifically the number of high-risk patients corresponding to each institution after grouping by institution code.

[0058] The above-mentioned aggregated data and statistical results are combined into a comprehensive response report for sales management of registration certificate changes, and the comprehensive response report for sales management of registration certificate changes is linked to the compliance processing records of the corresponding registration certificate change events in the cloud platform.

[0059] This invention performs sentence-level segmentation and longest common subsequence comparison on the text of the registration certificate's scope of application. It decomposes the wording changes at the natural language level into pure deleted sentence segments, modified sentence segments, and restrictive qualifiers, so that the implicit scope reduction is presented in a structured form. This overcomes the factor that manual comparison relies on overall semantic understanding and is prone to missing subtle wording differences. Therefore, it solves the problem that the implicit scope reduction is missed, resulting in the products on sale not being identified in a timely manner.

[0060] This invention pushes a structured set of patient exclusion feature constraints to the data exchange nodes of various medical institutions, where each institution completes the matching within its internal system and only returns the matching status. This achieves patient-level exclusion feature screening without the sales company directly obtaining patient diagnostic data. It overcomes the limitation that cloud platforms can only track down to the medical institution level and cannot extend to specific patients. Therefore, it solves the problem that potentially high-risk implanted patients cannot be discovered and tracked in the sales management process in non-recall registration certificate change scenarios.

[0061] On the other hand, the present invention also proposes a cloud-based medical device sales management system, comprising: The semantic difference module is used to obtain the text of the applicable scope before and after the registration certificate change, perform sentence-level semantic difference, and generate the semantic difference result of the applicable scope. The knowledge graph mapping module is used to semantically map the semantic difference results of the applicable scope through the medical device classification catalog knowledge graph to generate a structured scope reduction impact description. The Product Impact Analysis module is used to generate a comprehensive list of affected products based on structured descriptions of scope reduction impacts and specification change information. The order processing module is used to retrieve incomplete orders based on the comprehensive impact product list and generate a tiered disposal suggestion list; The implantation traceability retrieval module is used to retrieve implantation traceability records based on high-value implantable products in the comprehensive impact product list and generate a full list of implanted patients. The patient screening module is used to push the patient exclusion feature constraint set to the medical institution's data exchange node, receive the matching results, and generate screening results for potentially high-risk patients. The risk assessment module is used to calculate the comprehensive clinical risk score for each potentially high-risk patient and generate a patient-level clinical risk assessment list. The report generation module is used to summarize the results of each processing link and generate a comprehensive response report for registration certificate change sales management.

[0062] By way of example, the system proposed in this invention may also include various features and combinations thereof in the method embodiments, which will not be elaborated here.

[0063] See Figures 2-9 The following application examples are proposed: A medical device sales company manages the registration certificates, product master data, and UDI traceability records for its cardiovascular implantable product line through a cloud platform. In B month of 20XX, the cloud platform received a registration certificate change notification with the number REG-CV-0317, which covered Class I coronary drug-eluting stents. The previous scope of application allowed for primary coronary artery lesions and in-stent restenosis. The changed scope added the limitation of "focal lesions confirmed by cardiology evaluation and with a reference vessel diameter of not less than 2.5 mm" to the in-stent restenosis indication, and completely removed the statement "applicable to coronary artery disease patients with diabetes." The company has orders for this product and completed implantation records in multiple medical institutions nationwide, requiring a full-chain compliance response to this change.

[0064] The cloud platform extracted the applicable scope field text from the versions of REG-CV-0317 before and after the change, performed sentence-level segmentation, and then compared the longest common subsequence. The comparison results identified one purely deleted sentence segment (the original text's description of its applicability to diabetic patients) and one modified sentence segment (the description of in-stent restenosis indication). From the modified sentence segment, two restrictive qualifiers were extracted: "confirmed as focal lesion by cardiology evaluation" (conditional restriction category) and "reference vessel diameter not less than 2.5 mm" (range narrowing category).

[0065] Table 1. Scope of Application and Semantic Difference Results The deleted segments, modified segments, and restrictive qualifiers were input into the medical device classification catalog knowledge graph. Segment 3 was used to extract the terms "coronary artery" and "diabetic complications," which were matched to product classification code C1402; segment 5 was used to extract the term "in-stent restenosis," which was also matched to C1402. The two restrictive qualifiers were mapped to the applicable condition attribute nodes "lesion morphology restriction" and "lower limit of vessel diameter," respectively. After reasoning along the association path from applicable conditions to patient characteristics, a patient exclusion feature constraint set was obtained, including exclusionary comorbidity codes (diabetic complications ICD code E11), exclusionary indication codes (diffuse lesions ISC-DIF), and exclusionary vascular parameters (reference vessel diameter less than 2.5 mm).

[0066] Table 2 Description of the impact of structured scope reduction A set difference operation was performed on the specification and model lists before and after the change to identify the specification and model SP-2.0×18, which existed in the previous version but was deleted in the new version. This specification and model was then associated with SKU code SKU-4421, forming the first set of affected SKU codes. The product master data was then associated with product category code C1402 to obtain three codes: SKU-4421, SKU-4422, and SKU-4423, forming the second set of affected SKU codes. After merging and deduplication, the overall affected SKU code set includes SKU-4421, SKU-4422, and SKU-4423. SKU-4421 is simultaneously labeled as both a specification deletion type and a range narrowing type, while SKU-4422 and SKU-4423 are labeled as range narrowing types.

[0067] Table 3 List of Products with Comprehensive Impact Using SKU-4421, SKU-4422, and SKU-4423 as search criteria, incomplete orders were retrieved, yielding 4 incomplete sales orders. Order ORD-8801, where the affected product amount exceeded the high impact threshold and no alternative products were available, was marked as urgently paused. Order ORD-8802, where the affected product amount exceeded the high impact threshold, found an unaffected SKU-4425 with the same category code that could be used as a substitute, and product replacement suggestions were generated. Orders ORD-8803 and ORD-8804, where the affected product amount did not exceed the high impact threshold and no alternative products were available, were marked as low impact and awaiting observation.

[0068] Table 4. List of Recommendations for Tiered Treatment From the comprehensive impact product list, SKU codes for products with high-value implantation attributes were selected, and the corresponding product DI codes DI-CV317 were extracted. Using DI-CV317 as the query condition, implantation traceability records on the cloud platform were retrieved, yielding a total of 6 completed implantation UDI records for this product at Institution A, Institution B, and Institution C. The anonymized patient identifier, implantation surgery date, implantation site code, and medical institution code were extracted and summarized into a complete implantation patient list.

[0069] Table 5. List of patients who received full implantation (partial) The patient exclusion feature constraint set was encapsulated into structured query conditions, grouped by medical institution code, and pushed to the data exchange nodes of Institution A, Institution B, and Institution C respectively, along with a list of anonymized patient identifiers for each institution. Each institution performed exclusion feature matching queries on the received patient identifiers in its internal system, only sending the matching status back to the cloud platform. After summarizing the returned results, PAT-1103 matched the exclusionary comorbidity E11, PAT-3381 matched the exclusionary indication ISC-DIF, and PAT-4562 matched the exclusionary vascular parameter (diameter less than 2.5mm), for a total of 3 patients included in the potential high-risk patient screening results.

[0070] Three potentially high-risk patients were linked to the full list of implanted patients to obtain their implantation surgery dates and implantation site codes. PAT-1103 was matched to the excluded comorbidity E11, with a clinical severity level. The highest value of 3 is taken; the implantation site is LAD-PROX (proximal segment of the left anterior descending artery), and the anatomical risk level is [not specified]. The value is 3; the implantation duration falls within the interval index. The value is 2. Substitute it into the formula. The comprehensive score obtained from the table is used to classify the case into the urgent re-examination level. (PAT-3381 and PAT-4562) The values ​​were low, and the overall scores fell into the intensive follow-up and routine observation levels, respectively.

[0071] Table 6 Patient-Level Clinical Risk Assessment Checklist The report summarizes the compliance status of the comprehensive impact product list (SKUs 4421 to 4423 have all been marked as impactful), the tiered handling recommendation list (4 order handling results), and the patient-level clinical risk assessment list. The report also includes the percentage of high-risk patients: 1 patient requiring emergency re-examination and 1 patient requiring intensive follow-up, totaling 2 patients, representing the total of 6 patients on the full implantation patient list. The report further groups high-risk patients by medical institution: Institution A: 1 patient (emergency re-examination), Institution B: 1 patient (intensive follow-up), and Institution C: 1 patient (routine observation). The summarized statistical results are combined into a comprehensive sales management response report for the REG-CV-0317 change event and linked to the corresponding compliance processing records on the cloud platform.

[0072] The entire data flow in the implementation process has complete and coherent linkage. Step 1 extracts structured differential results from the original registration certificate. Step 2 maps the differential results to a knowledge graph to obtain an impact description containing patient exclusion features. Step 3 uses the impact description to associate with the product master data to generate a list of affected SKUs. Step 4 retrieves orders based on the SKU list and generates treatment suggestions. Step 5 filters high-value implantable products from the same SKU list and retrieves traceability records. Step 6 pushes patient exclusion features to medical institutions to complete patient screening under the premise of privacy protection. Step 7 performs three-dimensional risk scoring on the screening results. Step 8 summarizes the output of the entire linkage to form a final report. The output of each step directly serves as the input for subsequent steps. Data identifiers (such as SKU codes, de-identified patient identifiers, and medical institution codes) remain consistent throughout the entire linkage, ensuring end-to-end data coherence from changes to the registration certificate text to patient-level risk response.

[0073] The embodiments of the present invention have been described above. However, the embodiments are not limited to the specific implementation methods described above. The specific implementation methods described above are merely illustrative and not restrictive. Those skilled in the art can make more equivalent embodiments under the guidance of the present embodiments, and all of them are within the protection scope of the present embodiments.

Claims

1. A medical device sales management method based on a cloud platform, characterized in that, include: Sentence-level segmentation is performed on the text of the applicable scope before and after the registration certificate change. The longest common subsequence algorithm is performed on the sentence segment sequence before and after the change. Purely deleted sentence segments, modified sentence segments and restrictive words are identified by position alignment and word-by-word comparison to generate semantic difference results of the applicable scope. The semantic difference results of the applicable scope are semantically mapped in the medical device classification catalog knowledge graph to generate a scope reduction impact description that includes a set of affected product classification codes, a set of applicable condition constraints, and a set of patient exclusion feature constraints. Based on the cloud platform product master data associated with the scope reduction impact description, a comprehensive list of affected products is generated. High-value implantable products are selected from the comprehensive impact product list, and implantation traceability records are retrieved from the cloud platform to generate a full list of implanted patients. The patient exclusion feature constraint set is pushed to the medical institution's data exchange node, where each institution performs exclusion feature matching and returns the matching status to generate screening results for potentially high-risk patients.

2. The medical device sales management method based on a cloud platform according to claim 1, characterized in that, The method of identifying purely deleted segments, modified segments, and restrictive qualifiers through position alignment and word-by-word comparison includes: Segments in the original segment sequence that do not belong to the set of longest common sub-segments are marked as deleted segments, and segments in the original segment sequence that do not belong to the set of longest common sub-segments are marked as newly added segments. The deleted and newly added sentence segments are aligned according to their positions in the original sequence. The deleted sentence segments that can establish a positional correspondence are paired with the newly added sentence segments. Each pair is compared word by word. The newly added sentence segments in the pair are identified as modified sentence segments, and the deleted sentence segments that cannot establish a positional correspondence are retained as pure deleted sentence segments. For each modified sentence segment, extract the newly added words relative to the corresponding unmodified sentence segment to obtain a set of restrictive words.

3. The medical device sales management method based on a cloud platform according to claim 2, characterized in that, The restrictive qualifiers include conditional qualifiers and scope-narrowing qualifiers; wherein, the conditional qualifiers are new qualifier phrases with added preconditions added to the modified sentence segment; the scope-narrowing qualifiers are new words added to narrow the coverage when a generalized expression is replaced with a specialized expression in the modified sentence segment.

4. The medical device sales management method based on a cloud platform according to claim 1, characterized in that, The nodes of the medical device classification catalog knowledge graph include product classification code nodes, scope of application description nodes, applicable condition attribute nodes, and patient characteristic nodes, and the nodes are connected by semantic relationship edges; the step of performing semantic mapping on the semantic difference results of the scope of application in the medical device classification catalog knowledge graph includes: After extracting the medical device terms from each deleted and modified sentence segment, matching them with the corresponding product classification code nodes, and summarizing them, we obtain the set of affected product classification codes. Match each restrictive qualifier with its corresponding applicable condition attribute node to obtain the set of applicable condition constraints. Based on each constraint in the applicable condition constraint set, reasoning is performed along the association path from the applicable condition node to the patient feature node to obtain the patient exclusion feature constraint set, which includes exclusionary comorbidity codes, exclusionary age ranges, and exclusionary indication codes.

5. The medical device sales management method based on a cloud platform according to claim 1, characterized in that, The process of generating a comprehensive list of affected products based on the cloud platform product master data related to the scope reduction impact description includes: Perform a set difference operation on the list of specifications and models before and after the registration certificate change to identify the set of specifications and models that were deleted; Each specification in the deleted specification set is matched with the SKU code in the cloud platform product master data according to the specification field to obtain the first affected SKU code set; each code in the affected product category code set is matched with the SKU code in the cloud platform product master data according to the product category code field to obtain the second affected SKU code set. A merge and deduplication operation is performed on the first affected SKU code set and the second affected SKU code set to obtain a comprehensive affected SKU code set; wherein, the labels from the first affected SKU code set are of the specification deletion type, and the labels from the second affected SKU code set are of the range restriction type.

6. The medical device sales management method based on a cloud platform according to claim 1, characterized in that, Also includes: Based on the comprehensive impact product list, sales orders with an incomplete status are retrieved; for each incomplete order, the ratio of the amount of the affected products to the total order amount is calculated as the impact percentage, and available SKU codes with the same category code that are not in the comprehensive impact product list are queried as alternative products; When the impact percentage is greater than the preset high impact threshold and there is no alternative product, the order is marked as urgently suspended; when there is an alternative product, a product replacement suggestion containing the SKU code of the alternative product and the available inventory is generated; when the impact percentage is not greater than the preset high impact threshold and there is no alternative product, the order is marked as low impact pending observation; and a tiered handling suggestion list is compiled.

7. The medical device sales management method based on a cloud platform according to claim 1, characterized in that, The step of pushing the patient exclusion feature constraint set to the medical institution's data exchange node includes: The patient exclusion feature constraint set is encapsulated into structured query conditions, which include an exclusion comorbidity code list, an exclusion age range, and an exclusion indication code list. The records in the full list of implanted patients are grouped according to the medical institution codes. For each medical institution involved, the structured query conditions and the list of de-identified patients corresponding to that institution are pushed to the data exchange node of that medical institution. For each patient in the received desensitized patient identifier list, the data exchange nodes of each medical institution determine whether the patient has any comorbidity in the exclusion comorbidity code list, whether the patient is within the exclusion age range, or whether the patient has any indication in the exclusion indication code list, and then send the matching status back to the cloud platform.

8. The medical device sales management method based on a cloud platform according to claim 1, characterized in that, Also includes: The screening results of the potential high-risk patients are associated with the full list of implanted patients according to the desensitized patient identifiers to obtain the implantation surgery date and implantation site code for each potential high-risk patient; The implantation duration is calculated based on the difference between the implantation surgery date and the current date; A lookup table operation is performed on each potentially high-risk patient using a pre-configured three-dimensional scoring matrix. The row dimension of the three-dimensional scoring matrix corresponds to the clinical severity level, the column dimension corresponds to the implantation duration interval, and the layer dimension corresponds to the anatomical risk level. The comprehensive score is determined by the following formula: in, This is the overall score. It is a three-dimensional rating matrix. This is the highest severity level value among all the exclusion features matched for this patient. The interval number to which the implantation duration falls after being divided into preset intervals. The anatomical risk level value is assigned to the corresponding code of the implantation site; According to the comprehensive score value The intervals divide patients into three levels: emergency re-examination, intensive follow-up, and routine observation, generating a patient-level clinical risk assessment list.

9. The medical device sales management method based on a cloud platform according to claim 8, characterized in that, The clinical severity level The determination method is as follows: when a patient matches multiple exclusion features simultaneously, the severity level corresponding to each matching exclusion feature is obtained, and the highest value is taken as the clinical severity level. Among them, the severity of excluded comorbidities is higher than that of excluded age ranges and excluded indications.

10. A cloud-based medical device sales management system, used to execute the cloud-based medical device sales management method according to any one of claims 1 to 9, characterized in that, include: The semantic difference module is used to perform sentence-level segmentation on the text of the applicable scope before and after the registration certificate change, and to perform the longest common subsequence algorithm on the sentence segment sequence before and after the change. It identifies purely deleted sentence segments, modified sentence segments and restrictive words through position alignment and word-by-word comparison, and generates semantic difference results of the applicable scope. The knowledge graph mapping module is used to perform semantic mapping on the semantic difference results of the applicable scope in the medical device classification catalog knowledge graph to generate a scope reduction impact description that includes a set of affected product classification codes, a set of applicable condition constraints, and a set of patient exclusion feature constraints. The product impact analysis module is used to generate a comprehensive list of impacted products based on the cloud platform product master data associated with the scope reduction impact description. The implantation traceability retrieval module is used to filter high-value implantable products from the comprehensive impact product list and retrieve implantation traceability records from the cloud platform to generate a full list of implanted patients. The patient screening module is used to push the patient exclusion feature constraint set to the medical institution's data exchange node, whereby each institution performs exclusion feature matching and returns the matching status to generate screening results for potentially high-risk patients.

Images