A pre-pregnancy carrier screening genetic counseling system based on a large language model

Abstract

The disclosure provides a pre-pregnancy carrier screening genetic counseling system based on a large language model, applied to the technical field of gene sequencing analysis, comprising a variation acquisition module for screening the user's to-be-sequenced gene data to obtain pathogenic abnormal site information, and converting the pathogenic abnormal site information, disease clinical phenotype, family history and personal past history into a high-dimensional vector; a variation retrieval module for dynamically retrieving the high-dimensional vector based on a semantic vector index database and recalling Top-K medical data in the semantic vector index database; an inference analysis module for evaluating the reproductive genetic variation risk according to the Top-K related medical data and the high-dimensional vector to obtain an evaluation result; and a multi-round dialogue module, through which the user can obtain the evaluation result through multi-round dialogue. The advantage of the disclosure is that the reproductive risk, intervention suggestions and other contents can be automatically generated, and the disclosure can also interact with the patient through dialogue, thereby reducing the clinical genetic counseling burden.

Description

Technical Field

[0001] This disclosure relates to the field of gene sequencing analysis technology, and in particular to a genetic counseling system for preconception carrier screening based on a large language model. Background Technology

[0002] With the rapid development of gene sequencing technology, preconception carrier screening plays a crucial role in preventing genetic diseases. Currently, various carrier screening products are available both domestically and internationally, primarily based on next-generation sequencing technology combined with the detection of specific genes to help couples understand their risk of carrying recessive genetic diseases. Carrier gene screening and genetic counseling, as important components of modern reproductive health management, aim to help individuals and families understand their risk of carrying genetic disease genes and make informed reproductive decisions accordingly.

[0003] Some domestic gene testing companies have launched carrier screening products, but their main shortcomings or drawbacks include the following:

[0004] 1) Most existing products target hot genes and hot variants, which may result in a high rate of missed detections;

[0005] 2) Limited gene coverage, and many products do not incorporate genetic variation databases specific to the Chinese population. Existing products largely rely on publicly available databases or international data, lacking efficient detection capabilities for variations specific to the Chinese population. While some products attempt to incorporate Chinese population data, they still have shortcomings in gene selection, detection methods, and clinical applicability.

[0006] 3) With the development of gene testing technology, the scope of screening is constantly expanding, and the demand for and challenges of genetic counseling are also increasing.

[0007] Therefore, there is an urgent need to develop a preconception carrier screening genetic counseling system based on a large-scale language model to solve the above problems. Summary of the Invention

[0008] This disclosure provides a genetic counseling system for preconception carrier screening based on a large-scale language model to solve the above-mentioned problems, and addresses the issues of limited gene detection, limited gene coverage, insufficient manpower for genetic counseling, and low efficiency in the prior art.

[0009] According to a first aspect of this disclosure, a genetic counseling system for preconception carrier screening based on a large-scale language model is provided. The system includes: a variant acquisition module for screening a user's gene data to be sequenced, obtaining pathogenic abnormality site information, and converting the pathogenic abnormality site information, disease clinical phenotype, family history, and personal medical history into a high-dimensional vector;

[0010] A mutation retrieval module, connected to the mutation acquisition module, is used to dynamically retrieve and recall Top-K medical data from the semantic vector index database based on the high-dimensional vector; where K is a non-zero natural number.

[0011] The reasoning and analysis module, connected to the variant retrieval module, is used to assess the risk of reproductive genetic variants based on the Top-K related medical data and high-dimensional vectors, and obtain the assessment results;

[0012] A multi-turn dialogue module is connected to the reasoning and analysis module, allowing users to obtain evaluation results and their interpretations through multi-turn dialogues.

[0013] Furthermore, the pathogenic abnormal site information is obtained through the following steps:

[0014] High-frequency variant genes are obtained to construct a gene library, and the target genes to be tested are screened out from the gene library;

[0015] The target gene region in the user's gene data to be sequenced is captured using a probe set.

[0016] The target gene region is automatically annotated with bioinformatics and interpreted using a genetic variation classification guide to obtain pathogenic abnormal sites;

[0017] Based on the identified pathogenic abnormal sites, gene names are confirmed and labeled to obtain pathogenic abnormal site information; wherein,

[0018] The probe set includes SEQ ID NO: 1-277.

[0019] Furthermore, the construction process of the semantic vector index database includes:

[0020] Collect multi-source medical data;

[0021] The medical data is automatically segmented into text blocks;

[0022] All text blocks are converted into vectors using an embedding encoding model to obtain a semantic vector index database; wherein the embedding encoding model is the BGE model.

[0023] Furthermore, the mutation retrieval module is used to perform the following steps:

[0024] Calculate the semantic similarity between the high-dimensional vector and the medical data in the semantic vector index database, and recall the K medical data with the highest semantic similarity to obtain the Top-K medical data.

[0025] Furthermore, the reasoning analysis model includes a large language model, which contains a preset genetic counseling reasoning chain. The genetic counseling reasoning chain is embedded in the model in the form of a rule chain or a Prompt template.

[0026] Furthermore, the genetic counseling reasoning chain includes:

[0027] Based on the aforementioned Top-K related medical data and pathogenic abnormal site information, the disease inheritance pattern is determined, and then the reproductive risk is assessed in combination with the disease clinical phenotype, family history and personal medical history.

[0028] Furthermore, the assessment results include fertility risk assessment and intervention recommendations;

[0029] The fertility risk assessment includes theoretical fertility risk calculated based on genetic patterns, as well as individual fertility risk adjusted for family history and personal history.

[0030] The intervention recommendations include emergency intervention recommendations, routine monitoring recommendations, long-term health management recommendations, and assisted reproductive recommendations, prioritized by level.

[0031] Further, determine whether both spouses have information on pathogenic abnormal loci on the same autosomal recessive gene, or whether the woman carries information on pathogenic abnormal loci on a sex-linked gene; if so, proceed to the clinical solution recommendation module.

[0032] Furthermore, the clinical solution recommendation module is used to perform the following steps:

[0033] The user's clinical resolution assessment score is calculated based on multiple dimensions;

[0034] Determine whether the clinical resolution assessment score is greater than or equal to the first preset value. If yes, proceed to the PGT-M assisted reproduction process; otherwise, proceed to the next step.

[0035] Determine whether the clinical resolution assessment score is less than the second preset value. If the second preset value is less than the first preset value, proceed to the natural pregnancy process; otherwise, proceed to the routine pre-pregnancy counseling process.

[0036] Genetic testing is performed in the PGT-M assisted reproduction process, natural pregnancy process, and routine preconception counseling process. If the test results are normal, the pregnancy continues; otherwise, the pregnancy is terminated or genetic counseling guidance is obtained.

[0037] Furthermore, it also includes a structure output module, used to generate and output a structured report from the evaluation results.

[0038] The beneficial effects of this disclosure are:

[0039] This disclosure integrates gene testing, semantic retrieval, large language model reasoning, and multi-turn dialogue technology to construct a fully automated system. Gene testing effectively detects complex variants, reducing missed and false positives, providing a more accurate genetic disease risk assessment tool to meet actual clinical needs. It also correlates with individual and family history, clinical manifestations, and constructs a semantic vector index database to ensure accurate interpretation rather than relying on empirical advice. The genetic counseling reasoning chain set by the large language model avoids subjective assumptions, especially for interpreting variants of unknown significance. The multi-turn dialogue module allows users to deeply understand the assessment results, and the system provides transparent explanations based on the genetic counseling reasoning chain, improving patient acceptance of the recommendations.

[0040] The genetic counseling system of this invention can improve the efficiency of genetic counseling in prenatal carrier screening and reduce the manual burden; by accurately combining knowledge base content, it provides the accuracy of genetic counseling, and can also perform automated report generation and conversational risk explanation, enhancing the patient's service experience;

[0041] This disclosure helps couples understand the risks of genetic diseases through precise carrier screening, providing a scientific basis for eugenics and reducing the birth rate of children with genetic diseases; it also provides early identification of health risks for some examinees, such as hereditary heart disease and tumors, and provides early warnings and timely prevention of related diseases and pregnancy risks.

[0042] It should be understood that the description in the Summary of the Invention is not intended to limit the key or essential features of the embodiments of this disclosure, nor is it intended to restrict the scope of this disclosure. Other features of this disclosure will become readily apparent from the following description. Attached Figure Description

[0043] The above and other features, advantages, and aspects of the embodiments of this disclosure will become more apparent from the accompanying drawings and the following detailed description. The drawings are provided for a better understanding of the invention and are not intended to limit the scope of this disclosure. In the drawings, the same or similar reference numerals denote the same or similar elements, wherein:

[0044] Figure 1 A framework diagram of a preconception carrier screening genetic counseling system based on a large language model, provided in an embodiment of this disclosure, is shown.

[0045] Figure 2 This diagram illustrates the steps for obtaining pathogenic abnormal site information according to an embodiment of the present disclosure.

[0046] Figure 3 This diagram illustrates a comparison of the detection coverage of the screening kit provided in this embodiment with that of other screening kits.

[0047] Figure 4A schematic diagram showing the comparison results of Fold80 values ​​detected by the screening kit provided in this embodiment of the present disclosure and Fold80 values ​​detected by other screening kits is shown.

[0048] Figure 5 A schematic diagram illustrating the detection results with the addition of the AR gene probe provided in an embodiment of this disclosure is shown.

[0049] Figure 6 This illustration shows the detection results after adding probes to the full length of the HBA1 and HBA2 genes, as well as to the intergenic regions and upstream and downstream regions, according to an embodiment of this disclosure.

[0050] Figure 7 This diagram illustrates the detection results after adding probes to exon7 of the SMN1 gene and its upstream and downstream regions, as provided in an embodiment of this disclosure.

[0051] Figure 8 This diagram illustrates the detection results after adding probes to the intron region of the DMD gene, as provided in an embodiment of this disclosure.

[0052] Figure 9 A flowchart of the clinical solution recommendation module provided in an embodiment of this disclosure is shown. Detailed Implementation

[0053] To make the objectives, technical solutions, and advantages of the embodiments of this disclosure clearer, the technical solutions of the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this disclosure, and not all embodiments. Based on the embodiments of this disclosure, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this disclosure.

[0054] Furthermore, the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.

[0055] This disclosure provides a genetic counseling system for preconception carrier screening based on a large language model, see [link to relevant documentation]. Figure 1 The system specifically includes the following modules:

[0056] The variant acquisition module 110 is used to screen the user's gene data to be sequenced, obtain pathogenic abnormal site information, and convert the pathogenic abnormal site information, disease clinical phenotype, family history and personal history into a high-dimensional vector.

[0057] Specifically, see Figure 2The pathogenic abnormal site information is obtained through the following steps:

[0058] S11. Obtain high-frequency variant genes to construct a gene library, and screen out the target to be detected from the gene library.

[0059] In this disclosure, high-frequency variant genes are statistically obtained from data of 8,000 East Asians in the gnomAD database and data from 5,000 samples accumulated by a research group. Furthermore, a large number of population databases and internal hospital data can be integrated to make the test results more consistent with the characteristics of ethnic groups and meet actual clinical needs.

[0060] S12. Capture the target gene region in the gene data to be sequenced using probe sets.

[0061] Genetic variation information obtained through preconception carrier screening kits, wherein the probe set in the screening kits includes SEQ ID NO: 1-277, where 1-277 represents an internationally recognized sequence identifier number used to uniquely identify a specific nucleic acid sequence.

[0062] The probe set of the present invention was used to capture gene mutations related to different single-gene genetic diseases, as shown in Table 1.

[0063] Table 1

[0064]

[0065] The 22 diseases mentioned above correspond to probe sequences SEQ ID NO: 1-277, and their corresponding chromosomal locations, in order from 1 to 277, as follows:

[0066] ALDH5A1 Chr6:24520597-24520709, 24503450-24503561, 24532290-24532401, 24505040-24505151, 24533817-24533928, 244952 44-24495355, 24528222-24528333, 24522937-24523048, 24515368-24515479, 24512042-24512153, 24502715-24502829;

[0067] AR ChrX:66943477-66943588、66905799-66905910、66941619-66941730、66766395-66766506、66863111-66863222、66942608-66942719、66937263-66937374、66931188-66931299、66765237-66765348;

[0068] ATP6V0A4 Chr7:138453469-138453580、138437310-138437421、138453870-138453982、138418823-138418934、138424292-138424403、138433856-138433967、138417603-138417714、138447634-138447749、138432120-138432234、138406589-138406700、138440378-138440489、138400453-138400564、138447125-138447236、138394323-138394434、138429811-138429922、138413473-138413584、138455833-138455944;

[0069] COG6 Chr13:40297502-40297613、40253621-40253732、40293860-40293971、40256284-40256395、40229920-40230031、40268715-40268826、40239196-40239307、40254081-40254195、40251632-40251743、40273559-40273670、40234891-40235002、40233471-40233582、40263894-40264005;

[0070] COL6A3 Chr2:238283174-238283285, 238234257-238234368, 238249223-238249334, 238253645-238253756, 238256400-238256511, 238274285-238274396, 238303701-238303812, 238270327-238270438, 238287781-238287892, 238243250-2 38243361, 238269738-238269849, 238244722-238244833, 238290007-238290118, 238271912-238272023, 238263510-238263621, 238277152-238277263, 238275739-238275850, 238249428-238249539, 238267840-238267951, 2382421 32-238242243, 238303295, 238303406, 238305330, 238305441, 238275295, 238275406, 238285810, 238285921, 238258755, 238258866, 238245054, 238245165, 238280319, 238280430, 238252940, 238253051, 238287290, 238287401, 238 289560-238289671, 238268710-238268821, 238257196-238257307, 238296617-238296728, 238296248-238296359, 238267120-238267231, 238250698-238250809, 238259747-238259858, 238285442-238285556, 238247689-238247800;

[0071] CUL7 Chr6:43010794-43010909、43013670-43013785、43019099-43019210、43011191-43011302、43017641-43017763、43019348-43019459、43006340-43006451、43018740-43018851、43008329-43008440、43013269-43013380、43006632-43006743、43020053-43020165、43016165-43016276、43015836-43015947、43017267-43017378、43012960-43013071、43005950-43006064、43008642-43008765、43008018-43008129、43012478-43012589、43010484-43010595;

[0072] DPYS Chr8:105478841-105478952、105459532-105459643、105441763-105441874、105463490-105463601、105436438-105436549、105393430-105393541、105404952-105405063、105440167-105440278;

[0073] ERCC5 Chr13:103524512-103524623、103513800-103513911、103508405-103508516、103525564-103525675、103517980-103518091、103514363-103514475、103504588-103504699、103506109-103506220、103506608-103506719、103518938-103519049、103520478-103520590、103515299-103515410;

[0074] FREM2 Chr13:39446918-39447029、39261825-39261936、39265187-39265298、39450453-39450564、39422559-39422672、39265822-39265933、39271870-39271981、39264115-39264226、39425072-39425183、39338431-39338542、39262173-39262284、39264528-39264640、39263632-39263743、39266587-39266698、39262692-39262803、39435535-39435646、39358824-39358935、39433630-39433742、39261439-39261550、39424124-39424235、39263356-39263467;

[0075] GTPBP3 Chr19:17449171-17449282、17449411-17449523、17448895-17449006、17449759-17449870、17450013-17450124、17451853-17451964;

[0076] HPRT1 ChrX:133609161-133609272、133607353-133607464、133634006-133634117、133620438-133620549、133632363-133632474、133627510-133627621;

[0077] IL10RA Chr11:117866316-117866427、117860214-117860325、117864736-117864847、117859041-117859152、117863971-117864095;

[0078] IL36RN Chr2:113818372-113818483、113819757-113819868、113820069-113820180、113816990-113817101;

[0079] MOCS2 Chr5:52402880-52403001、52397956-52398067、52404404-52404515、52394404-52394516、52405502-52405613、52397133-52397244;

[0080] MPZL2 Chr11:118133194-118133305、118133596-118133707、118130835-118130946;

[0081] MYO5B Chr18:47421324-47421435、47566483-47566595、47432818-47432929、47455855-47455966、47398535-47398646、47489321-47489432、47383189-47383300、47404199-47404310、47365546-47365657、47480720-47480831、47563310-47563421、47402037-47402148、47373554-47373665、47405346-47405457、47363025-47363136、47463604-47463715、47379817-47379928、47511081-47511192、47389613-47389724、47364065-47364176、47462565-47462676、47390476-47390587、47500684-47500795、47429079-47429190、47369708-47369819、47406770-47406881、47479607-47479718、47506794-47506905;

[0082] OTOF Chr2:26700023-26700134、26700538-26700649、26683472-26683583、26684984-26685095、26781309-26781420、26688525-26688636、26703598-26703709、26717813-26717924、26689999-26690110、26695331-26695442、26696085-26696196、26698725-26698836、26691275-26691386、26702138-26702249、26697422-26697533、26725154-26725265、26689544-26689655、26705231-26705357、26707320-26707431、26686920-26687031、26699724-26699835、26750712-26750823、26702397-26702508、26741821-26741932、26683710-26683821、26698249-26698360、26703015-26703126、26683016-26683127、26706274-26706385、26696812-26696923、26684511-26684622、26680922-26681033、26686309-26686420、26687681-26687792;

[0083] POR Chr7:75614886-75614997、75612811-75612922、75614222-75614333、75615426-75615537、75583445-75583556、75611485-75611596;

[0084] RSPH4A Chr6:116937742-116937853、116943879-116943990、116948824-116948936、116951556-116951667、116950718-116950829、116944110-116944224;

[0085] SASS6 Chr1:100588746-100588857, 100585947-100586058, 100573014-100573125;

[0086] SBDS Chr7:66456068-66456179, 66459142-66459253, 66453403-66453514, 66458154-66458265, 66460230-66460341

[0087] SERAC1 Chr6:158538756-158538867, 158534416-158534543, 158569886-158570000, 158571 429-158571540, 158535803-158535914, 158549134-158549245, 158579320-1585794 31. 158541409-158541520, 158567804-158567915, 158551425-158551536, 158564073-158564184, 158540047-158540158, 158565392-158565503, 158537160-158537271.

[0088] See Figure 3 The probe set disclosed herein exhibits superior performance in capturing the target gene coding sequence and its upstream and downstream 25bp regions, ensuring complete detection of the target gene and its surrounding regulatory elements, greatly reducing detection blind spots, and laying a solid foundation for subsequent gene analysis and interpretation. This makes the probes disclosed herein highly competitive in the field of gene detection and can meet the needs of scientific research and clinical practice for high-precision and comprehensive gene detection.

[0089] See Figure 4 The sequencing depth uniformity of the target gene coding sequence and the corresponding upstream and downstream 25bp regions captured by the probe set in the screening kit reached an extremely high level, with a significantly lower Fold80 value compared to other screening kits, indicating a more uniform distribution of sequencing data.

[0090] Therefore, this probe set can achieve stable and reliable sequencing depth across the entire target region, including both the target gene coding sequence and the upstream and downstream 25bp regions. This high uniformity not only improves the accuracy of sequencing results but also enhances the ability to detect low-frequency variations, ensuring that potentially important genetic information is not missed due to depth fluctuations.

[0091] Different probes can be added to meet different needs. For example, see the following:

[0092] (1) See Figure 5 By adding probes to specific genes, it is possible to capture deep intron regions and cover pathogenic variants that cannot be detected by other screening kits, such as the AR gene.

[0093] (2) See Figure 6 In the α-thalassemia region, probes were added to the full length of the HBA1 and HBA2 genes, as well as to intergenic regions and upstream and downstream regions, enabling precise analysis of common deletion variants, such as -α. 3.7 -α 4.2 -- SEA .

[0094] (3) See Figure 7 Probe encryption was performed on exon7 and its upstream and downstream intron regions of the SMN1 gene to improve the sequencing depth and coverage of this region, thereby enabling accurate analysis of the deletion status of exon7 in the SMN1 gene, which is of great significance for the detection of spinal muscular atrophy.

[0095] (4) See Figure 8 By adding probes to the intron regions of the DMD gene, it is possible to comprehensively analyze the exon deletions and duplications of the DMD gene, providing more accurate support for the diagnosis of Duchenne muscular dystrophy.

[0096] In one specific embodiment, this disclosure employs targeted capture technology to detect target gene regions. Based on statistical analysis of measured data from more than 200 samples, the coverage rate of target gene regions with sequencing depths greater than or equal to 30× is as high as 99%, and the coverage rate of pathogenic mutation sites with sequencing depths greater than or equal to 50× is as high as 99%.

[0097] In summary, based on the above structural design, it is possible to specifically bind to the high-frequency variant regions of 22 genetic disease-related genes, and accurately enrich the target gene regions through targeted capture technology.

[0098] S13. Perform automatic bioinformatics annotation and genetic variation interpretation on the target gene region to obtain pathogenic abnormal sites.

[0099] Automated bioinformatics annotation, or automated bioinformatics workflow, compares, filters, analyzes, annotates, and interprets genetic variations of the target gene region with the reference genome, ultimately screening out all pathogenic abnormal sites that meet the criteria.

[0100] S14. Label the gene names corresponding to the pathogenic abnormal sites.

[0101] The gene names of the pathogenic abnormal sites are labeled to accurately locate the target in subsequent analysis, avoiding interpretation errors caused by ambiguous site location.

[0102] Clinical manifestations refer to the observable or measurable symptoms, signs, and behavioral abnormalities exhibited by patients due to disease or pathological conditions. They are an important basis for doctors to diagnose diseases, assess conditions, and formulate treatment plans.

[0103] Family history refers to a confirmed hereditary disease in the family. Individual past medical history information includes whether the woman has experienced two or more miscarriages, repeated embryo transfer failures, or fertility disorders in the male or female partner (such as fallopian tube factors in the woman or oligospermia or asthenospermia in the man).

[0104] Information on pathogenic abnormal sites, clinical phenotypes of diseases, family history, and personal medical history are all natural language content, which needs to be vectorized for subsequent processing.

[0105] The mutation retrieval module 120, connected to the mutation acquisition module 110, is used to dynamically retrieve and recall Top-K medical data from the semantic vector index database based on the high-dimensional vector; where K is a non-zero natural number.

[0106] The semantic vector index database is used for knowledge augmentation interpretation to improve the system performance of natural language processing systems. Specifically, the construction process of the semantic vector index database includes:

[0107] Collect multi-source medical data;

[0108] The medical data is automatically segmented into text blocks;

[0109] All text blocks are converted into vectors using an embedding encoding model to obtain a semantic vector index database; wherein the embedding encoding model is the BGE model.

[0110] The medical data includes medical guidelines (such as the ACMG guidelines and the WHO prenatal screening guidelines), standard clinical pathway texts (such as the operating procedures for prenatal genetic clinics), and genetic disease knowledge bases (such as the ClinVar database, the OMIM database, and the pubMed database), integrating multi-source and authoritative medical data.

[0111] Medical data is automatically segmented into text blocks, ensuring that each text block is between 512 and 1024 tokens in length. Then, an embedding encoding model converts all text blocks into vectors for easy comparison with higher-dimensional vectors.

[0112] In one specific embodiment, the semantic similarity between the high-dimensional vector and the medical data in the semantic vector index database is calculated, and the K medical data with the highest semantic similarity are retrieved to obtain the Top-K medical data. For example, an index between the high-dimensional vector and the semantic vector index database can be established using tools such as FAISS to achieve efficient semantic segment retrieval.

[0113] For example, if a user detects that they carry a BRCA1 gene variant, FAISS will recall the following content related to BRCA1 from the semantic vector index database: pathogenicity ratings in ClinVar, associated diseases in OMIM, and the latest research in PubMed.

[0114] In summary, by dynamically introducing information from external knowledge bases, the accuracy, timeliness, and interpretability of large language models in knowledge-intensive tasks can be improved.

[0115] The reasoning and analysis module 130, connected to the variation retrieval module 120, is used to assess the risk of reproductive genetic variations based on the Top-K related medical data and high-dimensional vectors, and obtain the assessment results.

[0116] In one embodiment, the reasoning analysis model includes a large language model, which contains a pre-defined genetic counseling reasoning chain embedded in the model as a rule chain or a Prompt template. The large language model can be obtained by training GPT-4, ChatGLM, or BioGPT, etc.

[0117] By inputting the Top-K related medical data and high-dimensional vectors into the large language model, the knowledge limitations of the large language model are resolved, ensuring that the input to the large language model contains the latest and most authoritative external knowledge.

[0118] The direct generation of consultation conclusions by large language models may lead to problems such as knowledge bias and logical jumps. Therefore, this disclosure sets up a genetic counseling reasoning chain in the large language model, which is preset by experts and embedded into the large language model. This prompts the large language model to reason step by step to simulate the clinical counseling process, such as performing steps such as genetic pattern judgment, database evidence retrieval, risk assessment, and giving suggestions, and outputting consultation conclusions to improve the transparency and interpretability of the conclusions.

[0119] In one specific embodiment, the genetic counseling reasoning chain includes: determining the disease inheritance pattern based on the Top-K related medical data and pathogenic abnormality site information, and then assessing fertility risks and providing intervention suggestions in combination with the disease clinical phenotype, family history and personal history.

[0120] When the large language model identifies a variant site, the system automatically links to the backend database or internal knowledge base to retrieve information about the corresponding variant and generates intermediate inference text for the large language model to reference.

[0121] The backend knowledge database refers to a locally accumulated genetic disease knowledge base, containing information on all known genetic diseases, genetically determined traits, and their genes. The large language model retrieves medically similar semantics from the index database, including information on diseases, genes, relationships between diseases and genes, disease inheritance patterns, and locus pathogenicity assessments, and references them.

[0122] The assessment results include fertility risk assessment and intervention recommendations;

[0123] The fertility risk assessment includes theoretical fertility risk calculated based on genetic patterns, as well as individual fertility risk adjusted for family history and personal history.

[0124] The intervention recommendations include emergency intervention recommendations, routine monitoring recommendations, long-term health management recommendations, and assisted reproductive recommendations, prioritized by level.

[0125] The multi-turn dialogue module 140 is connected to the reasoning and analysis module 130, and the user can obtain the evaluation results and their interpretations through multi-turn dialogue.

[0126] The evaluation results generated by the large language model can be output through a multi-turn dialogue module. Based on the evaluation results, users can interact with the large language model in the multi-turn dialogue module using natural language to obtain more explanations and other suggestions about the consultation conclusions.

[0127] In summary, based on the genetic information input by the user, the large-scale language model combines information such as the severity of the disease phenotype, the level of genetic variation sites, reproductive risk indicators, and the patient's self-selection, and automatically outputs the assessment results based on the reasoning logic chain.

[0128] In addition, this disclosure also includes: determining whether both spouses have pathogenic abnormality information on the same autosomal recessive gene, or determining whether the female carries pathogenic abnormality information on a sex-linked gene; if so, proceeding to the clinical solution recommendation module; if either of the two determination conditions is true, the clinical solution recommendation module is considered to be activated.

[0129] Based on the test results, the risk level of the couples is identified. Those with a high risk level will enter the intervention clinical solution recommendation module to develop a personalized clinical solution based on the couple's actual fertility situation and wishes.

[0130] Specifically, see Figure 9 This includes the following steps:

[0131] The clinical resolution assessment score for users is calculated based on multiple dimensions; for example, the scoring is performed with reference to the score settings in Table 2.

[0132] Table 2 Reference Table for Clinical Resolution Assessment

[0133]

[0134] Determine whether the clinical resolution assessment score is greater than or equal to the first preset value. If so, proceed to the PGT-M assisted reproduction process; otherwise, proceed to the next step.

[0135] For example, the first preset value is 7 points. When the clinical resolution assessment score is ≥7 points, the patient enters the PGT-M (Preimplantation Genetic Testing for Monogenic Disorders) assisted reproductive process, which is a preimplantation genetic testing (PGD) assisted reproductive technology process. Specifically, it uses assisted reproductive technology to select embryos without pathogenic mutations for transfer. The PGT-M assisted reproductive process includes the following steps:

[0136] In a laboratory setting, a woman's egg and a man's sperm combine to form a fertilized egg. After culturing the fertilized egg for several days, in vivo genetic testing is performed. Typically, a very small number of cells are used as the test sample, focusing on detecting known pathogenic mutations in both partners to determine if the embryo carries these mutations. If it does, the embryo is excluded from implantation, and further assisted reproductive technology (ART) is discussed, or the embryo is abandoned. Otherwise, a single embryo transfer is performed, and prenatal diagnosis is conducted after pregnancy to confirm the fetus's genetic status and avoid missed detections due to errors in the in vivo testing sample. PGT-M assisted reproduction blocks the possibility of genetic transmission at its source.

[0137] Determine whether the clinical resolution assessment score is less than a second preset value. If the second preset value is less than the first preset value, proceed to the routine preconception counseling process; otherwise, proceed to the natural pregnancy process.

[0138] For example, the second preset value is 4 points. If the score is less than 4 points, it means that the risk of genetic diseases is small, so you can proceed to the regular pre-pregnancy counseling process and only need to consult some routine pre-pregnancy guidance.

[0139] If the score is greater than or equal to 4 and less than 7, it indicates that the genetic risk is moderate, and the couple can proceed with the natural pregnancy process. The couple can conceive naturally without the need for assisted reproductive technology. However, after successful conception, they must strictly follow the requirements for prenatal diagnosis starting from the 12th week of pregnancy. That is, preliminary screening can be carried out through non-invasive prenatal testing around the 12th week of pregnancy.

[0140] Genetic testing is performed in the PGT-M assisted reproduction process, natural pregnancy process, and routine preconception counseling process. If the test results are normal, the pregnancy continues; otherwise, the pregnancy is terminated or genetic counseling guidance is obtained.

[0141] If the results indicate abnormalities, further diagnostic tests such as chorionic villus sampling or amniocentesis are needed to determine whether the fetus carries the target pathogenic mutation. Prenatal monitoring should be strengthened throughout the pregnancy to ensure timely detection of fetal abnormalities and take appropriate measures.

[0142] In addition, this disclosure also includes a structure output module for generating and outputting a structured report of the evaluation results for repeated review of the relevant content.

[0143] Based on the above technical solutions, this disclosure helps couples understand the risks of genetic diseases through accurate carrier screening, providing a scientific basis for eugenics and reducing the birth rate of children with genetic diseases; it also provides early identification of some health risks of the examinees, such as hereditary heart disease and tumors, and provides early warning of the examinees' subsequent related disease risks and pregnancy risks, enabling timely prevention.

[0144] It should be noted that, for the foregoing embodiments, for the sake of simplicity, they are all described as a series of actions. However, those skilled in the art should understand that this disclosure is not limited to the described order of actions, because according to this disclosure, some steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also understand that the embodiments described in the specification are all optional embodiments, and the actions and modules involved are not necessarily essential to this disclosure.

[0145] The acquisition, storage, and application of user personal information involved in the technical solution disclosed herein comply with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0146] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working process of the described module can be referred to the corresponding process in the foregoing embodiments, and will not be repeated here.

[0147] The electronic devices described in this disclosure are intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic devices may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the disclosure described and / or claimed herein.

[0148] Electronic devices include a computing unit, which can perform various appropriate actions and processes based on a computer program stored in ROM or loaded into RAM from a storage unit. RAM can also store various programs and data required for the operation of the electronic device. The computing unit, ROM, and RAM are interconnected via a bus. I / O interfaces are also connected to the bus.

[0149] Multiple components in an electronic device are connected to an I / O interface, including: input units such as keyboards and mice; output units such as various types of displays and speakers; storage units such as disks and optical discs; and communication units such as network interface cards (NICs), modems, and wireless transceivers. The communication unit allows the electronic device to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0150] The computing unit can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of computing units include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit performs the various processes described above. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device via ROM and / or a communication unit.

[0151] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0152] The program code used to implement the schemes of this disclosure may be written in any combination of one or more programming languages. This program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing apparatus, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0153] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0154] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including voice input, speech input, or tactile input).

[0155] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

[0156] Computer systems can include clients and servers. Clients and servers are generally located far apart and typically interact via communication networks. Client-server relationships are created by computer programs running on the respective computers and having a client-server relationship with each other. Servers can be cloud servers, servers in distributed systems, or servers incorporating blockchain technology.

[0157] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this disclosure can be achieved, and this is not limited herein.

[0158] The specific embodiments described above do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure should be included within the scope of protection of this disclosure.

Claims

1. A pre-pregnancy carrier screening genetic counseling system based on a large language model, characterized in that, The application comprises: a variation acquisition module for screening user's to-be-sequenced gene data to obtain pathogenic abnormal site information, converting the pathogenic abnormal site information, disease clinical phenotype, family history and personal history into a high-dimensional vector; a variation retrieval module connected with the variation acquisition module, for dynamic retrieval of the high-dimensional vector based on a semantic vector index database and recalling Top-K medical data in the semantic vector index database; wherein K is a natural number not equal to 0; an inference analysis module connected with the variation retrieval module, for evaluating the risk of genetic variation of reproduction according to the Top-K related medical data and the high-dimensional vector to obtain an evaluation result; a multi-round dialogue module connected with the inference analysis module, for the user to obtain the evaluation result and its explanation through multi-round dialogue; the pathogenic abnormal site information is obtained by the following steps: acquiring a high-frequency variation gene to construct a gene library, and screening the detection target in the gene library; capturing the target gene region in the user's to-be-sequenced gene data through a probe set; automatically annotating the target gene region combined with genetic variation classification guidelines to obtain pathogenic abnormal sites; confirming the gene name based on the pathogenic abnormal site and marking to obtain the pathogenic abnormal site information; wherein the probe set comprises SEQ ID NO: 1-277; the construction process of the semantic vector index database comprises: collecting multi-source medical data; the multi-source medical data comprises medical guidelines, clinical path standard texts, genetic disease knowledge base; automatically segmenting the medical data to obtain text blocks; converting all text blocks into vectors through an embedding coding model to obtain a semantic vector index database; wherein the embedding coding model is a BGE model.

2. The preconception carrier screening genetic counseling system based on a large language model according to claim 1, characterized in that, The variation retrieval module is used to perform the following steps: calculating the semantic similarity between the high-dimensional vector and the medical data in the semantic vector index database, and recalling the K medical data with the highest semantic similarity to obtain Top-K medical data. 3.The pre-pregnancy carrier screening genetic counseling system based on a large language model of claim 1, wherein, The inference analysis module comprises a large language model, and a preset genetic counseling inference chain is arranged in the large language model, which is embedded in the model in the form of a rule chain or a Prompt template.

4. The preconception carrier screening genetic counseling system based on a large language model according to claim 3, characterized in that, The genetic counseling inference chain comprises: judging the disease inheritance mode according to the Top-K related medical data and the pathogenic abnormal site information, and evaluating the reproduction risk in combination with the disease clinical phenotype, family history and personal history.

5. The pre-pregnancy carrier screening genetic counseling system based on a large language model according to claim 1, wherein, The evaluation result comprises reproduction risk evaluation and intervention suggestion; the reproduction risk evaluation comprises theoretical reproduction risk calculated in combination with the genetic mode, and personal reproduction risk modified based on the family history and personal history; the intervention suggestion comprises emergency intervention suggestion, routine monitoring suggestion, long-term health management suggestion and assisted reproduction suggestion according to priority.

6. The pre-pregnancy carrier screening genetic counseling system based on a large language model according to claim 1, wherein, determine whether both husband and wife users include pathogenic abnormal site information on the same autosomal recessive genetic gene, or whether the female carries pathogenic abnormal site information on the sex-linked genetic gene; if yes, enter the clinical solution recommendation module.

7. The preconception carrier screening genetic counseling system based on a large language model according to claim 6, characterized in that, The clinical solution recommendation module is used to perform the following steps: calculating a clinical resolution evaluation score of the user based on multiple dimensions; determining whether the clinical resolution evaluation score is greater than or equal to a first preset value, if yes, entering a PGT-M assisted pregnancy process, otherwise, entering the next step; determining whether the clinical resolution evaluation score is less than a second preset value, the second preset value being less than the first preset value, if yes, entering a natural pregnancy process; otherwise, entering a conventional pre-pregnancy counseling process; genetic detection is performed in the PGT-M assisted pregnancy process, the natural pregnancy process and the conventional pre-pregnancy counseling process, if the detection result is normal, pregnancy is continued, otherwise, pregnancy is terminated or genetic counseling guidance is obtained.

8. The preconception carrier screening genetic counseling system based on a large language model of claim 1, wherein, Further comprising: a structure output module configured to generate a structured report based on the evaluation result and output the structured report.

Images