CN111968715A - Drug recommendation modeling method based on medical record data and drug interaction risk - Google Patents
Drug recommendation modeling method based on medical record data and drug interaction risk Download PDFInfo
- Publication number
- CN111968715A CN111968715A CN202010623195.0A CN202010623195A CN111968715A CN 111968715 A CN111968715 A CN 111968715A CN 202010623195 A CN202010623195 A CN 202010623195A CN 111968715 A CN111968715 A CN 111968715A
- Authority
- CN
- China
- Prior art keywords
- drug
- medical record
- record data
- recommendation
- medicine
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H20/00—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance
- G16H20/10—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance relating to drugs or medications, e.g. for ensuring correct administration to patients
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H70/00—ICT specially adapted for the handling or processing of medical references
- G16H70/40—ICT specially adapted for the handling or processing of medical references relating to drugs, e.g. their side effects or intended usage
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Physics & Mathematics (AREA)
- Chemical & Material Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Medicinal Chemistry (AREA)
- Theoretical Computer Science (AREA)
- Epidemiology (AREA)
- Medical Informatics (AREA)
- Primary Health Care (AREA)
- Public Health (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Toxicology (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Pharmacology & Pharmacy (AREA)
- Medical Treatment And Welfare Office Work (AREA)
Abstract
The invention discloses a medicine recommendation modeling method based on medical record data and medicine interaction risks, which specifically comprises the following steps: s100: heterogeneous feature extraction operation is carried out on medical record data, outpatient medical record data and TWOSIDES drug interaction data; s200: constructing a drug recommendation sequence generation model according to the extracted multi-source heterogeneous characteristics; s300: and inputting the patient medical record into the medicine recommendation sequence generation model, and calculating and outputting the recommended medication information by the model. According to the medicine recommendation modeling method based on medical history data and medicine interaction risks, medicine recommendation information is given by automatically learning and modeling from massive historical case data and medicine interaction information, accurate and effective decision basis is provided for reasonable combination of clinical medicines, the incidence rate of adverse medicine reactions is reduced, and effective guarantee is provided for medication safety of users.
Description
Technical Field
The invention relates to the field of information processing, in particular to a medicine recommendation modeling method based on medical record data and medicine interaction risks.
Background
According to investigation, the disease and symptoms are not simply in a one-to-one relationship, and the occurrence of one disease may cause multiple symptoms to occur simultaneously, so that doctors must cure patients by matching multiple medicines. Likewise, it is increasingly common for two or more chronic or acute diseases to occur simultaneously in individual patients. Therefore, how to select the most suitable treatment scheme for the disease condition of the patient becomes one of the most important factors in the hospitalization of the patient. The combination of drugs should take into account the medical condition of the patient, the effectiveness of the treatment and the potential adverse effects of the treatment. Physicians typically prescribe them based on their own experience and intuition. However, due to knowledge gaps or unexpected prejudices, these clinical decisions may often not be the best choices. With the development of big data technology, the construction of the electronic medical record system and the wide use of the electronic health record provide an excellent opportunity for improving clinical decision making by using medical care data. In this case, a computer-assisted treatment recommendation system based on electronic medical record data may provide powerful assistance. The present method attempts to provide clinical medication decisions by discovering hidden clinical knowledge from historical medical record data and utilizing it to form effective and safe medication recommendations.
Disclosure of Invention
Aiming at the technical problems in the prior art, the invention provides a medicine recommendation modeling method based on medical history data and medicine interaction risks, which provides medicine recommendation information by automatically learning and modeling from massive historical case data and medicine interaction information, provides an accurate and effective decision basis for reasonable combination of clinical medicines, reduces the incidence rate of adverse drug reactions, and provides effective guarantee for the medication safety of users.
The technical scheme adopted by the invention for solving the technical problems is as follows:
a drug recommendation modeling method based on medical record data and drug interaction risks comprises the following steps:
s100: heterogeneous feature extraction operation is carried out on medical record data, outpatient medical record data and TWOSIDES drug interaction data;
s200: constructing a drug recommendation sequence generation model according to the extracted multi-source heterogeneous characteristics;
s300: and inputting the patient medical record into the medicine recommendation sequence generation model, and calculating and outputting the recommended medication information by the model.
Further, step S100 specifically includes the following steps:
s110: and extracting physiological information and diagnosis information of the patient from medical record data and outpatient medical record data as the characteristics of reasonable medication recommendation, wherein the physiological information is three characteristics of sex, age and weight, and the diagnosis information is diagnosis sequence characteristics.
S120: drug interaction relationships were established from the twosid drug interaction data as another feature of rational drug recommendations.
Further, step S110 specifically includes the following steps:
s111: the gender of the patient in the historical medical record data set is distinguished by F and M, which are designated by 0 and 1, respectively, for ease of computer processing. And code conversion is performed on the medicines, and ATC4 codes are used for representing different medicines.
S112: dictionaries are separately built for all diagnoses and drugs in the data set, and integers are used to represent the diagnoses and drugs in the finally generated data set.
Further, step S120 specifically includes the following steps:
s121: for case history data and the MIMIC _ III dataset, two drugs in the drug set are arbitrarily selected, assuming the encoded value is i, j, and then their ATC4 codes are mapped into the CID classification.
S122: searching the drug interaction database for the two types of drugs by using CID codes as key words to determine whether the two types of drugs have adverse drug interaction risks, and if so, Aij1, otherwise Aij0. And sequentially traversing all the medicine pairs in the database to generate a final adjacency matrix of the medicine interaction graph.
S123: and performing one-dimensional space transformation on the adjacent matrix characteristics of the drug interaction map. A graph neural network is built with two GCN layers, and after feature aggregation, non-linear permutations such as ReLU are applied to the generated output, and the final hidden representation of each node in the graph gets information from the subsequent neighborhood through stacking of multiple layers of GCNs. And finally, accessing a full-connection network to obtain one-dimensional vector output.
Further, step S200 specifically includes the following steps:
s210: symbols are defined such that X represents a diagnosis space and Y represents a medication space. R is a set of prescription records, whereinIs a diagnostic sequence, andis a drug sequence. I XkI and YkEach is XkAnd YkThe length of the sequence of (c).
S220: constructing a two-layer Transformer model, projecting Query, Key and Value to different spaces h times through linear transformation, and then obtaining h self-attribute matrixes through calculation. And finally, splicing the h matrixes, and multiplying the h matrixes by a weight matrix to generate a final attention matrix.
S230: based on a Transformer model, the extracted features are spliced and used as an input sequence, elements with the maximum probability in probability distribution are selected for each position through a multi-step coding and decoding process to serve as prediction results of the current position until an end symbol is generated or the set maximum sequence length is reached, and finally the generated sequence is the medication recommendation scheme of the model for the current patient.
Further, step S210 specifically includes the following steps:
s211: prescription record R { (X)1,Y1),(X2,Y2),…,(Xk,Yk)}
Further, step S220 specifically includes the following steps:
s221: when self-attention between input and output sequences is calculated, Q is equal to K and V, and the specific calculation process is as follows:
MultiHead(Q,K,V)=Concat(head1,…,headh)W0
headi=Attention(QWi Q,KWi K,VWi V
further, step S300 specifically includes the following steps:
s310: and reading the patient medical record as the input of the drug recommendation sequence generation model, performing model calculation, and finally outputting the recommended medication information.
Compared with the prior art, the invention has the following beneficial effects:
1. the medicine recommendation modeling method based on medical history data and medicine interaction risks can automatically learn and construct a medicine recommendation model from massive historical case data and medicine interaction information, can reduce the incidence rate of adverse drug reactions, can flexibly and reasonably provide medicine recommendation information, provides accurate and effective decision basis for reasonable combination of clinical medicines, and provides effective guarantee for medication safety of users.
The invention is further explained in detail with the accompanying drawings and the embodiments; however, the method for modeling recommendation of drugs based on medical history data and drug interaction risk is not limited to the embodiment.
Drawings
FIG. 1 is a flow chart of the method of the present invention.
Detailed Description
In an embodiment, referring to fig. 1, a method for modeling drug recommendation based on medical record data and drug interaction risk according to the present invention includes the following steps:
a drug recommendation modeling method based on medical record data and drug interaction risks comprises the following steps:
s100: heterogeneous feature extraction operation is carried out on medical record data, outpatient medical record data and TWOSIDES drug interaction data;
s200: constructing a drug recommendation sequence generation model according to the extracted multi-source heterogeneous characteristics;
s300: and inputting the patient medical record into the medicine recommendation sequence generation model, and calculating and outputting the recommended medication information by the model.
Further, step S100 specifically includes the following steps:
s110: and extracting physiological information and diagnosis information of the patient from medical record data and outpatient medical record data as the characteristics of reasonable medication recommendation, wherein the physiological information is three characteristics of sex, age and weight, and the diagnosis information is diagnosis sequence characteristics.
S120: drug interaction relationships were established from the twosid drug interaction data as another feature of rational drug recommendations.
Further, step S110 specifically includes the following steps:
s111: the gender of the patient in the historical medical record data set is distinguished by F and M, which are designated by 0 and 1, respectively, for ease of computer processing. And code conversion is performed on the medicines, and ATC4 codes are used for representing different medicines.
S112: dictionaries are separately built for all diagnoses and drugs in the data set, and integers are used to represent the diagnoses and drugs in the finally generated data set.
Further, step S120 specifically includes the following steps:
s121: for case history data and the MIMIC _ III dataset, two drugs in the drug set are arbitrarily selected, assuming the encoded value is i, j, and then their ATC4 codes are mapped into the CID classification.
S122: searching the drug interaction database for the two types of drugs by using CID codes as key words to determine whether the two types of drugs have adverse drug interaction risks, and if so, Aij1, otherwise Aij0. And sequentially traversing all the medicine pairs in the database to generate a final adjacency matrix of the medicine interaction graph.
S123: and performing one-dimensional space transformation on the adjacent matrix characteristics of the drug interaction map. A graph neural network is built with two GCN layers, and after feature aggregation, non-linear permutations such as ReLU are applied to the generated output, and the final hidden representation of each node in the graph gets information from the subsequent neighborhood through stacking of multiple layers of GCNs. And finally, accessing a full-connection network to obtain one-dimensional vector output.
Further, step S200 specifically includes the following steps:
s210: symbols are defined such that X represents a diagnosis space and Y represents a medication space. R is a set of prescription records, whereinIs a diagnostic sequence, andis a drug sequence. I XkI and YkEach is XkAnd YkThe length of the sequence of (c).
S220: constructing a two-layer Transformer model, projecting Query, Key and Value to different spaces h times through linear transformation, and then obtaining h self-attribute matrixes through calculation. And finally, splicing the h matrixes, and multiplying the h matrixes by a weight matrix to generate a final attention matrix.
S230: based on a Transformer model, the extracted features are spliced and used as an input sequence, elements with the maximum probability in probability distribution are selected for each position through a multi-step coding and decoding process to serve as prediction results of the current position until an end symbol is generated or the set maximum sequence length is reached, and finally the generated sequence is the medication recommendation scheme of the model for the current patient.
Further, step S210 specifically includes the following steps:
s211: prescription record R { (X)1,Y1),(X2,Y2),…,(Xk,Yk)}
Further, step S220 specifically includes the following steps:
s221: when self-attention between input and output sequences is calculated, Q is equal to K and V, and the specific calculation process is as follows:
MultiHead(Q,K,V)=Concat(head1,…,headh)W0
headi=Atteation(QWi Q,KWi K,VWi V
further, step S300 specifically includes the following steps:
s310: and reading the patient medical record as the input of the drug recommendation sequence generation model, performing model calculation, and finally outputting the recommended medication information.
The above embodiments are only for further illustrating the present invention, but the present invention is not limited to the embodiments, and any simple modification, equivalent change and modification made to the above embodiments according to the technical spirit of the present invention fall within the protection scope of the technical solution of the present invention.
Claims (8)
1. A drug recommendation modeling method based on medical record data and drug interaction risks comprises the following steps:
s100: heterogeneous feature extraction operation is carried out on medical record data, outpatient medical record data and TWOSIDES drug interaction data;
s200: constructing a drug recommendation sequence generation model according to the extracted multi-source heterogeneous characteristics;
s300: and inputting the patient medical record into the medicine recommendation sequence generation model, and calculating and outputting the recommended medication information by the model.
2. The medical record data and drug interaction risk based drug recommendation modeling method of claim 1, wherein the step S100 specifically comprises the steps of:
s110: and extracting physiological information and diagnosis information of the patient from medical record data and outpatient medical record data as the characteristics of reasonable medication recommendation, wherein the physiological information is three characteristics of sex, age and weight, and the diagnosis information is diagnosis sequence characteristics.
S120: drug interaction relationships were established from the twosid drug interaction data as another feature of rational drug recommendations.
3. The medical record data and drug interaction risk based drug recommendation modeling method of claim 2, wherein the step S110 specifically comprises the steps of:
s111: the gender of the patient in the historical medical record data set is distinguished by F and M, which are designated by 0 and 1, respectively, for ease of computer processing. And code conversion is performed on the medicines, and ATC4 codes are used for representing different medicines.
S112: dictionaries are separately built for all diagnoses and drugs in the data set, and integers are used to represent the diagnoses and drugs in the finally generated data set.
4. The medical record data and drug interaction risk based drug recommendation modeling method of claim 2, wherein the step S120 specifically comprises the steps of:
s121: for case history data and the MIMIC _ III dataset, two drugs in the drug set are arbitrarily selected, assuming the encoded value is i, j, and then their ATC4 codes are mapped into the CID classification.
S122: searching the drug interaction database for the two types of drugs by using CID codes as key words to determine whether the two types of drugs have adverse drug interaction risks, and if so, Aij1, otherwise Aij0. And sequentially traversing all the medicine pairs in the database to generate a final adjacency matrix of the medicine interaction graph.
S123: and performing one-dimensional space transformation on the adjacent matrix characteristics of the drug interaction map. A graph neural network is built with two GCN layers, and after feature aggregation, non-linear permutations such as ReLU are applied to the generated output, and the final hidden representation of each node in the graph gets information from the subsequent neighborhood through stacking of multiple layers of GCNs. And finally, accessing a full-connection network to obtain one-dimensional vector output.
5. The medical record data and drug interaction risk based drug recommendation modeling method of claim 1, wherein the step S200 specifically comprises the steps of:
s210: symbols are defined such that X represents a diagnosis space and Y represents a medication space. R is a set of prescription records, whereinIs a diagnostic sequence, andis a drug sequence. I XkI and YkEach is XkAnd XkThe length of the sequence of (c).
S220: constructing a two-layer Transformer model, projecting Query, Key and Value to different spaces h times through linear transformation, and then obtaining h self-attribute matrixes through calculation. And finally, splicing the h matrixes, and multiplying the h matrixes by a weight matrix to generate a final attention matrix.
S230: based on a Transformer model, the extracted features are spliced and used as an input sequence, elements with the maximum probability in probability distribution are selected for each position through a multi-step coding and decoding process to serve as prediction results of the current position until an end symbol is generated or the set maximum sequence length is reached, and finally the generated sequence is the medication recommendation scheme of the model for the current patient.
7. The medical record data and drug interaction risk based drug recommendation modeling method of claim 5, wherein the step S220 specifically comprises the steps of:
s221: when self-attention between input and output sequences is calculated, Q is equal to K and V, and the specific calculation process is as follows:
MultiHead(Q,K,V)=Concat(head1,...,headh)W0
headi=Attention(QWi Q,KWi K,VWi V
8. the medical record data and drug interaction risk based drug recommendation modeling method of claim 1, wherein the step S300 specifically comprises: and reading the patient medical record as the input of the drug recommendation sequence generation model, performing model calculation, and finally outputting the recommended medication information.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010623195.0A CN111968715B (en) | 2020-06-30 | 2020-06-30 | Drug recommendation modeling method based on medical record data and drug interaction risk |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010623195.0A CN111968715B (en) | 2020-06-30 | 2020-06-30 | Drug recommendation modeling method based on medical record data and drug interaction risk |
Publications (2)
Publication Number | Publication Date |
---|---|
CN111968715A true CN111968715A (en) | 2020-11-20 |
CN111968715B CN111968715B (en) | 2022-11-01 |
Family
ID=73362387
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010623195.0A Active CN111968715B (en) | 2020-06-30 | 2020-06-30 | Drug recommendation modeling method based on medical record data and drug interaction risk |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111968715B (en) |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112420153A (en) * | 2020-11-26 | 2021-02-26 | 济南浪潮高新科技投资发展有限公司 | Method for improving traditional Chinese medicine prescription based on GAN |
CN112802573A (en) * | 2021-01-26 | 2021-05-14 | 中国科学技术大学 | Medicine package recommendation method and device, computer system and readable storage medium |
CN113314195A (en) * | 2021-05-28 | 2021-08-27 | 北京航空航天大学 | Staged dynamic drug administration matching system for chronic diseases |
CN113343590A (en) * | 2021-07-08 | 2021-09-03 | 华北电力大学(保定) | Wind speed prediction method and system based on combined model |
CN113724830A (en) * | 2021-08-31 | 2021-11-30 | 平安国际智慧城市科技股份有限公司 | Medicine taking risk detection method based on artificial intelligence and related equipment |
CN114628002A (en) * | 2022-03-18 | 2022-06-14 | 东南大学 | Drug recommendation method based on replication mechanism and hierarchy selection strategy |
CN114678140A (en) * | 2020-12-24 | 2022-06-28 | 宏碁股份有限公司 | Calculation method of high-risk medication path |
CN118197535A (en) * | 2024-05-17 | 2024-06-14 | 首都医科大学附属北京友谊医院 | Antibiotic combined medication recommendation model and application thereof |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107341345A (en) * | 2017-06-22 | 2017-11-10 | 厦门大学 | A kind of clinical rational drug use methods of risk assessment based on electronic health record big data |
CN108630322A (en) * | 2018-04-27 | 2018-10-09 | 厦门大学 | Drug interaction modeling and methods of risk assessment, terminal device and storage medium |
GB201904949D0 (en) * | 2019-04-08 | 2019-05-22 | Pankhania Anand Mansukh | Method of minimising patient risk |
CN111145906A (en) * | 2019-12-31 | 2020-05-12 | 清华大学 | Item determination method, related device and readable storage medium |
CN111191020A (en) * | 2019-12-27 | 2020-05-22 | 江苏省人民医院(南京医科大学第一附属医院) | Prescription recommendation method and system based on machine learning and knowledge graph |
-
2020
- 2020-06-30 CN CN202010623195.0A patent/CN111968715B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107341345A (en) * | 2017-06-22 | 2017-11-10 | 厦门大学 | A kind of clinical rational drug use methods of risk assessment based on electronic health record big data |
CN108630322A (en) * | 2018-04-27 | 2018-10-09 | 厦门大学 | Drug interaction modeling and methods of risk assessment, terminal device and storage medium |
GB201904949D0 (en) * | 2019-04-08 | 2019-05-22 | Pankhania Anand Mansukh | Method of minimising patient risk |
CN111191020A (en) * | 2019-12-27 | 2020-05-22 | 江苏省人民医院(南京医科大学第一附属医院) | Prescription recommendation method and system based on machine learning and knowledge graph |
CN111145906A (en) * | 2019-12-31 | 2020-05-12 | 清华大学 | Item determination method, related device and readable storage medium |
Non-Patent Citations (1)
Title |
---|
王磊等: "生物信息技术在药物相互作用风险研究中的应用进展", 《中国药理学与毒理学杂志》 * |
Cited By (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112420153A (en) * | 2020-11-26 | 2021-02-26 | 济南浪潮高新科技投资发展有限公司 | Method for improving traditional Chinese medicine prescription based on GAN |
CN112420153B (en) * | 2020-11-26 | 2022-11-15 | 山东浪潮科学研究院有限公司 | Method for improving traditional Chinese medicine prescription based on GAN |
CN114678140A (en) * | 2020-12-24 | 2022-06-28 | 宏碁股份有限公司 | Calculation method of high-risk medication path |
CN112802573A (en) * | 2021-01-26 | 2021-05-14 | 中国科学技术大学 | Medicine package recommendation method and device, computer system and readable storage medium |
CN112802573B (en) * | 2021-01-26 | 2023-06-20 | 中国科学技术大学 | Medicine package recommendation method, device, computer system and readable storage medium |
CN113314195A (en) * | 2021-05-28 | 2021-08-27 | 北京航空航天大学 | Staged dynamic drug administration matching system for chronic diseases |
CN113343590A (en) * | 2021-07-08 | 2021-09-03 | 华北电力大学(保定) | Wind speed prediction method and system based on combined model |
CN113343590B (en) * | 2021-07-08 | 2022-03-22 | 华北电力大学(保定) | Wind speed prediction method and system based on combined model |
CN113724830A (en) * | 2021-08-31 | 2021-11-30 | 平安国际智慧城市科技股份有限公司 | Medicine taking risk detection method based on artificial intelligence and related equipment |
CN113724830B (en) * | 2021-08-31 | 2024-04-30 | 深圳平安智慧医健科技有限公司 | Medication risk detection method based on artificial intelligence and related equipment |
CN114628002A (en) * | 2022-03-18 | 2022-06-14 | 东南大学 | Drug recommendation method based on replication mechanism and hierarchy selection strategy |
CN118197535A (en) * | 2024-05-17 | 2024-06-14 | 首都医科大学附属北京友谊医院 | Antibiotic combined medication recommendation model and application thereof |
Also Published As
Publication number | Publication date |
---|---|
CN111968715B (en) | 2022-11-01 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111968715B (en) | Drug recommendation modeling method based on medical record data and drug interaction risk | |
CN111191020B (en) | Prescription recommendation method and system based on machine learning and knowledge graph | |
Banerjee et al. | Comparative effectiveness of convolutional neural network (CNN) and recurrent neural network (RNN) architectures for radiology text report classification | |
Shickel et al. | Deep EHR: a survey of recent advances in deep learning techniques for electronic health record (EHR) analysis | |
Harerimana et al. | Deep learning for electronic health records analytics | |
Scheurwegs et al. | Selecting relevant features from the electronic health record for clinical code prediction | |
Wu et al. | Medical knowledge graph: Data sources, construction, reasoning, and applications | |
Gangavarapu et al. | FarSight: long-term disease prediction using unstructured clinical nursing notes | |
Sundararaman et al. | Novel approach to predict hospital readmissions using feature selection from unstructured data with class imbalance | |
CN112035627B (en) | Automatic question and answer method, device, equipment and storage medium | |
Haque et al. | Identifying health insurance claim frauds using mixture of clinical concepts | |
CN113223656A (en) | Medicine combination prediction method based on deep learning | |
Stylianou et al. | EBM+: Advancing Evidence-Based Medicine via two level automatic identification of Populations, Interventions, Outcomes in medical literature | |
CN114783603A (en) | Multi-source graph neural network fusion-based disease risk prediction method and system | |
CN114912887A (en) | Clinical data entry method and device based on electronic medical record | |
Cheung et al. | Deep learning from electronic medical records using attention-based cross-modal convolutional neural networks | |
US11544619B2 (en) | Dimension reduction of claims data | |
CN115240873A (en) | Medicine recommendation method based on machine learning, electronic equipment and computer-readable storage medium | |
Teng et al. | Few-shot ICD coding with knowledge transfer and evidence representation | |
CN113704481A (en) | Text processing method, device, equipment and storage medium | |
Ren et al. | Mortality prediction in ICU using a stacked ensemble model | |
US20130253892A1 (en) | Creating synthetic events using genetic surprisal data representing a genetic sequence of an organism with an addition of context | |
Yu et al. | AKA-SafeMed: A safe medication recommendation based on attention mechanism and knowledge augmentation | |
CN114627993A (en) | Information prediction method, information prediction device, storage medium and computer equipment | |
Ma et al. | Predicting heart failure in-hospital mortality by integrating longitudinal and category data in electronic health records |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |