A hospital library retrieval system and method

Abstract

This invention relates to a hospital library retrieval method, wherein the hospital library communicates with at least one superior digital library. The method includes the following steps: obtaining a first retrieval result from the superior digital library based on retrieval content information; obtaining a second retrieval result from an identity book list database based on the retrieval content information and retrieval identity information, wherein the identity book list database is constructed based on medical staff-related book data and medical record data; the medical staff-related book data includes academic book data and academic book data; and merging the first retrieval result and the second retrieval result to obtain the user retrieval result. The hospital library retrieval method of this invention considers the hospital's patient medical records and the knowledge structure of the hospital's medical staff in the retrieval results, enabling hospitals, medical staff, and patients to benefit from improved retrieval performance and coverage of their digital libraries.

Description

Technical Field

[0001] This invention belongs to the field of information retrieval technology, and specifically relates to a hospital library retrieval system and method. Background Technology

[0002] A digital library is a type of distributed information system (Distributed Software Systems). To meet the clinical and research needs of medical staff, many hospitals have built hospital digital libraries. However, current hospital digital library search results only consider search efficiency, accuracy, and comprehensiveness, without taking into account the patient's medical history and the knowledge structure of the hospital's medical staff. This means that while hospitals may improve the search performance and scope of their digital libraries, it may not necessarily benefit the hospital, medical staff, and patients. Summary of the Invention

[0003] To address at least one technical problem proposed in this invention, a hospital library retrieval method is proposed, wherein the hospital library communicates with at least one superior digital library, and the method includes the following steps:

[0004] Based on the search content information, obtain the first ordered set of search results from the superior digital library;

[0005] The second search result set is obtained from the identity book list database based on the search content information and the search identity information; the identity book list database is constructed based on medical staff-related book data and case data, and the medical staff-related book data includes academic book data and academic book data.

[0006] Calculate the intersection of the first ordered set of search results and the second ordered set of search results to generate the first ordered set of user search results.

[0007] To address at least one technical problem proposed in this invention, this invention also proposes a hospital library retrieval system, the system comprising at least one processor; and a memory storing instructions that, when executed by the at least one processor, implement the steps of the aforementioned method.

[0008] To address at least one technical problem proposed in this invention, this invention also provides a computer-readable storage medium storing a computer program / instructions thereon, which, when executed by a processor, implements the steps of the aforementioned method.

[0009] To address at least one technical problem proposed in this invention, this invention also provides a computer program product comprising a computer program / instructions that, when executed by a processor, implement the steps of the aforementioned method.

[0010] The beneficial effects of the present invention are as follows: the retrieval results of the hospital library retrieval method of the present invention take into account the medical records of patients in the hospital and the knowledge structure of the hospital's medical staff. While improving the retrieval performance and coverage of the hospital's digital library, it can benefit the hospital, medical staff and patients. Attached Figure Description

[0011] Figure 1 A topology diagram of the superior digital library and the hospital library;

[0012] Figures 2-4 A flowchart illustrating the hospital library's search methods;

[0013] Figure 5 This is a schematic diagram of the ordered set of the first search results;

[0014] Figure 6 This is a schematic diagram of the ordered set of the second search results;

[0015] Figure 7 This is a schematic diagram of the ordered set of search results for the first user.

[0016] Figure 8 This is a schematic diagram of the ordered set of search results for the second user.

[0017] Figure 9 The sorting of matching results indicates the intended meaning;

[0018] Figure 10 To correct the set diagram;

[0019] Figure 11 A schematic diagram of the ordered set of search results for the second user (after correction);

[0020] Figure 12 The sorting of matching results indicates the intended meaning;

[0021] Figure 13 This is a diagram illustrating search results for books with negative case studies removed. Detailed Implementation

[0022] In some embodiments, a hospital library retrieval method is involved, such as... Figure 1 As shown, the hospital library 1 communicates with at least one superior digital library 2, such as... Figure 2 As shown, the method includes the following steps:

[0023] S1: Obtain the first ordered set of search results from the superior digital library based on the search content information;

[0024] S2: Obtain a second ordered set of search results from the identity book list database based on the search content information and search identity information; the identity book list database is constructed based on medical staff-related book data and case data, and the medical staff-related book data includes academic book data and academic book data;

[0025] S3: Calculate the intersection of the first ordered set of search results and the second ordered set of search results to generate the first ordered set of user search results.

[0026] The following is combined Figures 5-7 These embodiments are further illustrated. For example... Figure 5 As shown, the first search result ordered set 10 is recorded as {A, B, C, D, E, F, G, H, I}, including 9 ordered search results 11, each of which includes a title 12 and an abstract 13. For example... Figure 6 As shown, the ordered set 20 of the second search results is recorded as {A, B, o, p, E, F, q, r, i}, including 9 ordered search results 21. Each search result 21 includes a title 22 and an abstract 23. Figure 7 As shown, the ordered set 30 of the first user's search results is the intersection of the two sets mentioned above, and is recorded as {A, B, E, F}. It includes four ordered search results 31, and each search result 31 includes a title 32 and an abstract 33.

[0027] "Retrieval content information" refers to the information that users input into the retrieval system. The modalities of retrieval content information include written (such as keywords) or spoken natural language signals (such as speech), images, and video signals. It can be a single-modal retrieval or a cross-modal retrieval, the latter of which is generally used in multimedia digital libraries.

[0028] "Superior-level digital library" refers to a digital library whose retrieval system can perform combined searches on a large number of databases and display unified search results. Taking university digital libraries as an example, they typically purchase a large number of shared databases (such as MedSci, PubMed, CNKI, Chinese Biomedical Literature Database (CBM), Chinese Science Citation Database (CSCD), Wanfang Data, DuXiu Academic Search, VIP Information, Chaoxing Digital Library, TANet, PubChem, JSTOR, IEEE Xplore, EBSCOhost, ScienceDirect, SprungerLink, Web of Science, NSSD, etc.) or build several databases within their own institution (such as the library's collection). To improve retrieval efficiency, their retrieval system can perform combined searches on a large number of databases and display unified search results. In this invention, the retrieval system of the superior-level digital library is a known technology, such as the retrieval system in Chinese Patent CN118170816 A, which uses user search history data to perform database weight filtering to obtain a combination of target databases. Compared to the university's digital library, the affiliated hospital's digital library is a subordinate digital library. The affiliated hospital's digital library can obtain search results by accessing the university's digital library interface, rather than accessing the interface of the shared database, thus avoiding the duplication of digital library construction.

[0029] "Search results" refers to the search results that users obtain after a search tool or system has performed a search process, which meet the search requirements. These results include titles, abstracts, image thumbnails, etc.

[0030] "Ordered set of search results" refers to a data structure such as a table or array composed of ordered search results (data / elements).

[0031] "Search identity information" includes registration information that can distinguish a user's knowledge structure, such as medical / nursing role information, medical / nursing educational background information, and work experience information (e.g., clinical research papers published or supervised during work, academic conferences attended). The medical / nursing role information includes doctor identity information and nurse identity information. The hospital library search system can collect search identity information through user registration forms.

[0032] "Educational Book Data" refers to a dataset or catalog of books representing the knowledge structure of medical personnel during their studies, obtained through computer searches or manual compilation based on their educational background. This includes, but is not limited to, textbook data, dissertations, and their citation data. For example, the textbook data for Doctor B, who graduated from University A with a master's degree in clinical medicine, might include the following books: *Physiology* (X Publishing House, Y edition), *Internal Medicine* (X Publishing House, Y edition), *Surgery* (X Publishing House, Y edition), *Preventive Medicine* (X Publishing House, Y edition), *Neurology* (X Publishing House, Y edition), *Biochemistry and Molecular Biology* (X Publishing House, Y edition), *Pharmacology* (X Publishing House, Y edition), *Pathology* (X Publishing House, Y edition), *Patient Physiology* (X Publishing House, Y edition), *Diagnostics* (X Publishing House, Y edition), *Medical Microbiology* (X Publishing House, Y edition), *Traditional Chinese Medicine* (X Publishing House, Y edition), *Obstetrics and Gynecology* (X Publishing House, Y edition), *Histology and Embryology* (X Publishing House, Y edition), *Pediatrics* (X Publishing House, Y edition), *Medical Advanced Mathematics* (X Publishing House, Y edition), and *Doctor-Patient Communication*. *Organic Chemistry* (X Publishing House, Y Edition), *Medical Biology* (X Publishing House, Y Edition), *Basic Chemistry* (X Publishing House, Y Edition), *Medical Genetics* (X Publishing House, Y Edition), *Medical Psychology* (X Publishing House, Y Edition), *Anesthesiology* (X Publishing House, Y Edition), *Internal Medicine* (X Publishing House, Y Edition), *Surgery* (X Publishing House, Y Edition), *Preventive Medicine* (X Publishing House, Y Edition), *Neurology* (X Publishing House, Y Edition), *Forensic Medicine* (X Publishing House, Y Edition), *Medical Statistics* (X Publishing House, Y Edition), *Ophthalmology* (X Publishing House, Y Edition), *Clinical Epidemiology and Evidence-Based Medicine* (X Publishing House, Y Edition), *Human Parasitology* (X Publishing House, Y Edition), *Dermatology* (X Publishing House, Y Edition), *Systemic Anatomy* (X Publishing House, Y Edition), *Patophysiology* (X Publishing House, Y Edition).

[0033] "Academic book data" refers to a dataset of books that represents the knowledge structure of medical personnel during their work, obtained through computer searches or manually compiled based on their academic information. This academic book data includes data from papers published after their employment and their citations, academic conference data, etc.

[0034] The "academic book data" and "academic book data" of this invention are obtained using known computer technologies, such as data crawling (e.g., DFS, BFS crawlers), data filtering, text description supplementation, and data annotation, to establish a dataset that can be used for digital libraries.

[0035] "Medical staff-related book data" refers to a dataset that combines "academic book data" and "educational book data" after data preprocessing. This dataset reflects the relationship between "retrieval identity information" and "book data," representing the sum of medical staff's knowledge structure. The data preprocessing method includes one or more steps such as data cleaning, data standardization, data normalization, category coding, and feature selection.

[0036] "Case data" includes, but is not limited to, digitized biochemical test reports, discharge summaries, imaging test reports, medical records, emergency room progress notes, admission records, case investigations, and case study reports. The "case data" of this invention can be directly obtained from a digital hospital system or acquired using machine learning techniques. Digital hospital systems include, but are not limited to, Hospital Information Systems (HIS), Telemedicine Systems, Online Business Processing Systems (OLTP), Clinical Information Systems (CIS), Online Analytical Processing Systems (OLAP), and Intranet / Internet systems. Healthcare information technology refers to information and communication technologies specifically used for handling or processing medical or health data, such as ICT technologies for medical simulation or medical data mining based on patient-specific data (e.g., electronic medical records). There are no prior art reports on combining hospital digital libraries and healthcare information technology; however, to address the technical problems of this invention, several embodiments of this application combine digital libraries and healthcare information technology.

[0037] "Intersection" refers to a set of elements (search results) that belong to both the ordered set of the first search results and the ordered set of the second search results, and the elements of the intersection are also ordered.

[0038] Based on the above embodiments, a digital library platform is provided that is tailored to the current resources of the hospital and takes into account clinical needs. Because the search results take into account the hospital's patient medical records and the knowledge structure of its medical staff, improving the search performance and coverage of the hospital's digital library benefits the hospital, medical staff, and patients. The hospital does not need to blindly pursue search efficiency, accuracy, and comprehensiveness, thus avoiding wasting medical funds (if a database is purchased that does not match the knowledge structure of the hospital's medical staff, it is equivalent to the hospital's investment in the library not being reflected in patient services). The search results obtained by medical staff are more targeted and accessible (reducing information redundancy; for example, nurses' search results will not include certain foreign language books or books with high professional depth; doctors' search results will not include books with poorly recorded treatment plans). Patients can obtain more advanced treatment services.

[0039] In other embodiments of the present invention, a hospital library retrieval method is also involved, such as... Figure 3As shown, the method further includes the following steps:

[0040] S4: Calculate the relative complement of the ordered set of the second search results in the ordered set of the first search results, generate the ordered set of the second user search results, and output the ordered set of the first user search results before the ordered set of the second user search results.

[0041] "The relative complement of the ordered set of the second search results in the ordered set of the first search results" means that the elements of the relative complement belong to the ordered set of the first search results but not to the ordered set of the second search results.

[0042] "The first user's ordered search results are prioritized over the second user's ordered search results" means that in the matching results ranking table presented to the user, the search results (data / elements) in the first user's ordered search results are ranked first, while the search results (data / elements) in the second user's ordered search results are ranked last.

[0043] A "matching result ranking table" refers to a visual form presented to users through a human-computer interaction interface (GUI), including unimodal and mixed-modal matching result ranking tables. Traditional digital libraries typically use unimodal ranking tables, such as ranking tables of excerpts from digital journals, dissertations, or books. Multimedia digital libraries typically use mixed-modal ranking tables, such as ranking journals, papers, or books alongside videos. The aforementioned methods of presenting data in matching result ranking tables are known technologies.

[0044] The following is combined Figures 5-9 These embodiments are further illustrated. For example... Figure 5 As shown, the first search result ordered set 10 is recorded as {A, B, C, D, E, F, G, H, I}, including 9 ordered search results 11, each of which includes a title 12 and an abstract 13. For example... Figure 6 As shown, the ordered set 20 of the second search results is recorded as {A, B, o, p, E, F, q, r, i}, including 9 ordered search results 21. Each search result 21 includes a title 22 and an abstract 23. Figure 7 As shown, the ordered set 30 of the first user's search results is the intersection of the two sets mentioned above, recorded as {A, B, E, F}, including four ordered search results 31. Each search result 31 includes a title 32 and a summary 33. Figure 8As shown, the ordered set 40 of the second user search results is the relative complement of the ordered set of the second search results in the ordered set of the first search results, recorded as {C, D, G, H, I}, including 5 ordered search results 41, each of which includes a title 42 and an abstract 43. For example... Figure 9 As shown, in the matching result ranking table 50, the search results (data / elements) in the ordered set {A, B, E, F} of the first user's search results are ranked first, while the search results (data / elements) in the ordered set {C, D, G, H, I} of the second user's search results are ranked last.

[0045] Based on the above embodiments, the ordered set of first user search results preserves the technical effect of making the search results obtained by medical staff more targeted and accessible, while the ordered set of second user search results ensures that users have the need for in-depth browsing of the superior digital library.

[0046] In other embodiments of the present invention, a hospital library retrieval method is also involved, such as... Figure 4 As shown, the method further includes the following steps:

[0047] S5: Calculate the relative complement of the ordered set of the first search results in the ordered set of the second search results, and generate a corrected set;

[0048] S6: Based on the element relationship between the modified set and the ordered set of the second user's search results, adjust the order of the search results in the ordered set of the second user's search results.

[0049] "Calculate the relative complement of the first search result ordered set in the second search result ordered set" means that the elements of the relative complement belong to the second search result ordered set but not to the first search result ordered set.

[0050] "The relevance between the modified set and the ordered set of the second user's search results" refers to the relevance of the elements (search results) in the two sets, specifically the relevance of the titles and abstracts. The algorithm for determining the relevance of the titles and abstracts of the two search results is a known technique. For example, if the ordered set of the second user's search results includes "Epidemiology (9th Edition) published by People's Medical Publishing House," while the modified set includes "Epidemiology (8th Edition) published by People's Medical Publishing House," the two are highly similar only because they differ in edition. Conversely, if the ordered set of the second user's search results includes "Cell Biology (4th Edition) published by Higher Education Press," while the modified set includes nursing books such as "Introduction to Nursing (5th Edition)" and "Surgical Nursing (7th Edition)," the two have low similarity.

[0051] The following is combined Figures 5-8 , Figures 10-12 These embodiments are further illustrated. For example... Figure 5 As shown, the first search result ordered set 10 is recorded as {A, B, C, D, E, F, G, H, I}, including 9 ordered search results 11, each of which includes a title 12 and an abstract 13. For example... Figure 6 As shown, the ordered set 20 of the second search results is recorded as {A, B, o, p, E, F, q, r, i}, including 9 ordered search results 21. Each search result 21 includes a title 22 and an abstract 23. Figure 7 As shown, the ordered set 30 of the first user's search results is the intersection of the two sets mentioned above, recorded as {A, B, E, F}, including four ordered search results 31. Each search result 31 includes a title 32 and a summary 33. Figure 8 As shown, the ordered set 40 of the second user search results is the relative complement of the ordered set of the second search results in the ordered set of the first search results, recorded as {C, D, G, H, I}, including 5 ordered search results 41, each of which includes a title 42 and an abstract 43. For example... Figure 10 As shown, the correction set 60 is the relative complement of the ordered set of the first search results to the ordered set of the second search results, recorded as {o, p, q, r, i}, including 5 ordered search results 61, each of which includes a title 62 and an abstract 63. In this embodiment, the total relevance score between each search result of the ordered set of the second user search results {C, D, G, H, I} and each search result of the correction set {o, p, q, r, i} is calculated. First, the total similarity score of search result C is calculated with search results o, p, q, r, and i respectively, recorded as {Co, Cp, Cq, Cr, Ci} -> {34 points, 65 points, 28 points, 96 points, 18 points} -> C total score: 241 points. Then, the total similarity scores of D, G, H, and I are obtained in sequence: D total score: 414 points; G total score: 119 points; H total score: 208 points; I total score: 239 points. like Figure 11 As shown, the ordered set of second user search results (corrected) 40' is formed by rearranging the search results of the ordered set of second user search results {C, D, G, H, I}, and is recorded as {D, C, I, H, G}. Each search result 41' includes a title 42' and an abstract 43'. Figure 12 As shown, in the matching result ranking table 70, the search results (data / elements) in the ordered set {A, B, E, F} of the first user's search results are ranked first, while the search results (data / elements) in the ordered set {D, C, I, H, G} of the second user's search results (after correction) are ranked last.

[0052] Based on the above embodiments, the ordered set of search results from the first user retains the technical effect of making the search results obtained by medical staff more targeted and accessible, while the ordered set of search results from the second user ensures that users have the need for in-depth search of the superior digital library.

[0053] In other embodiments of the present invention, a hospital library retrieval method is also involved, wherein,

[0054] The first ordered set of search results is split into multiple ordered subsets of the first search results;

[0055] The ordered set of the second search results is split into multiple ordered subsets of the second search results;

[0056] Calculate the intersection of the corresponding ordered subset of the first search result and the ordered subset of the second search result to generate the ordered subset of the first user search result.

[0057] "First Search Result Ordered Set Splitting" is based on set partitioning technology. It calculates the number of search results (elements) (n) displayed per page in the matching result ranking table of the user's search device, and the number of search results (elements) (m) in the first ordered set of search results for this search. The first ordered set of search results is then split into m / n, rounded up. For example, if a user's search device is a desktop computer, and its matching result ranking table displays 15 search results (elements) per page, and the first ordered set of search results for a certain search includes 144 search results (elements), then this first ordered set of search results will be split into 10 ordered subsets of the first search results. As another example, if a user's search device is a mobile phone, and its matching result ranking table displays 9 search results (elements) per page, and the first ordered set of search results for a certain search includes 144 search results (elements), then this first ordered set of search results will be split into 16 ordered subsets of the first search results.

[0058] "Second search result ordered set splitting" is also based on set partitioning technology. According to the number n (elements) of search results (elements) presented in the matching result sorting table of each page of the user's search device, and the number m (elements) of the second search result ordered set in this search, the second search result ordered set is split into m / n rounded up (elements).

[0059] "Calculate the intersection of the corresponding ordered subsets of the first search results and the ordered subsets of the second search results" means that the intersection of one ordered subset of the first search results and one ordered subset of the second search results is taken (1 to 1), or the intersection of one ordered subset of the first search results and multiple ordered subsets of the second search results is taken (1 to many).

[0060] In the above embodiments, when the data volume of the first ordered set of search results and the second ordered set of search results is large, dividing the set into subsets and then performing the intersection operation greatly reduces the processing time. Specifically, when taking the intersection of one ordered subset of the first search results with multiple ordered subsets of the second search results, priority is given to presenting the elements that meet the conditions as early as possible from the ordered subsets of the second search results. For example, if there are 7 ordered subsets of the first search results and 16 ordered subsets of the second search results, according to the 1-to-2 correspondence, 14 elements of the ordered subsets of the second search results can be included in the intersection, while according to the 1-to-1 correspondence, only 7 elements of the ordered subsets of the second search results are included in the intersection. The former prioritizes the ordered subsets of the second search results.

[0061] In other embodiments of the present invention, a hospital library retrieval method is also involved, wherein the method further includes the following steps:

[0062] Calculate the relative complement of the ordered subset of the second search result to the corresponding ordered subset of the first search result, and generate the ordered subset of the second user search result. The ordered subset of the first user search result is output with priority over the ordered subset of the second user search result.

[0063] In the above embodiment, the difference from the previous embodiment's "relative complement of the second search result ordered set in the first search result ordered set" is that the elements of the relative complement in the previous embodiment considered all elements belonging to the first search result ordered set but not to the second search result ordered set. However, the elements of the relative complement in this embodiment only consider elements in the matching result sorting table of each page that belong to the first search result ordered set but not to the second search result ordered set, which can reduce the computational load when the data volume is large.

[0064] In other embodiments of the present invention, a hospital library retrieval method is also involved, wherein the method further includes the following steps:

[0065] Calculate the relative complement of the ordered subset of the first search result in the ordered subset of the second search result, and generate a corrected subset;

[0066] Based on the relevance of the modified subset to several ordered subsets of the second user search results, including the corresponding ordered subsets of the second user search results, the order of the search results in the ordered subsets of the second user search results is adjusted.

[0067] In the above embodiment, the difference from the previous embodiment's "based on the element relationship between the modified set and the ordered set of the second user's search results" is that the elements of the relative complement in the previous embodiment considered all elements belonging to the ordered set of the second search results but not to the ordered set of the first search results. However, the elements of the relative complement in this embodiment only consider the elements of the matching result sorting table on each page, which reduces the computational load when the data volume is large.

[0068] In other embodiments of the present invention, a hospital library retrieval method is also involved, wherein a case-irrelevant book dataset is constructed based on the case data, and search results in the ordered set of second user search results that are identical to the case-irrelevant book dataset are deleted.

[0069] "Constructing a case-irrelevant book dataset based on case data" refers to constructing a dataset based on negative case data (such as medical malpractice cases) and the book data that matches them.

[0070] In the above embodiments, the probability of medical staff receiving incorrect information is reduced. See also Figure 13 As shown, if the associated book for a negative case is "Guidelines for the Diagnosis, Treatment and Rehabilitation of Cervical Spondylosis" (2018), then in the ordered set of the second user's search results, the search results containing "Guidelines for the Diagnosis, Treatment and Rehabilitation of Cervical Spondylosis" (2018) will be deleted, but "Guidelines for the Diagnosis, Treatment and Rehabilitation of Cervical Spondylosis" 2016, 2017, 2019, and 2022 will still be retained.

[0071] In other embodiments of the present invention, a hospital library retrieval method is also involved, wherein the correlation between the modified set and the ordered set of the second user's retrieval results is recorded each time a retrieval is performed, and the high-frequency and low-relevance retrieval results in the modified set are used to generate an ordered set of candidate books for the hospital library.

[0072] In the above embodiment, "high-frequency and low-relevance search results in the correction set" refers to search results (elements) that appear multiple times in the ordered set of second user search results, but each time have a low relevance to the correction set. The books corresponding to these search results are books of interest to the hospital's medical staff, but are not included in the database of the superior digital library. Therefore, the ordered set of candidate books for the hospital library can serve as a reference for the hospital to expand the scope of its digital library in the future.

[0073] In other embodiments of the present invention, a hospital library retrieval method is also involved, wherein the method for constructing the identity-based book list database based on the medical staff-associated book data and the medical record data includes the following steps:

[0074] 1. Construct a medical staff-related book dataset based on the aforementioned medical staff-related book data;

[0075] 2. Construct a case-related book dataset based on the aforementioned case data;

[0076] 3. The medical staff-associated book dataset and the case-associated book dataset are merged to obtain the identity book list library.

[0077] These embodiments are described below with reference to Table 1.

[0078] Table 1. Element attributes of data and datasets

[0079]

[0080]

[0081] Furthermore, "constructing a medical staff-related book dataset based on medical staff-related book data" includes the following steps:

[0082] 1.1 Obtain the hospital's educational background and academic data;

[0083] The educational background data includes the educational information (school, major, year of enrollment, etc.) of all current and historical medical staff in the hospital, and multiple educational background records are entered for each medical staff member; for example, XXX, a neurosurgeon, studied clinical medicine (undergraduate) at University A in 1996 and neurosurgery (master's) at University B in 2002, which is recorded as two educational background records: {XXX-1, University A, Clinical Medicine, 1996} and {XXX-2, University B, Neurosurgery, 2002}.

[0084] The academic data includes the academic information (published papers, books, conference papers, etc.) of all current and historical medical staff of the hospital, and multiple academic records are entered for the same academic resume of different medical staff; for example, if Dr. XXX and Dr. YYY jointly published paper S in 2010, it is recorded as two academic records: {XXX, paper S} and {YYY, paper S}.

[0085] 1.2 Based on the deduplicated educational information of medical staff, the data of academic books representing the knowledge structure of medical staff during their studies were obtained by computer search or manually organized; based on the deduplicated academic information of medical staff, the data of academic books representing the knowledge structure of medical staff's academic research were obtained by computer search or manually organized.

[0086] For example, continuing from the previous example, if neurosurgeon XXX studied clinical medicine (undergraduate) at University A in 1996, and cardiac surgeon YYY has the same first degree as neurosurgeon XXX, then the search system only needs to search for books related to "1996 University A Clinical Medicine (undergraduate)" (removing duplicate academic information).

[0087] For example, continuing the previous example, since Dr. XXX and Dr. YYY jointly published paper S in 2010, the search system only needs to use relevant citations of "paper S" (deduplicated academic information).

[0088] 1.3 After cleaning the data of academic books, the data was standardized and an educational background influence factor was assigned to obtain the academic-related book dataset. After cleaning the data of academic books, the data was standardized and an academic resume influence factor was assigned to obtain the academic-related book dataset.

[0089] The educational background influence factor α is determined based on the number of times the book is repeated. For example, continuing the previous example, the relevant books for "Clinical Medicine (Undergraduate) at University A in 1996" include book M. Since both neurosurgeon XXX and cardiac surgeon YYY have studied book M, the influence value of the educational background influence factor α for book M is increased. As another example, continuing the previous example, a physical diagnostician studied Medical Imaging (Undergraduate) at University C in 1989, and his relevant books also include book M. Therefore, the influence value of the educational background influence factor α for book M is increased.

[0090] The academic resume impact factor β is also determined based on the number of times the book is repeated. For example, continuing the previous example, the search system needs to search for relevant cited books "Paper T" using "Paper S" (deduplicated academic information). Since both neurosurgeon XXX and cardiac surgeon YYY have cited Paper T, the impact value of Paper T's academic resume impact factor β is increased. As another example, continuing the previous example, if a physical diagnostician's "Paper O" cites "Paper T", the impact value of "Paper T's" academic resume impact factor β is increased.

[0091] 1.4 The educational background-related book dataset and the academic background-related book dataset are merged after deduplication to obtain the medical staff-related book dataset. In a specific embodiment, as shown in Table 1, the merged educational background-related book dataset can simultaneously represent the relationship between the educational background, academic background, and books of the hospital's medical staff.

[0092] Furthermore, "constructing a case-related book dataset based on case data" includes the following steps:

[0093] 2.1 Obtain case data from the digital hospital system, and extract advanced semantic information from the cases to obtain empirical data;

[0094] The experience data includes all experience information (medical care plans, outcomes, etc.) of the hospital's current and historical medical staff involved in diagnosis and treatment, with multiple experience records entered for each patient's multiple cases. The digital hospital system includes, but is not limited to, Hospital Information System (HIS), Telemedicine System, Online Service Transaction Processing System (OLTP), Clinical Information System (CIS), Online Analytical Processing System (OLAP), and Internet system (Intranet / Internet). Well-known feature extraction techniques are used when extracting high-level semantic information, including but not limited to clustering algorithms (such as K-MEANS), Natural Language Processing (NPL), and Machine Learning (ML).

[0095] For example, patient A with cervical spondylosis visited the orthopedics department of this hospital in 2001 and was treated by doctor B. Patient A's medical data includes medical records, imaging reports, biochemical test reports, and case study reports. The extracted high-level semantic information for patient A is: {b, doctor, treatment plan (vertebral artery type cervical spondylosis; dizziness accompanied by headache, with blood stasis, suitable for removing blood stasis and unblocking collaterals, resolving dampness and calming the liver; Xuefu Zhuyu Decoction treatment; treatment results (basically normal; assessment of spinal cord function status in cervical spondylosis patients (40-point scale); I. Upper limb function basically normal; II. Lower limb function basically normal; III. Spinal sphincter function basically normal; VI. Sensation in all four limbs basically normal; V. Girdle sensation basically normal)}.

[0096] 2.2 Experience book data representing the knowledge structure of medical staff's clinical experience, obtained by computer search based on the deduplication experience information of medical staff or by matching with the book dataset associated with medical staff;

[0097] For example, continuing the previous example, the search system used "vertebral artery type cervical spondylosis; dizziness accompanied by headache, with blood stasis, it is advisable to remove blood stasis and unblock the meridians, resolve dampness and calm the liver; Xuefu Zhuyu Decoction for treatment" and "treatment results (basically normal; assessment of spinal cord function status of cervical spondylosis patients (40-point method); I. upper limb function is basically normal; II. lower limb function is basically normal; III. sphincter function is basically normal; VI. limb sensation is basically normal; V. ligament sensation is basically normal" as search text to crawl related book information;

[0098] For example, continuing the previous example, the matching system uses the following text as a reference text: “vertebral artery type cervical spondylosis; dizziness accompanied by headache, with blood stasis, it is advisable to remove blood stasis and unblock the meridians, resolve dampness and calm the liver; Xuefu Zhuyu Decoction for treatment” and “treatment results (basically normal; assessment of spinal cord function status of cervical spondylosis patients (40-point scale); I. upper limb function is basically normal; II. lower limb function is basically normal; III. sphincter function is basically normal; VI. limb sensation is basically normal; V. band sensation is basically normal” to perform semantic matching with the book summary text information of the book dataset associated with medical staff.

[0099] 2.3 After data cleaning, the experience-based book data was standardized and assigned a work experience influence factor to obtain a case-related book dataset;

[0100] The work experience impact factor γ is determined based on the number of times the book is repeated. For example, continuing the previous example, the relevant books for the advanced semantic information of Dr. B's "basically normal" patient a include N books, and the relevant books for the advanced semantic information of Dr. B's "basically normal" patient c also include N books. Since both patients a and c benefited from Dr. B studying N books, the impact value of the work experience impact factor γ for those N books is increased. As another example, continuing the previous example, the relevant books for the advanced semantic information of Dr. B's "basically normal" patient a include N books, and the relevant books in the medical staff association book dataset for Dr. D's "basically normal" patients also include N books. Since both patients a and c benefited from doctors (Dr. B and Dr. D) studying N books at the same hospital, the impact value of the work experience impact factor γ for those N books is increased. Additionally, the experience information of "abnormal" doctors in treating patients will be included in the case-irrelevant book dataset.

[0101] Furthermore, "merging the medical staff-associated book dataset and the case-associated book dataset to obtain the identity book list library" includes the following steps:

[0102] 3.1 Obtain the data sets of books associated with medical staff and the data sets of books associated with medical cases;

[0103] 3.2 The medical staff-associated book dataset and the case-associated book dataset are merged to obtain the identity book list library.

[0104] In one specific embodiment, the greater the influence values ​​of the education background influence factor α, academic resume influence factor β, and work experience influence factor γ on a book, the higher the book's ranking in the identity book list library. However, this invention does not limit the way in which the education background influence factor α, academic resume influence factor β, and work experience influence factor γ affect the ranking; any technique that adjusts the ranking by introducing weights, machine learning, or other methods is within the scope of this invention.

[0105] In the above embodiments, the patients in a hospital are fixed, and the information a hospital can obtain from a digital library is also fixed. Some of this information is incorrect, and some is useful. The construction of the identity-based book list library fully reflects the relationship between the identities (education, academic background, experience) of the hospital's medical staff and the books. The identity-based book list library of this invention is a knowledge graph of the hospital's medical staff and a case knowledge graph. To optimize the matching results, any known knowledge graph construction technology is within the scope of this invention.

[0106] The embodiments and functional operations of the subject matter described in this specification can be implemented in the following ways: digital electronic circuits, tangibly implemented computer software or firmware, computer hardware, including the structures disclosed in this specification and their structural equivalents, or combinations thereof. Embodiments of the subject matter described in this specification can be implemented as one or more computer programs, i.e., one or more modules of computer program instructions encoded on one or more tangible, non-transitory program carriers, for execution by a data processing device or to control the operation of a data processing device. A computer program (which may also be referred to or described as a program, software, software application, module, software module, script, or code) can be written in any form of programming language, including compiled or interpreted languages, declarative languages, or procedural languages, and can be expanded in any form, including as a standalone program or as a module, component, subroutine, or other unit suitable for use in a computing environment. A computer program may, but does not necessarily, correspond to a file in a file system. A program can be stored in a portion of a file that stores other programs or data, for example, stored in one or more scripts as follows: in a markup language document; in a single file dedicated to a related program; or in multiple collaborating files, for example, a file storing one or more modules, subroutines, or code portions. A computer program can be deployed to execute on one or more computers, which are located in one place or distributed to multiple locations and interconnected through a communication network.

Claims

1. A method for retrieving information in a hospital digital library, wherein the hospital digital library communicates with at least one superior digital library, and the hospital digital library obtains search results by accessing the interface of the superior digital library instead of accessing the interface of the superior digital library's shared database, characterized in that... The method includes the following steps: Based on the search content information, obtain the first ordered set of search results from the superior digital library; Based on the search content information and search identity information, a second ordered set of search results is obtained from the identity book list library; the identity book list library is constructed based on medical staff-related book data and case data, and the method for constructing the identity book list library includes the following steps: constructing a medical staff-related book dataset based on the medical staff-related book data, constructing a case-related book dataset based on the case data, and extracting and fusing features from the medical staff-related book dataset and the case-related book dataset to obtain the identity book list library; The medical staff-related book data is a dataset that combines academic and educational book data after data preprocessing, reflecting the relationship between the search identity information and the book data. The medical staff-related book data includes academic and educational book data. The academic book data refers to a dataset or catalog of books representing the knowledge structure of medical staff during their studies, obtained through computer search or manually compiled based on their educational information. The academic book data refers to a dataset of books representing the knowledge structure of medical staff during their work period, obtained through computer search or manually compiled based on their academic information. The intersection of the first ordered set of search results and the second ordered set of search results is calculated to generate the first ordered set of user search results.

2. The hospital digital library retrieval method as described in claim 1, characterized in that, The method further includes the following steps: Calculate the relative complement of the ordered set of the second search results in the ordered set of the first search results, and generate the ordered set of the second user search results. The ordered set of the first user search results is output before the ordered set of the second user search results.

3. The hospital digital library retrieval method as described in claim 2, characterized in that, The method further includes the following steps: Calculate the relative complement of the ordered set of the first search results in the ordered set of the second search results, and generate a corrected set; Based on the correlation between the modified set and the ordered set of the second user's search results, the order of the search results in the ordered set of the second user's search results is adjusted.

4. The hospital digital library retrieval method as described in any one of claims 1 to 3, characterized in that, The first ordered set of search results is split into multiple ordered subsets of the first search results; The ordered set of the second search results is split into multiple ordered subsets of the second search results; Calculate the intersection of the corresponding ordered subset of the first search result and the ordered subset of the second search result to generate the ordered subset of the first user search result.

5. The hospital digital library retrieval method as described in claim 4, characterized in that, The method further includes the following steps: Calculate the relative complement of the ordered subset of the second search result to the corresponding ordered subset of the first search result, and generate the ordered subset of the second user search result. The ordered subset of the first user search result is output with priority over the ordered subset of the second user search result.

6. The hospital digital library retrieval method as described in claim 5, characterized in that, The method further includes the following steps: Calculate the relative complement of the ordered subset of the first search result in the ordered subset of the second search result, and generate a corrected subset; Based on the relevance of the modified subset to several ordered subsets of the second user search results, including the corresponding ordered subsets of the second user search results, the order of the search results in the ordered subsets of the second user search results is adjusted.

7. The hospital digital library retrieval method as described in claim 2, characterized in that, Construct a case-irrelevant book dataset based on the case data, and delete the search results in the ordered set of the second user search results that are the same as the case-irrelevant book dataset.

8. The hospital digital library retrieval method as described in claim 3, characterized in that, Record the correlation between the modified set and the ordered set of search results of the second user during each search, and generate an ordered set of candidate books for the hospital library from the high-frequency and low-relevance search results in the modified set.

9. A hospital digital library retrieval system, characterized in that, The system includes at least one processor; and a memory storing instructions that, when executed by the at least one processor, perform the steps of the method according to any one of claims 1-8.

Images