An electronic medical record data governance method and device for childhood epilepsy

Abstract

The present application belongs to the technical field of medical data governance, and provides an electronic medical record data governance method and device for children with epilepsy. The method can acquire electronic medical record data from multiple heterogeneous data sources, construct multiple keyword dictionaries related to epilepsy for identifying and extracting information, thereby governing data from each data source through the keyword dictionaries, further generating structured data for clinical analysis of epilepsy, and displaying the epilepsy-related data of the patient in the form of a visualized graph based on the structured data. The scheme proposed in the present application converts the original electronic medical record data, which is scattered, chaotic and difficult to directly utilize, into structured data that is centralized, regularized and directly usable for clinical analysis and decision-making of children with epilepsy, solves the pain point of difficulty in integrating and converting medical data, improves the intuitiveness and efficiency of epilepsy diagnosis and treatment through visual means and provides strong data support, and can also provide analyzable and high-quality data for scientific research.

Description

Technical Field

[0001] This invention belongs to the field of medical data governance technology, specifically relating to a method and apparatus for governing electronic medical record data of children with epilepsy. Background Technology

[0002] Epilepsy is a common chronic neurological disorder caused by abnormal electrical activity in the brain. Its clinical characteristic is recurrent seizures without apparent cause. Symptoms are diverse and may include altered consciousness, involuntary movements, and abnormal behavior. Notably, the incidence of epilepsy in children is significantly higher than in adults, and seizures can have serious negative impacts on a child's developing cognitive, emotional, and social abilities.

[0003] Currently, anti-epileptic drug therapy is the most important treatment for epilepsy. The process usually follows a phased strategy: it starts with monotherapy, and if it is ineffective or causes side effects, the drug is changed or combination therapy is used. If multiple drug treatments fail, it may be diagnosed as drug-resistant epilepsy, and then the ketogenic diet or surgery may be considered.

[0004] In clinical practice, especially during initial monotherapy, selecting the most suitable regimen for patients from more than 30 candidate drugs is a huge challenge. Achieving personalized medication and improving treatment outcomes requires the support of large-scale clinical data, especially real-world data. Electronic medical record systems, as the core of hospital informatization, record massive amounts of patient information, including outpatient and emergency medical records, inpatient medical records, pharmacy records, diagnostic information, and EEG and imaging examinations. They should be a valuable resource to support clinical research and practice. However, if existing electronic medical record data is to be used for in-depth analysis and decision support in childhood epilepsy, challenges such as highly heterogeneous and fragmented data, difficulty in data utilization, and insufficient data presentation capabilities are encountered. Summary of the Invention

[0005] To overcome the shortcomings of the prior art, this invention proposes a method for managing electronic medical record data in children with epilepsy, the method comprising:

[0006] Electronic medical record data from multiple heterogeneous data sources are acquired, and multiple epilepsy-related keyword dictionaries are constructed for identification and / or information extraction. The data sources include outpatient and emergency medical records storing unstructured data, inpatient medical records storing unstructured data, and / or pharmacy records storing semi-structured data.

[0007] Based on the keyword dictionary, unstructured data in the outpatient and emergency medical records and / or the inpatient medical records are processed to obtain medical record data, and semi-structured data in the pharmacy records are processed based on the keyword dictionary to obtain medication data;

[0008] Structured data for epilepsy clinical analysis is generated based on the medical record data and medication data, and the patient's epilepsy-related data is displayed in the form of visualizations based on the structured data.

[0009] Specifically, the outpatient and emergency medical records and / or the inpatient medical records include several data entries, each data entry corresponding to a record time information. The method for managing unstructured data in the outpatient and emergency medical records and / or the inpatient medical records includes:

[0010] Data entries with the same specified information in the same outpatient or inpatient medical record are grouped into the same duplicate entry group. The data entry with the latest recording time information in each duplicate entry group is retained, and the remaining data entries are deleted.

[0011] The system uses a custom function based on regular expressions to identify the inherent information in each data entry, confirms whether there are any anomalies in the inherent information of each data entry, and deletes the data entry with the anomaly when anomalies are found, or supplements and / or corrects the inherent information in the data entry with the anomaly based on data information from other data sources.

[0012] Preferably, each of the outpatient / emergency medical records or the inpatient medical records corresponds to identification information, the inherent information including name, outpatient number, inpatient number, gender, age, height, and weight. A method for confirming whether the inherent information in any of the data entries is abnormal includes:

[0013] Confirm whether there are any missing inherent information in the data entry and the corresponding record time information of the data entry;

[0014] Confirm whether the name, outpatient number, and inpatient number in the data entry are consistent with the identification information corresponding to the outpatient or inpatient medical record from which the data entry originates;

[0015] Confirm whether the name, outpatient number, and inpatient number recorded in the data entry are consistent with other data entries contained in the outpatient or inpatient medical record that is the source of the data entry;

[0016] And / or, confirm whether the name, outpatient number, and inpatient number recorded in the data entry are consistent with the data information from other data sources.

[0017] Furthermore, based on the keyword dictionary, unstructured data in the outpatient and emergency medical records and / or the inpatient medical records is processed to obtain medical record data, including:

[0018] The free text in each data entry is segmented into short sentences according to punctuation marks. A string matching method based on regular expressions is used to sequentially identify whether each short sentence contains an event description based on the keyword dictionary. Specified data is extracted from the short sentences containing the event description, the terminology of all events is standardized, and the data is summarized to generate the medical record data. The event description includes descriptions of epileptic seizures, epileptic seizure states without seizure descriptions, and drug treatment-related events such as the initiation of anti-epileptic drug use, drug reduction, drug discontinuation, missed doses, or self-discontinuation of drug use.

[0019] Specifically, the pharmacy records include several medication entries. Medication data is obtained by processing the semi-structured data in the pharmacy records based on the keyword dictionary, including:

[0020] The drug information of each medication entry in the pharmacy record is identified. The medication information in each medication entry and the medication information keywords in the keyword dictionary are compared sequentially using a string comparison method based on regular expressions. The drug name information is confirmed and standardized through the comparison results. Medication entries that meet the first preset condition are deleted, and medication entries that meet the second preset condition are supplemented and / or corrected.

[0021] The medication information in each medication entry is compared with the medication information keywords in the keyword dictionary using a string comparison method based on regular expressions. The comparison results are used to determine whether there are any abnormalities in the medication information in each medication entry. If there are any abnormalities, the medication entry with abnormalities is deleted, or the medication information in the medication entry with abnormalities is supplemented based on data information from other data sources.

[0022] Based on each of the aforementioned medication entries, the administration method, frequency of administration, and dosage unit are standardized, and the average daily dose and prescription duration for each medication entry are calculated. At the same time, emergency medication and long-term medication are distinguished, and based on the administration method, oral medications and injectable medications used in emergency situations to control epileptic seizures are also distinguished, thereby generating the corresponding medication data.

[0023] Preferably, the method for calculating the average daily dose for each medication item is based on the normalized drug dose according to the patient's weight value on the query date corresponding to each medication item, and the method for determining the patient's weight value on the query date includes:

[0024] If the outpatient or inpatient medical records contain the patient's weight information, determine the start and end time periods and each record date corresponding to the weight information, filter the weight information, fit a locally weighted regression model with the patient's age as the independent variable and the filtered weight information as the dependent variable, identify and delete abnormal weight values ​​in the weight information by calculating the residuals of the locally weighted regression model, and generate a fitted curve based on the weight values ​​retained in the weight information.

[0025] If the query date is included in the recording time period and there is a corresponding recording date for the query date, the patient's weight value on the query date is obtained from the fitted curve; if the query date is included in the recording time period but there is no corresponding recording date for the query date, linear interpolation is performed through the fitted curve to obtain the patient's weight value on the query date.

[0026] If the query date is not included in the recorded time period or the patient's outpatient and inpatient medical records do not contain the patient's weight information, and the patient's age is not greater than a preset age value, the patient's weight value on the query date is determined based on the weight value obtained by fitting the weight data of all children with epilepsy of the same sex as the patient.

[0027] If the query date is not included in the recorded time period or the patient's outpatient or inpatient medical records do not contain the patient's weight information, and the patient's age is greater than a preset age value, the patient's weight on the query date is determined based on the weight reference value for healthy children.

[0028] Furthermore, the data source also includes medical record diagnoses, electroencephalogram (EEG) examinations, and imaging examinations. The medical record diagnoses include several diagnostic data entries. The EEG examinations include several EEG examination data in the form of routine EEG, evoked EEG, and / or video EEG. The imaging examinations include several brain CT and brain MRI imaging data. The method further includes:

[0029] Diagnostic data containing the keyword "epilepsy" is found by string matching based on regular expressions, and the epileptic seizure type, accompanying non-epileptic seizure, status epilepticus, epilepsy syndrome and epilepsy comorbidity corresponding to each diagnostic data containing the keyword "epilepsy" are identified based on the keyword dictionary to obtain diagnostic data.

[0030] The EEG data are cleaned, and the examination results, background, waveforms, and lobes are identified. The EEG results are classified as normal, abnormal, and borderline. The background is classified as normal, slowed, and poor. The waveforms are classified as epilepsy-related or epilepsy-unrelated. The lobes are classified as frontal, temporal, parietal, occipital, frontotemporal junction, temporoparietal junction, frontoparietal junction, parieto-occipital junction, pituitary, hippocampus, insula, and pineal gland. EEG data are obtained based on the identification and classification results.

[0031] Clean the image data, identify the examination results and brain lobe regions in the cleaned image data to distinguish the image data as normal or abnormal images, identify the brain lobe regions where abnormal phenomena occur, and obtain image data based on the identification and distinction results.

[0032] The structured data is generated based on the diagnostic data, the electroencephalogram data, the imaging data, the medical record data, and the medication data.

[0033] Specifically, based on the structured data, the patient's epilepsy-related data are displayed in the form of visualizations, including:

[0034] Based on the structured data, a medication timeline for the patient is constructed in a visualization. The start date and duration of prescriptions for various anti-epileptic drugs corresponding to the patient are marked in the visualization. The medication events and the dates of seizures and seizure-free periods for the patient in the outpatient and inpatient medical records are also marked in the visualization. The medication timelines for emergency and long-term medications are drawn with different colors, and the outpatient and inpatient prescriptions are marked with different symbols.

[0035] The present invention also proposes an electronic medical record data management device for epilepsy, the device comprising:

[0036] The acquisition module acquires electronic medical record data from multiple heterogeneous data sources and constructs multiple keyword dictionaries for identifying and extracting information; the data sources include outpatient and emergency medical records storing unstructured data, inpatient medical records storing unstructured data, and / or pharmacy records storing semi-structured data.

[0037] The processing module is used to process unstructured data in the outpatient and emergency medical records and / or the inpatient medical records based on the keyword dictionary to obtain medical record data, and to process semi-structured data in the pharmacy records based on the keyword dictionary to obtain medication data.

[0038] The analysis module is used to generate structured data for clinical analysis of epilepsy based on the medical record data and the medication data, and to display the patient's epilepsy-related data in the form of visualizations based on the structured data.

[0039] The present invention also proposes a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the electronic medical record data management method for childhood epilepsy as described above.

[0040] The present invention has at least the following beneficial effects:

[0041] The proposed solution can acquire data from multiple independent systems such as outpatient, emergency, inpatient, pharmacy, and imaging, solving the most common data fragmentation problem within medical institutions. This allows each patient's medical records, medication information, and various examination information to be linked together to form a more complete personal profile. Furthermore, it can accurately extract relevant information from the data through a keyword dictionary related to epilepsy, transforming it into substantial structured data and displaying it in the form of visualizations. This greatly improves the efficiency of clinical diagnosis and treatment, supports more precise clinical decision-making and personalized treatment, and provides high-quality analyzable data for clinical research.

[0042] Furthermore, the proposed solution can effectively eliminate data redundancy in medical records, avoiding analytical biases caused by errors or outdated information not being covered in the medical records. Through regular expressions and user-defined functions, it can systematically and automatically detect missing, inconsistent, and abnormal information, discover and repair errors that are difficult to detect in a single data source, improve the accuracy and reliability of core information such as patient identification and epilepsy treatment outcomes, and segment free text into short sentences to accurately locate sentences describing key clinical events such as "epilepsy seizures" and "medication changes," and associate these events with specific times, generating highly structured medical record data that can be used to construct epilepsy course and medication timelines.

[0043] Regarding pharmacy records, the proposed solution can identify and delete invalid entries, and standardize the administration method, frequency of administration, and dosage unit, so that medication data of different specifications can be compared and analyzed together. By distinguishing medication types and calculating key indicators, pharmacy records can be transformed into quantifiable key pharmacological indicators for efficacy evaluation. This solution also provides the most accurate possible weight estimate for the timing of medication administration in children with epilepsy through a modeling strategy, thereby achieving precise medication analysis and enabling assessments of whether the drug has reached the therapeutic dose and whether there is a risk of insufficient or excessive dosage.

[0044] Building upon this foundation, this solution can further incorporate pathological diagnosis and other imaging examination data, enabling the final structured data to integrate information from multiple dimensions such as diagnosis, comorbidities, and imaging findings. This provides a data basis for constructing a comprehensive and three-dimensional clinical profile of the patient, while also aiding in the analysis of complex cases, the development of epilepsy treatment strategies, and clinical scientific research oriented towards precision medicine. The final visualization solution can integrate medication, clinical events, and diagnostic timelines into a single graph, facilitating healthcare professionals' understanding of the relationship between maintenance therapy and emergency interventions. It also helps analyze treatment patterns in different medical scenarios and provides an intuitive tool for assessing efficacy and identifying drug resistance.

[0045] Therefore, this invention proposes a method and device for managing electronic medical record data in children with epilepsy. The proposed solution transforms the scattered, chaotic, and difficult-to-use raw electronic medical record data into centralized, organized, and directly usable structured data for clinical analysis and decision-making in children with epilepsy. This solves the pain point of difficult integration and transformation of medical data, improves the intuitiveness and efficiency of epilepsy diagnosis and treatment through visualization, and provides strong data support. At the same time, it can provide analyzable and high-quality data for scientific research. Attached Figure Description

[0046] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0047] Figure 1 This is a schematic diagram of the process flow for the electronic medical record data management method for childhood epilepsy provided in Example 1;

[0048] Figure 2 A flowchart illustrating the methods for managing unstructured data in medical records;

[0049] Figure 3 A flowchart illustrating a method for managing semi-structured data in pharmacy records;

[0050] Figure 4 Example graph of patient P06831's weight record and LOWESS fitted curve;

[0051] Figures 5(a) and 5(b) are schematic diagrams containing the fitted values ​​of LOWESS weight for children with epilepsy and the reference values ​​of weight for normal children. Figure 5(a) is a schematic diagram containing the fitted values ​​of LOWESS weight for boys with epilepsy and the reference values ​​of weight for normal boys, and Figure 5(b) is a schematic diagram containing the fitted values ​​of LOWESS weight for girls with epilepsy and the reference values ​​of weight for normal girls.

[0052] Figure 6 Example graph showing the timeline of epileptic seizures and antiepileptic drug events for patient P06046;

[0053] Figure 7 Example graph of the timeline of anti-epileptic drug prescription dosage for patient P06046;

[0054] Figure 8 This is a schematic diagram of the module structure of the electronic medical record data management device for epilepsy provided in Example 3. Detailed Implementation

[0055] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0056] Various embodiments of the invention will be described more fully below. The invention may have various embodiments, and adjustments and changes may be made therein. However, it should be understood that there is no intention to limit the various embodiments of the invention to the specific embodiments disclosed herein, but rather the invention should be understood to cover all modifications, equivalents, and / or alternatives falling within the spirit and scope of the various embodiments of the invention.

[0057] In the following, the terms “comprising” or “may include” as used in various embodiments of the invention indicate the presence of the disclosed functions, operations, or elements, and do not limit the addition of one or more functions, operations, or elements. Furthermore, as used in various embodiments of the invention, the terms “comprising,” “having,” and their cognates are intended only to indicate a specific feature, number, step, operation, element, component, or combination of the foregoing, and should not be construed as primarily excluding the presence of one or more other features, numbers, steps, operations, elements, components, or combinations of the foregoing, or the possibility of adding one or more combinations of the foregoing.

[0058] In various embodiments of the invention, the expression "or" or "at least one of A and / or B" includes any combination or all combinations of the words listed simultaneously. For example, the expression "A or B" or "at least one of A and / or B" may include A, may include B, or may include both A and B.

[0059] The expressions used in the various embodiments of the present invention (such as "first," "second," etc.) may modify various constituent elements in the various embodiments, but do not limit the corresponding constituent elements. For example, the above expressions do not limit the order and / or importance of the elements. The above expressions are only used for the purpose of distinguishing one element from other elements. For example, a first user device and a second user device refer to different user devices, although both are user devices. For example, a first element may be referred to as a second element without departing from the scope of the various embodiments of the present invention, and similarly, a second element may also be referred to as a first element.

[0060] It should be noted that, in this invention, unless otherwise explicitly specified and defined, terms such as "installation," "connection," and "fixation" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0061] In this invention, those skilled in the art should understand that the terms indicating orientation or positional relationship in the text are based on the orientation or positional relationship shown in the accompanying drawings, and are only for the purpose of facilitating the description of this invention and simplifying the description, and are not intended to indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this invention.

[0062] The terminology used in the various embodiments of the invention is for the purpose of describing particular embodiments only and is not intended to limit the various embodiments of the invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Unless otherwise defined, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art to which the various embodiments of the invention pertain. The terms (such as those defined in a generally used dictionary) are to be interpreted as having the same meaning as in the context of the relevant technical field and are not to be interpreted as having an idealized or overly formal meaning, unless clearly defined in the various embodiments of the invention.

[0063] Example 1

[0064] Please see Figure 1This embodiment proposes a method for managing electronic medical record data in children with epilepsy. This method can cover multiple data sources, and the managed data includes important information needed for research. It can also be visually displayed in a graphical format, allowing medical staff and researchers to quickly understand the patient's seizure patterns and medication treatment. Simultaneously, it provides analyzable, high-quality data for research. The specific method includes:

[0065] S100: Acquire electronic medical record data from multiple heterogeneous data sources and construct multiple epilepsy-related keyword dictionaries for identification and / or information extraction.

[0066] In this embodiment, the data source includes outpatient and emergency medical records storing unstructured data, inpatient medical records storing unstructured data, and / or pharmacy records storing semi-structured data. The outpatient and emergency medical records and / or inpatient medical records include several data entries, each corresponding to a recording time information. The pharmacy records include several medication entries. In an optional embodiment, the data source also includes examination reports such as electroencephalogram (EEG) examinations and imaging examinations. The medical record diagnosis includes several diagnostic data entries, the EEG examination includes several EEG data sets, and the imaging examination includes several imaging data sets.

[0067] Each patient has demographic information such as gender, date of birth, outpatient number, inpatient number, and age at consultation. The patient's outpatient and emergency medical records can include the contents of each outpatient or emergency visit, such as the patient's name, consultation time, age at consultation, outpatient serial number, medical record recording time, chief complaint, present illness, past medical history, physical examination, clinical auxiliary examination, Western medicine diagnosis, and doctor's orders.

[0068] The patient's inpatient medical record can include the contents of each hospitalization, which is divided into three parts: admission record, progress record and discharge record. It includes information such as the patient's name, hospitalization time, age at hospitalization, hospitalization serial number, chief complaint, present illness, past medical history, and discharge instructions.

[0069] Patient pharmacy records can include every prescription item issued when the patient visits or is hospitalized, such as the time of visit, age of visit, outpatient or inpatient record number, type of prescription, name of medication, single dose, unit of dosage, frequency of use, drug specifications, total dose, etc.

[0070] The patient's examination report may include the results of electroencephalogram (EEG) and imaging examinations performed during the patient's outpatient visit or hospitalization, such as brain CT and brain MRI. It may also include information such as examination time, age of the patient, outpatient or inpatient record number, examination items, examination site, examination findings, and examination conclusions.

[0071] S200: Obtain medical record data by processing unstructured data in outpatient and / or inpatient medical records based on keyword dictionaries, and obtain medication data by processing semi-structured data in pharmacy records based on keyword dictionaries.

[0072] S300: Generates structured data for clinical analysis of epilepsy based on medical record data and medication data, and displays the patient's epilepsy-related data in the form of visualizations based on the structured data.

[0073] Step S300 describes displaying the patient's epilepsy-related data in the form of a visualization based on structured data. Specifically, the visualization can construct the patient's medication timeline based on the structured data, mark the start date and duration of the prescriptions for various anti-epileptic drugs for the patient, and mark the drug events and the dates of seizures and seizure-free periods in the corresponding outpatient, emergency, and inpatient medical records. The medication timelines for emergency and long-term medications have different drawing colors, and the outpatient, emergency, and inpatient prescriptions have different drawing markers.

[0074] Specifically, please see Figure 2 The method for managing unstructured data in outpatient and / or inpatient medical records as described in step S200 includes:

[0075] S210: Group data entries with the same specified information in the same emergency room medical record or the same inpatient medical record into the same duplicate entry group, retain the data entry with the latest record time information in each duplicate entry group, and delete the remaining data entries.

[0076] In this embodiment, the specified information includes outpatient and inpatient serial numbers. In step S210, data entries with the same outpatient information are classified into the same duplicate entry group, and data entries with inpatient serial number information are classified into the same duplicate entry group. Only the data entries with the latest recording time information in the same duplicate entry group are retained.

[0077] In an optional implementation, if the same duplicate entry group includes multiple data entries with the same and latest record time information, then these data entries will be retained in step S210.

[0078] S220: Identify the inherent information in each data entry based on a custom function using regular expressions, confirm whether there are any anomalies in the inherent information of each data entry, and delete the data entry with the anomaly when anomalies are found, or supplement and / or correct the inherent information in the data entry with the anomaly based on data information from other data sources.

[0079] In this embodiment, each outpatient or inpatient medical record has corresponding identification information. The inherent information includes name, outpatient number, inpatient number, gender, age, height, and weight. Step S220 can pre-mark inherent information that is inconsistent with the identification information corresponding to the outpatient or inpatient medical record that is the source of the data entry.

[0080] In this embodiment, the method for step S220 to confirm whether there is an anomaly in the inherent information of any data entry may include, but is not limited to:

[0081] Confirm whether there are any missing inherent information in the data entries and the corresponding record time information for the data entries;

[0082] Confirm that the name, outpatient number, and inpatient number in the data entry are consistent with the identification information corresponding to the outpatient or inpatient medical records from which the data entry is sourced;

[0083] Confirm that the name, outpatient number, and inpatient number recorded in the data entry are consistent with other data entries contained in the outpatient or inpatient medical records from which the data entry is sourced;

[0084] And / or, verify that the name, outpatient number, and inpatient number recorded in the data entry are consistent with the data information from other data sources.

[0085] It should be noted that, theoretically, each patient has a unique outpatient number and inpatient number, and usually uses the same name. However, it is possible that the same patient may use multiple outpatient numbers and inpatient numbers, and patients may change their names. Therefore, when the same name is mapped to multiple outpatient numbers or multiple inpatient numbers, or the same outpatient number or the same inpatient number is mapped to different names, it can be confirmed that the mapping between the name and the outpatient number or inpatient number in the medical record is inconsistent. That is, the name, outpatient number, and inpatient number in the data entry are inconsistent with the identification information, other data entries, or data information from other data sources.

[0086] Preferably, the method proposed in this embodiment, after checking and matching the patient's name, outpatient number, and inpatient number, generates a unique identifier code corresponding to that patient, and records all names, outpatient numbers, and inpatient numbers used corresponding to that code. Simultaneously, the matching list is encrypted and saved. In subsequent use and display of data, all patient privacy information is removed, and only the code is displayed. Figure 4 The image shows the weight record and LOWESS fitted curve of patient P06831. Figure 6 The figure shows a timeline of seizures and antiepileptic drug events for patient P06046. Figure 7 The middle section shows a timeline of the prescribed dosage of anti-epileptic drugs for the patient with the code P06046.

[0087] Furthermore, the encrypted matching list of the patient's unique identification code, name, outpatient number, and inpatient number saved in this embodiment can be used for patient identity recognition and matching when governing newly added electronic medical record data in the future. Specifically, first, confirm whether there is a corresponding code for the name, outpatient number, and inpatient number information extracted from the new data, so as to integrate the new data of the same patient. If there is a newly admitted patient in the new data, a corresponding unique identification code will be generated for them in the same way.

[0088] In this embodiment, the implementation method of obtaining medical record data by governing unstructured data in outpatient and emergency medical records and / or inpatient medical records based on a keyword dictionary includes: splitting the free text in each data entry into short sentences according to punctuation marks, and using a regular expression-based string matching method to sequentially identify whether each short sentence contains an event description based on the keyword dictionary. Extract specified data from the short sentences where event descriptions appear, standardize the terms of all events, and summarize them to generate medical record data. In this embodiment, event descriptions include seizure status such as seizure descriptions and non-seizure descriptions, as well as drug treatment-related events such as the start of anti-seizure drug use, drug dosage reduction, drug discontinuation, missed doses, or self-stop of drugs.

[0089] Thus, the method proposed in this embodiment can construct independent variables and outcome variables required for research in subsequent analyses. For example, the time of the first seizure, the time interval from the first seizure to the first treatment, the number of seizures / frequency before the first treatment, etc. can all be used as independent variables related to drug efficacy, and the occurrence of seizures after treatment can be used as an outcome variable for drug efficacy research, that is, drug treatment failure.

[0090] Exemplarily, for sentence segmentation in each data entry, the sentence can be split according to full stops and line breaks, and then further split according to commas and semicolons. For each short sentence, the time can be sequentially identified. In this embodiment, the priority of each short sentence is determined according to the accuracy of date description from high to low. The ways to identify time can include identifying Chinese time, such as X years (X months X days); identifying English time, such as YYYY-MM-DD, YYYY.MM.DD, MM-DD, MM.DD; identifying Chinese age descriptions, such as X years old (X months), and X months / weeks / days / hours after birth; identifying Chinese relative time descriptions, such as X years / months / weeks / days / hours ago (after), including the Chinese character "half".

[0091] Furthermore, the method proposed in this embodiment can also perform post-processing on each data entry. For example, when post-processing data entries by date, the year of the data entry can be controlled within a reasonable time range, the month can be 1-12, and the date can be 1-31. When it is necessary to fill in the missing date in the data entry, if the date information identified by each short sentence is missing the year, the year identified by the previous short sentence will be used. If the month is missing, it will be filled with January, and if the date is missing, it will be filled with the 1st. In the relative time description, the time base can be readjusted as appropriate, and the relative time base of the next short sentence is the time identified by the previous short sentence.

[0092] Specifically, the pharmacy record includes several medication entries, each with corresponding identification information. The pharmacy record includes every medication prescription issued to a patient during their visit or hospitalization. The pharmacy record is independent of the electronic medical record system and is semi-structured data. Compared to handwritten electronic medical records, it has fewer human input errors and higher data quality. Therefore, the method proposed in this invention prioritizes the use of anti-epileptic drug information from the medication entries. Only when some data items are missing from the pharmacy record will the corresponding medical record be searched for and filled in. Furthermore, when there is a conflict between the information in the medical record and the medication entries, the method proposed in this embodiment will prioritize the information in the medication entries over the information in the medical record. Please refer to [link to relevant documentation]. Figure 3 The step S200, which involves using a keyword dictionary to process semi-structured data in pharmacy records to obtain medication data, includes:

[0093] S230: Identify the drug information of each medication entry in the pharmacy record, and sequentially compare the medication information in each medication entry with the medication information keywords in the keyword dictionary using a string comparison method based on regular expressions. Confirm and standardize the drug name information through the comparison results, delete the medication entries whose identification results meet the first preset condition, and supplement and / or correct the medication entries whose identification results meet the second preset condition.

[0094] In this embodiment, the deletion of medication entries whose identification results meet the first preset condition in step S230 includes:

[0095] Delete medication entries that identify multiple drug names simultaneously; medication entries that identify multiple drug names simultaneously usually do not involve drug prescriptions.

[0096] Delete medication entries related to drug injection rate adjustment or drug dosage adjustment;

[0097] Delete medication entries that do not specify medication frequency or dosage, or those with special medication frequencies, such as those with keywords like prn (long-term prescription, to be used when needed), SOS (to be used when needed, limited to one use, effective within 12 hours), or when necessary, as well as those with keywords that appear to be typos, such as sq and sqbxs.

[0098] Remove medication entries for hospitalized patients with special routes of administration, such as standby, retention enema, other, and as directed by the physician.

[0099] If a medication entry lacks medication frequency and the medication frequency cannot be filled by confirming the patient's corresponding outpatient or emergency medical records and the dosage and frequency of medications before and after, the medication entry will be deleted.

[0100] Delete medication entries in medical order records that lack a single dose or whose single dose cannot be accurately estimated;

[0101] Delete medication entries whose names, outpatient numbers, and inpatient numbers do not match the identification information corresponding to the pharmacy records from which the medication entries are sourced.

[0102] Supplementing and / or correcting medication entries whose identification results meet the second preset condition includes:

[0103] If a medication entry is missing medication frequency, the medication frequency should be filled in by confirming the patient's corresponding outpatient or emergency medical records and the dosage and frequency of medications before and after the entry, and the name, outpatient serial number and recording time in the entry record should be verified.

[0104] If a single dose is missing from the medication order record, for example, for non-ketogenic diet records that are missing or have irregular single doses, such as no single dose in outpatient prescription records, special doses in inpatient prescription records (e.g., interval adjustments, X times of prescription execution, required medication, to be taken to the operating room, taking XX tablets or pills, only numbers without dosage units), or no discharge medication record in inpatient records, if the single dose can be accurately estimated, then the single dose in the medication entry can be filled in by estimation. In an optional implementation, the method proposed in this embodiment can use string matching based on regular expressions. The system identifies whether the words "morning / noon / evening" appear in the inpatient medication record, and further identifies the doses at different times. The doses are added together to obtain the patient's total daily dose for that day, and the corresponding daily doses are corrected to change the daily single dose to the correct total daily dose, while the medication frequency is changed to once. It can also attempt to fill in the missing frequency entries in the discharge medication record based on the discharge record. If it cannot be filled in successfully, it further confirms all records related to the patient's medication to infer the missing medication frequency. If the discharge medical record also shows different dosages for morning, noon and evening in the discharge medication, the single dose of the discharge medication can be corrected at the same time.

[0105] S240: The drug information in each drug entry and the drug information keywords in the keyword dictionary are compared sequentially using a string comparison method based on regular expressions. The comparison results are used to confirm whether there are any abnormalities in the drug information in each drug entry. If there are abnormalities, the drug entry with abnormalities is deleted, or the drug information in the drug entry with abnormalities is supplemented based on data information from other data sources.

[0106] In this embodiment, if the comparison result of step S240 is that the comparison is unsuccessful, the drug keywords in the medication entry can be determined by manual interpretation, and the drug keywords corresponding to the medication entry can be extracted and added to the drug dictionary. The extracted drug keywords corresponding to the medication entry may include typos or incorrect characters. If the corresponding natural language processing rules cannot be defined for the medication entry, the method proposed in this embodiment can treat the medication entry as a special case and skip the treatment step, and directly output the result of the pre-manual judgment.

[0107] S250: Based on each medication item, the administration method, frequency of administration, and dosage unit are standardized, and the average daily dose and prescription duration for each medication item are calculated. At the same time, emergency medication and long-term medication are distinguished, and based on the administration method, long-term oral medication and injectable medication used in emergency situations to control epileptic seizures are distinguished, thereby generating corresponding medication data.

[0108] In this embodiment, step S250 categorizes drug administration methods into five types: oral solution, oral tablets, oral capsules, injection, and diet. The administration method for the ketogenic diet is diet. Methods for standardizing the administration methods of other drugs may include:

[0109] The administration method is confirmed through outpatient records. For example, for oral medications, the administration method (oral solution, oral tablet, oral capsule) is determined according to the drug type (oral liquid, tablet, capsule); the administration method for medication entries recorded as intramuscular injection, intravenous injection, intravenous drip, and intravenous bolus is determined as injection category; the administration method for medication entries including unconventional administration methods such as rectal suppositories and nasal sprays is determined by drug type and name, and drugs of solution and tablet type are respectively determined as oral solution category and oral tablet category;

[0110] The administration method is confirmed through hospitalization records. For example, the administration method (oral solution, oral tablet, oral capsule) is determined by the type of oral and sublingual medication (oral liquid, tablet, capsule); the administration method recorded as subcutaneous, intramuscular, intravenous, intravenous drip, intravenous bolus, intravenous pump, or intravenous drip maintenance of X mg / kg·h is determined as the injection category; and drugs with unconventional administration methods such as nebulization, nasal mist, and nasogastric feeding are determined as the corresponding oral solution, oral tablet, and injection categories by drug name and type.

[0111] The way medication frequency is written in electronic medical record medication entries is very diverse, including the conventional Chinese writing of "daily X times", as well as English writing of "qXh", "bid", "tid", etc. The method proposed in this embodiment can identify the medication frequency through natural language processing and unify it into a numerical value, that is, the number of times to take the medication per day. Special frequencies, such as "immediately executed" and "st", are considered to be executed only once, and "discontinued" is considered to be executed 0 times per day. If the medication frequency is missing in hospitalization, it is considered to be executed only once, thereby achieving standardized medication frequency.

[0112] The way units are written in electronic medical record medication entries is also very diverse, including Chinese and English, such as milligrams (mg). The method proposed in this embodiment can unify them all to English, namely mg, g, and ml. For medication entries that lack dosage units in the medical order record, the dosage unit can be inferred from the drug name and drug specification. For example, the ketogenic diet has no dosage unit, while the dosage unit of corticotropin is unit. At the same time, in order to facilitate the study of drug dosage, this method can further unify all dosage units to mg. The conversion from ml to mg can be determined based on the specific drug specification, thereby achieving standardized dosage units.

[0113] For example, one bottle of levetiracetam oral solution contains 150 ml and a total of 15 g of drug components, which is equivalent to a 100 mg drug dose per ml; one bottle of depakine sodium valproate oral solution manufactured by Sanofi contains 300 ml and a total of 12 g of drug components, which is equivalent to a 40 mg drug dose per ml; one bottle of trileoxacarbazepine oral solution manufactured by Novartis contains 100 ml and a total of 60 mg of drug components per ml, which is equivalent to a 60 mg drug dose per ml.

[0114] The average daily dose of a medication item can be obtained by multiplying the single dose by the daily frequency of medication. The number of days a prescription lasts can be obtained by first estimating the total dose of outpatient prescriptions based on the drug specifications and quantity, and then dividing the total dose of outpatient prescriptions by the average daily dose.

[0115] The duration of temporary and long-term inpatient medical orders can be determined by the discontinuation time. If the discontinuation time is empty, it is counted as 1 time, that is, the prescription duration is 0 days. If the long-term oral medications in the discharge medications in the inpatient record can be regarded as the prescription duration of 30 days, if they include emergency medications such as diazepam tablets and phenobarbital tablets, they can be regarded as the prescription duration of 7 days, and all injections can be regarded as the prescription duration of 7 days.

[0116] Furthermore, the method proposed in this embodiment supplements some special cases. For the ketogenic diet, it can be defined as having no dosage, and the prescription duration can be calculated from the start and end dates noted in the medical record. For records where the medication frequency is discontinued, it can be considered as an average daily dose of 0 and a prescription duration of 0 days. Since injections are administered in units of vials, if a single dose of an injection does not reach the dosage of one vial, it is counted as one vial. Since tablets can be taken in multiple doses, the prescription duration for tablets is calculated based on daily records. If the prescription duration is greater than 180 days, it is corrected to 180 days.

[0117] The antiepileptic drugs used in clinical studies of childhood epilepsy are long-term oral medications. Therefore, the drugs in the medication entries need to be classified into two categories: emergency medications such as injectable drugs and long-term medications. In this embodiment, diazepam and other injectable solutions excluding corticotropin are considered emergency medications, while other long-term oral medications excluding diazepam, corticotropin injections, and ketogenic diets are considered long-term medications. For phenobarbital, which can be used for both emergency treatment and long-term use, the method proposed in this embodiment can define phenobarbital in medication entries containing prescription information for oral phenobarbital for more than or equal to 30 days as long-term medication.

[0118] Unlike adult epilepsy, the daily dose of antiepileptic drugs used in clinical practice and scientific research on pediatric epilepsy must be normalized to body weight. Preferably, in this embodiment, the corresponding average daily dose can be calculated using body weight normalization, thereby enabling the targeted determination of the dosage unit for pediatric epilepsy medication. Specifically, the average daily dose for each medication item can also be normalized based on the patient's weight value on the corresponding query date. In clinical practice and scientific research on pediatric epilepsy, the corresponding dosage unit is mg / (kg·d). The calculation method is to divide the average daily dose by the patient's weight at the start of the medication prescription. The weight value can be obtained from outpatient and inpatient medical records and identified from the records using natural language processing. The method proposed in this embodiment can merge weight records from outpatient and inpatient medical records to generate the patient's weight value recorded by date.

[0119] Furthermore, when multiple weight records appear on the same date, the method proposed in this embodiment will prioritize using the weight record in the inpatient record as the weight value for that date; when multiple weight records on the same date are all from outpatient or emergency medical records or all from inpatient medical records, the mode, mean, or average of the modes with the same frequency of occurrence of the weight records can be taken as the weight value for that date; when multiple weight records on the same date have large differences, the weight record with smaller differences from the weight values ​​of the earlier and later dates can be selected as the weight value for that date.

[0120] Preferably, the method proposed in this embodiment can identify abnormal weight values. For example, it can examine weight records where the patient weighs no less than 80 kg and is no older than 12 years old, or fit a locally weighted scatterplot smoothing (LOWESS) model with the patient's age as the independent variable and the filtered weight information as the dependent variable. The residual z-score is calculated using the locally weighted regression model, and weight records with a residual z-score greater than or equal to 3 are marked as abnormal weight values ​​and deleted. In an optional implementation, the method proposed in this embodiment can repeat the LOWESS modeling process twice. Figure 4 The image shows the weight record of patient P06046, and five abnormal weight values ​​were identified by LOWESS, which were distributed between the ages of 10 and 11.

[0121] For patients without any weight records, the method proposed in this embodiment can use the weight records of all other patients to estimate the weight of the patient at a certain age. Since the weight of children with epilepsy who are on long-term medication treatment is lower than that of normal healthy children, it is reasonable to use the weight of other children with epilepsy to estimate the weight of patients with unknown weight records. At the same time, since there are differences in weight between boys and girls of the same age (see Figures 5(a)-5(b)), the method proposed in this embodiment can fit weight curves for boys and girls separately. That is, after deleting abnormal values ​​from the weight records of all boys and girls, two weight curves are established using the LOWESS model. Then, linear internal interpolation is used to obtain the weight value of any age for the weight prediction value of patients with unknown weight values.

[0122] Optionally, the prediction range for boys is 0 to 24.0450 years, and the prediction range for girls is 0.0 to 20.8750 years, with an interval of 0.0025 years (approximately 0.9 days). This ensures that there is a corresponding fitted weight value for each age and each day. In this embodiment, examples of fitted curves are provided, which are based on 13,739 weight records from 2,558 boys with epilepsy and 10,399 weight records from 1,799 girls with epilepsy and 8,351 fitted weight values ​​for girls. All fitted values ​​can be saved as a table for easy access by users.

[0123] As can be seen from Figures 5(a) and 5(b), there are fewer weight records for patients over 15 years old. Therefore, the fitted values ​​for patients over 15 years old show an unreasonable trend, that is, a linear upward trend with age. Theoretically, weight should follow an S-shaped curve with age, and the rate of weight increase should gradually slow down with age. Given that the weight growth trend of children with epilepsy over 15 years old has slowed down and the difference with healthy children has decreased, the weight reference value of healthy children can be used as the estimation basis in this case. Therefore, for patients over 15 years old with missing weight values, the method proposed in this embodiment uses the median weight value of normal children to fill the missing values. For example, the weight percentile values ​​of boys aged 0-36 months, girls aged 0-36 months, boys aged 2-18 years, and girls aged 2-18 years can be used from the "Practical Pediatrics of Zhu Futang" published by People's Medical Publishing House and edited by Jiang Zaifang, Shen Kunling, and Shen Ying to determine the patient's weight value.

[0124] In one alternative factual approach, methods for determining a patient's weight on the date of the query include:

[0125] If the patient's weight information is available in the outpatient or inpatient medical records, determine the start and end time periods and the date of each record for the corresponding weight information, filter the weight information, fit a locally weighted regression model with the patient's age as the independent variable and the filtered weight information as the dependent variable, identify and delete abnormal weight values ​​in the weight information by calculating the residuals of the locally weighted regression model, and generate a fitted curve based on the weight values ​​retained in the weight information.

[0126] If the query date is included in the recording time period and there is a corresponding record date, the patient's weight value on the query date is obtained from the fitted curve; if the query date is included in the recording time period but there is no corresponding record date, linear interpolation is performed through the fitted curve to obtain the patient's weight value on the query date.

[0127] If the query date is not included in the recorded time period or the patient's outpatient and inpatient medical records do not contain the patient's weight information, and the patient's age is not greater than the preset age value, the patient's weight value on the query date is determined based on the weight value obtained by fitting the weight data of all children with epilepsy of the same sex as the patient; in this embodiment, the preset age value is 15 years old.

[0128] If the query date is not included in the recorded time period or the patient's outpatient or inpatient medical records do not contain the patient's weight information, and the patient's age is greater than the preset age value, the patient's weight value on the query date will be determined based on the weight reference value for healthy children.

[0129] It should be noted that the recording time period in this embodiment specifically includes the time period between the earliest and latest recording time of the patient's weight information in the outpatient or inpatient medical records.

[0130] Furthermore, after combining data sources such as medical record diagnosis, electroencephalogram (EEG) examination, and imaging examination, step S200 of the method proposed in this embodiment includes:

[0131] The system uses a keyword dictionary to process unstructured data from outpatient and / or inpatient medical records to obtain medical record data, and uses a keyword dictionary to process semi-structured data from pharmacy records to obtain medication data. It searches for diagnostic data containing the keyword "epilepsy" using regular expression-based string matching, and uses the keyword dictionary to identify the epileptic seizure type, accompanying non-epileptic seizures, status epilepticus, epilepsy syndrome, and epilepsy comorbidities corresponding to each diagnostic data containing the keyword "epilepsy" to obtain diagnostic data. It also cleans various electroencephalogram (EEG) examination data, identifying the examination results, EEG background, EEG waveforms, and brain lobe locations in the cleaned EEG data. Finally, it cleans various imaging data, identifying the examination results and brain lobe locations in the cleaned imaging data to distinguish between normal and abnormal images, and obtains imaging data based on the identification and distinction results.

[0132] The method proposed in this embodiment uses the latest clinical guidelines, "Clinical Practice Guidelines - Epilepsy Section" (2023 revised edition), and refers to the epilepsy seizure types defined by the International League Against Epilepsy (ILAE) in 2017. A corresponding keyword dictionary is defined for each seizure type. The seizure type in the diagnostic project is identified by a custom natural language processing function. Patients without any seizure type record can be defined as "unknown" seizure type. At the same time, according to the guidelines, status epilepticus (SE) is divided into two categories according to symptomatology: with prominent motor symptoms and without prominent motor symptoms. The method also identifies 26 epilepsy syndromes and 3 common psychiatric comorbidities in childhood epilepsy as defined in the guidelines.

[0133] In this embodiment, epileptic seizure types include four main categories: generalized seizures, focal seizures, epileptic spasms, and reflex seizures. Generalized seizures include generalized tonic-clonic seizures (GTCS, also known as grand mal seizures), tonic seizures, clonic seizures, myoclonic seizures, atonic seizures, myoclonic-tonic-clonic seizures, myoclonic-atonic seizures, and absence seizures. Absence seizures can be further subdivided into typical absence, atypical absence, myoclonic absence, and oculoclonic absence. eyelid myoclonia);

[0134] Non-epileptic seizure types include syncope, psychogenic non-epileptic seizures (PNES), migraine, breath-holding episodes, transient ischemic attacks, sleep disorders, and tic disorders.

[0135] Status epilepticus, according to guidelines, can be divided into two main categories based on symptomatology: Category A, with prominent motor symptoms, includes A.1 Convulsive SE (CSE, equivalent to tonic-clonic SE), A.2 Myoclonic SE, A.3 Focal motor SE, A.4 Tonic SE, and A.5 Hypermotor SE; Category B, without prominent motor symptoms (i.e., non-convulsive SE, NCSE), includes B.1 NCSE with coma and B.2 NCSE without coma. If no relevant keywords appear in the patient's medical record, it can be considered that status epilepticus has not occurred.

[0136] Epilepsy syndromes include benign familial neonatal epilepsy (BFNE), benign familial infantile epilepsy (BFIE), benign infantile epilepsy (BIE), Ohtahara syndrome, early myoclonic encephalopathy (EME), epilepsy of infancy with migrating focal seizures (EIMFS), Dravet syndrome, infantile spasms, myoclonic epilepsy ininfancy, Lennox-Gastaut syndrome (LGS), myoclonic-atonic epilepsy (MAE), and childhood absence epilepsy. Epilepsy (CAE), eyelid myoclonic epilepsy (EME), myoclonic absence epilepsy, benign epilepsy in childhood with centrotemporal spikes (BECTS), Panayiotopoulos syndrome, late-onset occipital lobe epilepsy (Gastaut type), Landau-Kleffner syndrome (LKS), epileptic encephalopathy with continuous spike and waves during slow wave sleep (CSWS), juvenile absence epilepsy (JAE), and juvenile myoclonic epilepsy.The epilepsy syndromes identified include JME (Junior Myoclonus Epilepsy), epilepsy with generalized tonic-clonic seizures only, genetic epilepsy with febrile seizures plus (GEFS+), progressive myoclonus epilepsy (PME), Rasmussen syndrome, and febrile infection-related epilepsy syndrome (FIRES). In this embodiment, the keyword dictionary used to identify epilepsy syndromes includes the formal spellings, alternative names, common names, and earlier names of the syndromes mentioned in the guidelines, as well as corresponding English keywords, abbreviations, and misspellings.

[0137] Three common comorbid mental illnesses in childhood epilepsy include neurodevelopmental delay, cerebral palsy (CP), and autism spectrum disorder (ASD). Patients whose medical records do not contain relevant keywords can be considered to be free of these three mental illnesses. In addition, the method proposed in this embodiment can also identify whether a patient has a history of febrile seizures.

[0138] Electroencephalography (EEG) data includes various EEG examination data such as routine EEG, evoked EEG, and video EEG, all of which can reflect the abnormal state of the patient's brain.

[0139] For example, the background, waveforms, and cerebral lobe regions in the cleaned electroencephalogram are identified and distinguished, and the specific methods of distinction include:

[0140] Based on the examination results, the EEG is classified as normal, abnormal, or borderline; among them, the EEG with the examination result as borderline is the EEG with the examination result between normal and abnormal.

[0141] The method proposed in this embodiment can also distinguish the background of an EEG into normal background, slowed background, and poor background.

[0142] The method proposed in this embodiment can also identify whether the electroencephalogram (EEG) contains epilepsy-related abnormal waveforms such as spikes, sharp waves, sharp spikes, sharp-slow waves, spike-slow waves, sharp-slow waves, and high-grade dysrhythmia; whether it contains epilepsy-unrelated abnormal waveforms such as irregular delta waves, irregular theta waves, fast waves, slow waves, inhibitory brain waves, and lazy waves; and record abnormal brain regions in the EEG, such as the frontal lobe, temporal lobe, parietal lobe, occipital lobe, frontotemporal junction, temporoparietal junction, frontoparietal junction, parieto-occipital junction, pituitary gland, hippocampus, insula, and pineal gland.

[0143] Imaging data includes data from brain computed tomography (CT) and magnetic resonance imaging (MRI). Correspondingly, brain CT and MRI images can be classified as normal or abnormal, and the abnormal brain lobe locations can be recorded.

[0144] Step S300 includes the following:

[0145] Structured data for clinical analysis of epilepsy is generated by combining diagnostic data, electroencephalogram data, imaging data, medical record data, and medication data. Based on the structured data, the patient's epilepsy-related data is displayed in the form of visualization diagrams.

[0146] For example, the epilepsy-related keyword dictionary and entries used in the method proposed in this embodiment include:

[0147] The dictionary of antiepileptic drugs includes the following entries:

[0148] 'Valproic acid': ['Valproic acid', 'VPA', 'Depakine', 'DPK', 'DEPAKINE', 'Sanofi'],

[0149] 'Oxcarbazepine': ['Oxcarbazepine','OXC','Qulepi','Oxcarbazepine',],

[0150] 'Levobrinacitan': ['Levobrinacitan', 'LEV', 'Keppra', 'Giyk', 'Zoitan', 'Keppra'],

[0151] 'Topiramate': ['Topiramate', 'TPM', 'Topiramate', 'Topiramate', 'Topiramate'],

[0152] 'Lamotriazine': ['Lamotriazine', 'Lamotriazine', 'LTG', 'Lapiton'],

[0153] 'Carbamazepine': ['Carbamazepine', 'CBZ', 'Drido'],

[0154] 'Adrenocorticotropic hormone': ['Adrenocorticotropic hormone', 'ACTH'],

[0155] 'Ketogenic diet': ['ketogenic', 'KD', 'KDT'],

[0156] 'Clobarzan':['Clobarzan','CLB'],

[0157] 'Aminohexenoic acid': ['Aminohexenoic acid', 'VGB', 'Xibaoning', 'Xibaoning', 'SABRIL', 'Sabril', 'Aminohexenoic acid'],

[0158] Phenobarbital: [Phenobarbital, PB, Luminal]

[0159] 'Zonisamide': ['Zonisamide','ZNS','Fodining'],

[0160] 'Lacosamide': ['Lacosamide', 'LCM', 'Vipate', 'Xinkang', 'Lacoamide', 'Lacosamide', 'Lacosamide'],

[0161] 'Phenytoin':['phenytoin','PHT','Jinlixheng'],

[0162] 'Laurazepam':['Laurazepam','LRZ','Xinyi'],

[0163] 'Diazepam': ['Diazepam', 'DZP'],

[0164] 'Gabapentin': ['Gabapentin', 'Gabapentin', 'GBP'],

[0165] 'Nitrozepam': ['Nitrozepam', 'NZP'],

[0166] 'Clonazepam': ['Clonazepam', 'CZP'],

[0167] 'Midazolam':['Midazolam','MDZ'],

[0168] 'Propofol': ['Propofol', 'PRO', 'Yuejiabo', 'AstraZeneca'],

[0169] 'Perampanel': ['Perampanel', 'Vecta', 'PER']

[0170] The dictionary of medication frequency includes entries including...

[0171] "Twice a day": "twice a day", "twice a day", "twice a day", "twice a day", "twice every 12 hours", ...

[0172] "Once a day": "Once a night", "Before bed", "qn", "Once a night", "Once a day", "Once a day", "Once a day", "Once a day", " / d", "Once a day", "Once a day", "Once a day", "Once a day", "Operate before bed", "Once a day" "Once a night", "Once a day", "Immediately", "3am", "5am", "6am", "7am", "8am", "9am", "10am", "11am", "12am", "1pm", "2pm", "3pm", "4pm", "5pm", "6pm", "7pm", "8pm", "9pm", "10pm", "11pm", "qn", "qon", "qd", "qd8", "qd12", "qd21", "st";

[0173] "3 times a day": 1 time every 8 hours", "3 times a day", "3 times a day", "3 times a day", "q8h", "3 times a day", "once every 8 hours", "8 hours apart", "3 times / day", "8 hours apart", "tid", "3 times a day", "once every 8 hours", "3 times a day", "3 times a day", "3 times a day", "once every 8 hours", "q8h", "tid", "3 times a day";

[0174] “4 times a day”: 'q6h', '4 times / day', 'once every 6 hours', 'qid';

[0175] "Once every other day": "Once every other morning", "Once every other day", "Once every other day";

[0176] "When necessary": "When obviously agitated", "When necessary", "To prevent convulsions during fever", "On standby", "Not for oral administration".

[0177] The dictionary of epileptic seizures includes the following entries:

[0178] 'Still having seizures', 'Still having epileptic seizures', 'Sessions significantly reduced', 'Sessions occurring daily', 'Still having seizures', 'Having epileptic seizures', 'Still having intermittent convulsions', 'Head-nodding-like seizures', 'Having recurrent seizures', 'Still having head-nodding-like embracing seizures', 'Still having intermittent seizures', 'Sessions increased compared to before', 'Sessions more frequent than before', 'Sessions occurring', 'Still having occasional seizures', 'Occurring daily', 'Sessions reduced compared to before', 'No febrile seizures', 'Small seizures still present', 'Still having frequent seizures', 'Sessions increased', 'Chattering seizures', 'Still having frequent epileptic seizures', 'Recurring convulsions', 'With clinical convulsive seizures', 'Clinical seizures', 'Seizures still not relieved', 'Last seizure', 'Last seizure', 'No reduction in seizures', 'Having seizures', 'Seizures', 'Convulsions', 'Convulsive seizures'.

[0179] The dictionary of epileptic seizures includes entries including

[0180] 'No seizures', 'No seizures', 'No recurrence of seizures', 'No more seizures', 'No more seizures', 'No more seizures', 'No epileptic seizures', 'No more epileptic seizures', 'No more epileptic seizures', 'No more epileptic seizures', 'No more epileptic seizures', 'No more epileptic seizures', 'No more epileptic seizures', 'No more epileptic seizures', 'No more epileptic seizures', 'No more epileptic seizures', 'No more epileptic seizures', 'No more epileptic seizures', 'No more seizures', 'No more seizures', 'No more seizures', 'No more seizures', 'No more seizures', 'No more seizures', 'No more seizures', 'No more seizures', 'Stable condition', 'No more seizures', 'No more seizures', 'No more seizures', 'No more seizures', 'No more seizures' 'Seizures', 'Seizures have significantly improved', 'Frequency of seizures has improved', 'No further seizures', 'Seizures have not recurred to date', 'Seizures are under control and have not recurred', 'No seizures', 'Seizures have not occurred', 'No seizures', 'Seizures are under control', 'Seizures are under control', 'No seizures', 'No recurrence of seizures', 'No recurrence of seizures', 'No seizures', 'No recurrence of seizures', 'No recurrence of seizures', 'No recurrence of seizures', 'No epileptic seizures', 'Seizures have not yet occurred', 'No recurrence of seizures', 'No seizures observed', 'No recurrence of seizures', 'No recurrence of epileptic seizures', 'Seizures have not recurred to date', 'Seizures are not occurring'.

[0181] The dictionary of epileptic seizure symptoms includes the following entries:

[0182] 'Gaze fixed,' 'Gaze fixed,' 'Cyanosis of the lips,' 'Unresponsive,' 'Immuted consciousness,' 'Cyanosis of the face,' 'Cyanosis of the lips,' 'Gaze fixed to the right,' 'Gaze fixed to the left,' 'Tonic seizure,' 'Drooling,' 'Twitching and rigidity of the limbs,' 'Limb convulsions,' 'Limb convulsions,' 'Gaze fixed upward,' 'Frothing at the mouth,' 'Paroxysmal weakness of the right upper limb,' 'Deviation of the corner of the mouth to the left.'

[0183] A dictionary of special terms related to epileptic seizures, including the following entries:

[0184] 'Convulsions manifested as: no convulsions for 2-3 days', 'no / no loss of consciousness, pale and cyanotic face, sweating, no frothing at the mouth, weakness in the limbs, no obvious convulsions', 'accompanied by loss of consciousness, pale and cyanotic face, sweating, no frothing at the mouth, weakness in the limbs, no obvious convulsions', 'another episode of paroxysmal unresponsive seizures yesterday and today', 'myoclonic seizures and atypical absence seizures', 'seizures occurred in both febrile and non-febrile states', 'seizures decreased or ceased during infection or fever', 'two seizures occurred after the onset of fever', '6 atonic seizures occurred on June 8, 2022, totaling 6 seizures, and 2 atypical absence seizures occurred on June 8, 2022', 'diazepam was administered after the convulsions'. The patient has not had any further seizures. Parents reported no increase in seizure frequency (most frequent seizures occurred once every two weeks). Seizures occur occasionally when there is no fever. The patient experienced one seizure approximately 9 years ago (in 2013) without any obvious cause (parents did not witness the seizure). Present history: The patient experienced one seizure approximately 9 years ago (in 2013) without any obvious cause (parents did not witness the seizure). The patient experienced one seizure approximately 5 years ago (in 2013) without any obvious cause (parents did not witness the seizure). Present history: The patient experienced one seizure approximately 8 years ago (in 2013) without any obvious cause (parents did not witness the seizure). Parents reported no increase in seizure frequency (most frequent seizures occurred once every two weeks). There was no loss of consciousness after the seizure. 'Impaired', 'No loss of consciousness between the two seizures', 'No loss of consciousness during the two seizures', 'No convulsions (the last convulsion was on April 25, 2022)', 'No loss of consciousness or abnormal limb movement after the convulsions subsided', 'No loss of consciousness or limb movement disorder after the seizures', 'No loss of consciousness, no drooping of the corner of the mouth, nystagmus, or eyelid twitching during the above tic episodes', 'No memory of the seizures after the seizures subsided', 'No memory of the seizures after the seizures subsided', 'No limb movement disorder or fatigue after the seizures', 'No loss of consciousness between seizures', 'Seizures occurred with and without fever', 'Constant convulsions occurred for several months afterward (except for a period without seizures in November 2016)' 'No loss of consciousness during the seizure', 'The child had another seizure 5 hours later without fever', 'The child had no seizures during this period. More than a day ago, the child was suspected of missing 3 ml of "Oxcarbazepine Oral Solution" three times and then suddenly had a seizure during sleep (around 05:40 yesterday)', 'Current seizure status: no seizures for 2-3 days', 'Seizure manifestations: no seizures for 2-3 days', 'Diagnosis: "1. Epilepsy 1) Focal seizures 2) Atypical absence seizures 2. Angelman syndrome 3. Psychomotor developmental delay', 'Diagnosis: "1. Epilepsy 1) Focal seizures 2) Atypical absence seizures 2. Angelman syndrome 3. Psychomotor developmental delay', 'Due to 1. Epilepsy: Focal seizures, atypical absence seizures';

[0185] A dictionary of special phrases related to non-epileptic seizures, including the following entries:

[0186] 'Longest seizure-free period approximately 10 days without obvious seizures', 'No seizures in recent months', 'This type of seizure has been seizure-free for 2 years', 'No history of this type of seizure', 'No seizures observed in the past month', 'Family members have not noticed any obvious seizures', 'The child has not noticed any seizures', 'No seizures since discharge', 'Present medical history: Family members have not noticed any seizures in the child since the last visit', 'The child has not had any seizures since the last visit', 'Compared to the previous 4 weeks: No seizures in the last 12 weeks', 'No convulsions or altered consciousness in the child in the past six months', 'Consecutive days of seizure relief without seizures', 'No family members noticed any seizures in the child during the medication reduction period (approximately 2 months)', '1.23' The child has no seizures after oral topiramate was added to control the epileptic seizures. The child has no seizures after oral topiramate was added on January 23rd. Currently, there are no seizures or absence seizures. The child had several seizures but no further seizures. The child's seizures have subsided and have not recurred since. The child's seizures have not recurred since discharge (approximately 1 year and 11 months without seizures). The child had no seizures for over 3 months after discharge (100 days) without seizures. There was no loss of consciousness, vomiting, limb weakness, seizures, or incontinence. The child has not experienced any seizures or tremors since discharge. No seizures were observed. The child was diagnosed with "focal epilepsy" and treated with levetiracetam 0.15g twice daily without seizures. There were no seizures or startle reflexes.

[0187] Drug Events Dictionary, entries include

[0188] "Discontinue medication": "reduce or stop", "discontinue", "stop taking", "stop medication", "adjust", "reduce dosage", "abruptly stop", "reduced or stopped", "discontinued", "discontinued", "discontinued", "adjusted", "reduced dosage";

[0189] "Start taking medication": "Add", "New", "Add", "Increase", "Change", "Change", "Change to", "Change to", "Start oral administration", "Start taking", "Give";

[0190] "Missed medication": "Missed a dose";

[0191] "Discontinue medication on your own": "Discontinue medication on your own", "Reduce dosage on your own", "Discontinue use on your own", "Gradually reduce dosage on your own".

[0192] A dictionary of epileptic seizure types, including the following entries:

[0193] Non-epileptic seizures: paroxysmal dizziness, reflex non-epileptic tonic seizures, non-epileptic seizures, non-epileptic tonic seizures, psychogenic non-epileptic seizures, PNES, pnes, psychogenic.

[0194] Generalized seizures: generalized seizures, epilepsy (generalized seizures, epilepsy (generalized seizures);

[0195] Generalized tonic-clonic seizures: Epilepsy (generalized tonic-clonic seizures with eyelid myoclonus), grand mal seizures, generalized tonic-clonic seizures, generalized tonic seizures, generalized tonic seizures, tonic-clonic seizures, tonic-clonic seizures, tonic-clonic seizures.

[0196] Tonic seizures: focal tonic seizures, 5. epileptic spasms with tonic seizures, epileptic spasms with tonic seizures, tonic seizures, tonic seizures, tonic;

[0197] Clonic seizures: Epilepsy (clonic seizures, clonic attacks, clonic attacks);

[0198] Myoclonus-tonic-clonic seizures: Myoclonus-tonic-clonic, Myoclonus-tonic;

[0199] Myoclonic seizures: Myoclonic seizures, myoclonus;

[0200] Atonic seizures: myoclonus-atonicity, negative myoclonus, atonic seizures;

[0201] Focal seizures: focal epilepsy in children with central and temporal spikes, complex partial seizures, epilepsy (focal), laughing seizures, focal seizures, epilepsy (focal), laughing seizures, focal spasms, epilepsy (focal), epilepsy (focal), focal, focal seizures, epilepsy (focal), epilepsy (focal origin), 2. Paroxysmal motor-evoked choreoathetosis [paroxysmal motor neuron movement disorder] focal hypermotor seizures, epilepsy (focal) / brain injury, epilepsy (partial seizures), partial seizures), focal epilepsy, focal seizures, focal seizures, partial seizures, focal electrical seizures (right anterior head), complex seizures, focal seizures Focal seizures, focal electrical seizures (right anterior head), focal type seizures, complex partial seizures, epileptic seizures / epilepsy (frontal lobe), epilepsy (focal) occipital lobe, epilepsy (first seizure) / (focal), focal seizures, focal origin, focal epilepsy, focal tonic seizures, epileptic spasms-focal seizures, epileptic spasms with focal seizures, epileptic spasms mixed with focal seizures with speech and intellectual developmental delay, epileptic spasms mixed with focal seizures, focal seizures secondary to generalized tonic-clonic seizures, ①focal seizures ②generalized tonic-clonic seizures, focal seizures secondary to generalized seizures, epilepsy (focal + generalized), focal / generalized seizures, generalized focal seizures, focal seizures.

[0202] Epileptic spasms: Epileptic spasms, epileptic spasm seizures, epileptic spasm seizures with tonic seizures, 5. Epileptic spasm seizures with tonic seizures, epileptic spasms with focal seizures, epileptic spasms with focal seizures, epileptic spasms with mixed focal seizures, speech and intellectual developmental delay, epileptic spasms with mixed focal seizures.

[0203] Reflex seizures: reflex epileptic seizures, focal seizures.

[0204] Absence seizures: blank stare seizures, frontal lobe absence seizures, absence epilepsy, epilepsy (childhood absence seizures), adolescent absence epilepsy, childhood absence epilepsy, epilepsy (absence), epilepsy (childhood absence seizures), absence petit mal seizures, childhood absence epilepsy, absence, absence seizures.

[0205] Atypical absence seizures: atypical absence seizures, atypical absence seizures, atypical absence seizures with eyelid myoclonus, eyelid myoclonus with atypical absence seizures, atypical absence seizures with eyelid myoclonus.

[0206] Typical absence of consciousness: Typical absence of consciousness;

[0207] Myoclonic absence of eyelids: Epilepsy (generalized tonic-clonic seizures with myoclonic eyelids), myoclonic absence of eyelids, myoclonic absence of eyelids, myoclonic seizures of eyelids, myoclonic eyelids.

[0208] Myoclonic absence: Myoclonus with absence seizures, absence-like seizures with myoclonus, myoclonic absence, myoclonus with absence.

[0209] Syncope: What is the cause of syncope?

[0210] Migraine: migraine;

[0211] Breath-holding spells: Breath-holding spells;

[0212] Transient ischemic attack (TIA): Transient ischemic attack (TIA)

[0213] Sleep disorders: sleep disorders;

[0214] Tic disorder: Tic disorder.

[0215] The dictionary of epilepsy syndromes includes the following entries:

[0216] Benign familial neonatal epilepsy: familial neonatal epilepsy, familial neonatal seizures, neonatal epilepsy, BFNE;

[0217] Benign familial infantile epilepsy: familial infantile epilepsy, familial infantile seizures, BFIE;

[0218] Benign infantile epilepsy: Benign infantile epilepsy, benign infantile seizures, BIE;

[0219] Ohtahara syndrome: Ohtahara, OHTAHARA, Ohtahara syndrome;

[0220] Early myoclonic encephalopathy: Myoclonic encephalopathy, EME;

[0221] Epilepsy with migrating focal seizures in infancy: Epilepsy with migrating focal seizures in infancy, Epilepsy with migrating partial seizures in infancy, EIMFS;

[0222] Dravet syndrome: DRAVET, Severe myoclonic epilepsy in infancy, DRAVEAT syndrome;

[0223] Infantile spasms: WEST, Infantile spasms, WSET syndrome, West syndrome, Wess syndrome;

[0224] Myoclonic epilepsy in infancy: Myoclonic epilepsy in infancy;

[0225] Lennox-Gastaut syndrome: LENNOX, LGS, LGS syndrome;

[0226] Myoclonic-atonic epilepsy: Myoclonic-atonic epilepsy, DOOSE, MAE;

[0227] Childhood absence epilepsy: Childhood absence epilepsy, CAE;

[0228] Eyelid myoclonic epilepsy: Eyelid myoclonic epilepsy, JEAVONS, EME;

[0229] Myoclonic absence epilepsy: Myoclonic absence epilepsy;

[0230] Benign childhood epilepsy with centrotemporal spikes: Benign childhood epilepsy with centrotemporal spikes, BECTS, ROLANDIC, Focal epilepsy with centrotemporal spikes, Childhood focal epilepsy with centrotemporal spikes, Childhood epilepsy with centrotemporal spikes, Focal epilepsy in children with centrotemporal spikes, Epilepsy with centrotemporal spikes in children, Focal epilepsy with centrotemporal spikes;

[0231] Panayiotopoulos syndrome: Early-onset benign occipital epilepsy in children, PANAYIOTOPOULOS;

[0232] Late-onset childhood occipital epilepsy (Gastaut type): Late-onset childhood occipital epilepsy, GASTAUT, Late-onset occipital epilepsy;

[0233] Landau-Kleffner syndrome: LANDAU, KLEFFNER, Acquired epileptic aphasia, LKS, LKS syndrome;

[0234] Epilepsy encephalopathy with persistent spike-and-wave complexes during slow-wave sleep: Epilepsy encephalopathy with persistent spike-and-wave complexes during slow-wave sleep, CSWS;

[0235] Juvenile absence epilepsy: Juvenile absence epilepsy, JAE;

[0236] Juvenile myoclonic epilepsy: Juvenile myoclonic epilepsy, JME, Juvenile myoclonic epilepsy;

[0237] Epilepsy with only generalized tonic-clonic seizures: only generalized tonic-clonic seizures;

[0238] Hereditary epilepsy with febrile seizures, GEFS+, febrile seizures, GEFS+;

[0239] Progressive myoclonic epilepsy: Progressive myoclonic epilepsy, PME;

[0240] Rasmussen syndrome: Rasmussen;

[0241] Febrile Infection-Related Epilepsy Syndrome: Febrile infection-related epilepsy, fever-induced, fulminant inflammation, FIRES.

[0242] The dictionary of epilepsy comorbidities includes the following entries:

[0243] Neurodevelopmental delay: 'developmental delay', 'developmental retardation', 'developmental retardation', 'motor developmental delay', 'developmental disorder', 'intellectual disability', 'motor dysfunction', 'cognitive dysfunction', 'speech dysfunction', 'learning disability', 'emotional disorder', 'social dysfunction', 'brain dysplasia', 'cortical dysplasia', 'abnormal brain development', 'cortical malformation', 'motor disorder', 'cognitive impairment', 'speech disorder', 'social disorder', 'cortical dysplasia', 'corpus callosum dysplasia', 'temporal lobe dysplasia', 'language disorder', 'communication disorder', 'communication impairment' Attention deficit hyperactivity disorder (ADHD), brain dysplasia, developmental coordination disorder, psychomotor developmental disorder, hemispheric dysplasia, myelin dysplasia, communication disorder, communication problems, developmental abnormality, intellectual disability, brain malformation, speech disorder, dysarthria, intellectual disability, attention deficit disorder, below-average growth and development, cortical dysplasia, below-average intellectual development, ventricular malformation, Poirier-Bienvenu neurodevelopmental syndrome, gyri malformation;

[0244] Cerebral palsy: 'cerebral palsy', 'cerebral palsy';

[0245] Autism Spectrum Disorder: 'Autism Spectrum Disorder', 'Autism Spectrum Disorder', 'Autism', 'ASD';

[0246] Febrile seizures: febrile seizures, high fever seizures, high temperature seizures, neonatal seizures.

[0247] A dictionary of EEG, CT, and MRI examinations, including the following entries:

[0248] CT scan items: 'Cranial CT plain scan': 'Cranial CT plain scan', 'Cranial CT spiral scan': 'Cranial CT plain scan', 'Cranial CT plain scan + 3D reconstruction': 'Cranial CT plain scan', 'Cranial CT axial scan': 'Cranial CT plain scan', 'Cranial CT plain scan + enhancement + 3D reconstruction': 'Cranial CT enhancement', 'Cranial CT plain scan + enhancement': 'Cranial CT enhancement', 'Cranial CTA': 'Cranial CTA';

[0249] MRI examination items: 'Cranial MR plain scan 3.0T':'Cranial MR plain scan','Cranial MR plain scan':'Cranial MR plain scan','Cranial MR enhanced 3.0T':'Cranial MR enhanced','Cranial MR plain scan + enhanced':'Cranial MR enhanced','Cranial MR plain scan + enhanced 3.0T':'Cranial MR enhanced','Cranial MRI brain functional imaging (DTI) (1.5T)':'Cranial DTI','Cranial MR + vein imaging (MRV)':'Cranial MRV','Cranial MRV (3.0T)':'Cranial MRV','MRI brain functional imaging (DTI) (3.0T)':'Cranial DTI','Cranial MR+MRA (3.0T)':'Cranial MRA','Cranial MR plain scan 3.0T + DWI':'Cranial DWI','Cranial MRI brain 'Perfusion 3.0T':'Cranial MR Perfusion','Cranial MR Plain Scan + Enhanced Angiography':'Cranial MRA','MRA Cranial MRA (3.0T)':'Cranial MRA','Cranial MR Enhanced Angiography':'Cranial MRA','Cranial Plain Scan 3.0T; MRI Brain Functional Imaging (DTI) (3.0T)':'Cranial DTI','MRI Brain Functional Imaging (MRS) (3.0T)':'Cranial MRS','Cranial MR Enhanced':'Cranial MR Enhanced','Brain Functional Imaging':'Cranial MRS','Cranial MR Plain Scan + DWI':'Cranial DWI','Cranial MR Plain Scan + Enhanced + DWI':'Cranial DWI','Cranial MRI Brain Perfusion':'Cranial MR Perfusion','Cranial MR Plain Scan 3.0T + Enhanced + DWI':'Cranial DWI';

[0250] EEG examination items: '15-hour bedside video EEG (including brain topography + special evoked responses)': 'Video EEG', 'Wake-up + Sleep + Video EEG': 'Video EEG', 'Routine Wake-up EEG': 'Wake-up EEG', 'Brain Topography': 'Brain Topography', 'Brain Topography (including 2D brain topography (at least 16 channels))': 'Brain Topography', 'Long-term Video EEG': 'Video EEG', '3-hour bedside video EEG (including brain topography + special evoked responses)': 'Video EEG', '3-hour Video EEG (including brain topography + special evoked responses): 'Video EEG', 'Awake EEG (including special evoked responses)': 'Awake EEG + Evoked Response', 'Awake + Sleep EEG': 'Awake / Sleep EEG', '12-Hour Video EEG (including brain topography + special evoked responses)': 'Video EEG', 'Awake EEG': 'Awake EEG', 'Awake + Sleep EEG (including special evoked responses)': 'Awake / Sleep EEG + Evoked Response', '24-Hour Ambulatory EEG': 'Ambulatory EEG', '24-Hour Ambulatory EEG' Electroencephalogram (including brain topography + special evoked responses): 'Dynamic EEG + Evoked Response', '14-hour Video EEG (including brain topography + special evoked responses)': 'Video EEG', 'Special Evoked EEG': 'Awake EEG + Evoked Response', 'Routine Awake + Video EEG': 'Video EEG', '7-hour Bedside Video EEG (including brain topography): 'Video EEG', 'EEG Video Monitoring': 'Video EEG', 'Bedside EEG Detection': 'EEG', 'Special Electrode EEG': 'EEG', 'EEG Video Recording' Monitoring: Video EEG, Bedside EEG Monitoring: EEG, 3-hour Bedside Video EEG (including brain topography + special evoked responses): Video EEG, 12-hour Video EEG (including brain topography + special evoked responses) with diffuse alpha and theta frequency band rhythmic activity mixed with delta and beta activity in each channel, anterior head: Video EEG, Brain Topography (including 2D brain topography, at least 16 channels): Brain Topography, 3-hour Video EEG (including brain topography and special evoked responses): Video EEG

[0251] The EEG waveform dictionary includes the following entries:

[0252] Slow wave: slow wave;

[0253] Spike wave: a slow spike wave;

[0254] Spike-slow-wave complex: spike-slow-wave complex, spike (slow) wave, spike-slow-wave complex;

[0255] Spike wave: spike wave, spike (slow) wave, spike (slow) wave, spike (slow) wave, spike (slow) wave, spike (slow) wave, spike (slow) wave, spike (slow) emission;

[0256] Spike-slow wave: spike-slow wave, spike (slow) wave, spike (slow) wave, spike (slow) emission;

[0257] Spike wave: spike wave, spike tip (slow) wave, spike (spin) wave, spike (slow) wave, spike (spin) (slow) wave, spike (spin) (slow) emission;

[0258] Spike-slow wave: Spike-slow wave, spike (slow) wave, spike (slow) wave, spike (slow) wave, spike (slow) wave, spike (slow) emission;

[0259] fast wave: fast wave;

[0260] Irregular theta waves: theta waves, mixed activity in the theta and delta frequency bands, theta activity;

[0261] Irregular delta waves: delta waves, delta activity, mixed activity in the theta and delta frequency bands;

[0262] Lazy wave phenomenon: lazy wave phenomenon;

[0263] Inhibitory brain waves;

[0264] High-level rhythm disorder: high-level rhythm disorder, peak rhythm disorder, high-amplitude rhythm disorder, high-level rhythm disorder.

[0265] EEG background dictionary, entries include:

[0266] Poor background: disordered background, poor background, bad background, poor background structure, poor background, abnormal changes in background EEG.

[0267] Background slows down: background slows down, background activity slows down, background is slightly slow, background is slow, background activity is slightly slow, background is slightly slow, background activity is slightly slow, background rhythm is slow, background activity slows down.

[0268] The dictionary of brain lobe regions includes the following entries:

[0269] Frontal lobe: frontal lobe, frontal region, anterior head, frontal area, middle frontal region, frontal pole;

[0270] Top leaf: top leaf, posterior top leaf, upper top leaf, top, apex area;

[0271] Temporal lobe: temporal lobe, temporal region, temporal parietal, posterior temporal region, middle temporal region, anterior temporal region, temporal pole;

[0272] Occipital lobe: occipital lobe, occipital region, back of the head;

[0273] Temporoparietal junction: temporoparietal junction, temporoparietal lobe, parietotemporal region, temporoparietal region;

[0274] Temporoparietal junction | Insula: Temporoparietal insula;

[0275] Frontotemporal junction: frontotemporal region;

[0276] Frontoparietal junction: frontoparietal lobe, frontoparietal region, center, midline;

[0277] Parietal-occipital junction: Parietal-occipital lobes;

[0278] Seahorse: Seahorse;

[0279] Island leaf: island leaf, island cap;

[0280] pituitary gland;

[0281] Pineal gland: Pineal gland;

[0282] Temporal lobe | Parietal lobe | Insula: temporal, parietal, and insula lobes;

[0283] Frontal lobe | Parietal lobe | Temporal lobe: frontal, parietal, and temporal regions.

[0284] Example 2

[0285] To verify the feasibility of the method proposed in Example 1, this example applies the method proposed in Example 1 to the analysis of electronic medical records in a hospital in Shenzhen. The electronic medical records span from November 1, 2013 to March 31, 2023. The criteria for inclusion of patients are children who come to see a doctor for epilepsy, including outpatients, emergency rooms and inpatients, totaling 40,931 children.

[0286] In the pharmacy medication record management process, a total of 7,610,433 pharmacy medication records were exported from the electronic medical record database, corresponding to 737,430 outpatient, emergency, or inpatient records. After data management, it was found that 121 records were missing names, and 20,463 records were missing the doctor's prescription date; 584,005 records involved the use of anti-epileptic drugs, accounting for 7.67%, and 132,517 prescription records did not have corresponding electronic medical record texts. In the quality control phase of data management, the quality control of medication entries specifically included:

[0287] Twenty-four records containing the same medical order that identified multiple drug names were deleted; these records typically did not involve drug prescriptions.

[0288] Delete 181 records related to injection rate adjustment or drug dosage adjustment;

[0289] Delete 2,126 records with special medication frequencies, including the three medication frequency keywords prn (long-term prescription, to be used when needed), SOS (to be used when needed, limited to one use, effective within 12 hours), and when necessary, as well as the frequency keywords sq and sqbxs that are suspected of being misspelled;

[0290] Delete 50 records of hospitalized patients with special administration routes, including the keywords "alternative", "retention enema" (all records of diazepam injection), "other", and "as prescribed";

[0291] Quality control was performed on entries in outpatient medication records that lacked medication frequency. A total of 17 records were missing medication frequency, of which 15 records were successfully filled in, and 2 records that could not be filled in were deleted.

[0292] Quality control was performed on the prescription dates in medication records. 1,011 prescription records were missing prescription dates. The missing dates were filled in based on the visit dates in the visit information. The prescription dates of all 1,011 records were successfully filled in. The prescription dates of the prescription records related to discharge medication were also changed to the discharge date. As a result, the prescription dates of 21,350 medication records were corrected.

[0293] Quality control was conducted on medication records for entries lacking single doses. A total of 785 non-ketogenic diet (ketogenic diets do not require dosage) records were found to be missing or had non-standard single doses. This included 10 outpatient prescription records without single doses, inpatient medication records with special dosages such as interval adjustments, orders executed X times, required medications, medications taken to the operating room, dosages of XX tablets or pills, and records with only numbers without dosage units, as well as inpatient records lacking discharge medication records. By referencing other information, the single doses of 12 medications were corrected, and the remaining 773 records were deleted.

[0294] Dosage corrections were made for hospitalized patients taking medication multiple times a day, with a total of 174 single doses and 315 frequency of medication records being corrected.

[0295] Quality control was performed on the medication frequency in discharge medication records. A total of 30,002 discharge medication records were identified, of which 1,762 records lacked medication frequency information. Through quality control, the single-dose information in 128 records and the medication frequency in 1,614 records were corrected. 148 records that still lacked medication frequency information after quality control were deleted, along with 10 records with a medication frequency of prn (if necessary).

[0296] Delete one record that contains only the drug name and no other information.

[0297] After quality control, a total of 580,690 medication records remained, including 481,309 outpatient and emergency room prescriptions (82.89%) and 99,381 inpatient medication records (17.11%), covering 17 types of anti-epileptic drugs, corticotropin, and the ketogenic diet—three types of epilepsy treatments. The relevant records are listed below in order of the number of records from most to least:

[0298] • Valproic acid: 140,943 (24.27%)

[0299] •Levetiracetam: 119,616 (20.60%)

[0300] • Oxcarbazepine: 115,003 (19.81%)

[0301] • Topiramate: 63,017 (10.85%)

[0302] Lamotrigine: 48,262 (8.31%)

[0303] • Nitrazepam: 24,546 (4.23%)

[0304] • Clonazepam: 21,584 (3.72%)

[0305] Phenobarbital: 15,485 (2.67%)

[0306] • Diazepam: 11,339 (1.95%)

[0307] • Carbamazepine: 10,183 (1.75%)

[0308] Lacosamide: 2,516 (0.43%)

[0309] • Ketogenic diet: 2,339 (0.40%)

[0310] • Zonisamide: 2,210 (0.38%)

[0311] • Adrenocorticotropic hormone: 1,397 (0.24%)

[0312] • Clobar: 730 (0.13%)

[0313] • Gabapentin: 658 (0.11%)

[0314] • Vigabatrin: 449 (0.08%)

[0315] • Phenytoin: 400 (0.07%)

[0316] • Lorazepam: 13 (0.00%)

[0317] This shows that the commonly used anti-epileptic drugs for children in this hospital are valproic acid, levetiracetam, oxcarbazepine, topiramate, and lamotrigine. These five drugs account for 83.84% of all medical orders recorded.

[0318] After standardizing the administration methods of anti-epileptic drugs, the number of medical order records corresponding to each administration method is as follows:

[0319] • Oral tablets: 405,808 (69.88%)

[0320] • Oral solution: 150,738 (25.96%)

[0321] • Oral capsules: 12,133 (2.09%)

[0322] • Injection: 9,672 (1.67%)

[0323] • Diet: 2,339 (0.40%)

[0324] Oral tablet medications were the most frequently prescribed, accounting for 97.93% of all prescriptions. The top three frequency of medication use were twice daily (495,967 records, 85.39%), once daily (64,502 records, 11.11%), and three times daily (17,582 records, 3.03%), totaling 99.53%. Additionally, 558,831 (96.24%) of the prescriptions were for long-term medication.

[0325] In the management of outpatient and emergency medical records, a total of 593,636 records were retrieved from the electronic medical records, involving 517,533 outpatient and emergency visits. The record time for 29,306 outpatient and emergency visit entries was inconsistent with the visit time, indicating that the medical records were modified after the visit, resulting in the record time being updated. For each outpatient and emergency visit, only the latest record was retained, leaving 566,201 records, involving 517,533 outpatient and emergency visits, accounting for 95.38%.

[0326] Name, outpatient number, gender, age, height, and weight measurements were extracted from the full text of electronic medical records, including 268,028 name records, 260,993 age records, 267,658 gender records, 255,758 outpatient number records, 15,131 height records, and 34,845 weight records.

[0327] Quality control removed entries in the medical record text where the names and outpatient numbers did not match those in the electronic medical record, leaving 554,156 records. During the management of the medical record text, the process identified epileptic seizures, anti-epileptic drug treatments, and corresponding dates, identifying 75,524 epileptic seizure events and 253,289 seizure-free events, as well as 4,356 medication initiation events, 1,352 medication discontinuation events, 705 missed medication doses, and 1,051 instances of self-discontinuation of medication.

[0328] In the process of managing inpatient medical records, the latest entries of inpatient records are retained after data is read from the electronic inpatient medical records, totaling 40,846 records involving 25,918 inpatient records. Since inpatient records are stored separately as admission records, progress notes, and discharge records, all three types of records are processed during data management and integrated according to the inpatient serial number.

[0329] Names, genders, and ages were extracted from the full text of electronic medical records, including admission records, progress notes, and discharge records. This included 10,876 name records, 10,882 age records, and 10,885 gender records. The names identified from the admission records, progress notes, and discharge records were integrated, records with inconsistent names were identified, and the correct names were generated after data processing.

[0330] Names containing words such as "son," "daughter," "old baby," "young boy," and "BB" are considered informal names. The correct name after treatment is the formal name identified in the records. 70 patients had inconsistent names in their medical records, the reason being that the patients changed their names. No one-to-many mapping between names and hospital registration numbers was found, i.e., one name corresponding to one hospital registration number in the hospitalization record.

[0331] The height and weight information in the hospitalization record has two data sources: height and weight measurements identified from the hospitalization medical record text and height and weight records extracted from the patient's vital signs records during hospitalization. The height and weight records from the two sources are then integrated according to the hospitalization serial number. The integrated height and weight records are saved in the format of "measurement date: value".

[0332] 68,851 height / weight records were identified from vital sign records, involving 22,994 hospitalization records. Specifically, 22,982 hospitalization records had weight records, and 16,951 hospitalization records had height records. After adding the height / weight values ​​identified from the medical record text, 26,762 hospitalization records had weight records, and 17,201 hospitalization records had height records.

[0333] During the management of inpatient medical records, epileptic seizures, anti-epileptic drug treatments, and corresponding dates were identified. A total of 212,283 epileptic seizure events and 62,418 seizure-free events were identified, along with 114,782 drug initiation events, 24,837 drug discontinuation events, 625 missed drug doses, and 4,910 self-discontinuation events.

[0334] Among the 40,931 children, most had incomplete medical or medication information or lacked long-term follow-up records. To better illustrate the effectiveness of data governance in this embodiment, a cohort of children newly diagnosed with epilepsy and receiving antiepileptic drug monotherapy for the first time was defined, including 8,853 children. The data governance of diagnosis, EEG and imaging (CT / MRI) results was then described based on this cohort.

[0335] In the process of managing diagnostic data, 6,955 diagnostic records were obtained from 8,235 records corresponding to 7,430 hospitalizations of 2,510 patients, and 124,521 diagnostic records were obtained from 170,399 records corresponding to 158,398 outpatient and emergency visits of 7,951 patients. After merging these records, 4,046 diagnoses related to neurodevelopmental delay in 1,332 patients, 481 diagnoses related to cerebral palsy in 141 patients, and 271 diagnoses related to autism spectrum disorder in 133 patients were identified. 405 patients had a history of febrile seizures.

[0336] Of the 8,853 children, 7,426 had a confirmed diagnosis of epilepsy. Further, 4,613 seizure type records were identified from 2,974 patients, including 514 records of generalized seizures, 2,323 records of focal seizures, 167 records of epileptic spasms, 2 records of reflex seizures, and 486 records of non-epileptic seizures. 316 patients had status epilepticus. 619 epilepsy syndrome diagnoses were identified, involving 593 patients. The top five syndromes with the most patients were hereditary epilepsy with febrile seizures (151), infantile spasms (129), benign childhood epilepsy with centrotemporal spikes (117), childhood absence epilepsy (72), and Dravet syndrome (50).

[0337] In the process of managing the imaging data, a total of 150,210 EEG and imaging examination records were identified in the electronic medical records. Entries from 8,853 included pediatric patients who underwent CT / MR / EEG examinations were retained, leaving 18,199 entries, including 7,837 EEG records, 3,889 MRI records, and 2,040 examination results.

[0338] After removing duplicate entries and performing quality control by examination date, the remaining data consisted of 2,482 EEG records from 1,599 patients, 2,189 MRI records from 2,029 patients, and 1,002 CT records from 994 patients. Due to incomplete EEG and image recording in electronic medical records, the missing results rate was very high. After combining CT and MRI results, 5,344 patients (68%) had missing results, 1,426 patients (18%) had normal results, and 1,072 patients (14%) had abnormal results. For EEG results, 6,204 patients (79%) had missing results. Among the remaining patients with results, 178 patients (2%) had normal results, 196 patients (2%) had borderline results, and 1,264 patients (16%) had abnormal results. 509 patients (6%) had poor backgrounds, and 80 patients (1%) had slowed backgrounds. 1,315 patients (17%) had epilepsy-related abnormal waveforms (epilepsy-like waveforms), and 737 patients (9%) had epilepsy-unrelated abnormal waveforms.

[0339] Based on the above data, this embodiment defines a visualization function to more comprehensively reflect the patient's epileptic seizures and medication information, such as the timeline of epileptic seizures and antiepileptic drug events for patient P06046, as shown below. Figure 6 As shown, the patient began experiencing epileptic seizures at the end of 2015. After two seizures, the patient did not seek medical attention. On the day of the third seizure in 2016, the patient visited an outpatient clinic and began taking levetiracetam for anti-epileptic treatment. The medical record also identified the initiation of levetiracetam. Approximately six months later, the patient experienced another seizure, and the treatment plan was adjusted to a combination of levetiracetam and valproic acid. The medical record again identified the reduction of levetiracetam and the initiation of valproic acid, consistent with the doctor's orders.

[0340] During subsequent follow-ups, the patient reported no seizures until early 2019 when a missed dose of medication triggered a seizure. In mid-2020, the patient experienced another seizure and was immediately hospitalized. Treatment involved levetiracetam and valproic acid. The patient remained seizure-free until 2022 when the family discontinued the medication, resulting in two more seizures after that.

[0341] This embodiment also defines a visualization function to plot the timeline of anti-epileptic drug dosages. For example, for patient P06046, the initial treatment drug used was levetiracetam. The average daily dose and weight-normalized daily dose for each prescription are shown below. Figure 7As shown, the patient's initial prescription dose reached the maintenance dose of levetiracetam, which is also reasonable in clinical practice. Subsequently, the average daily dose of the drug prescription increased from 500 mg / day to 1000 mg / day and was maintained for a period of time. It can be seen that although the dose of each prescription was 1000 mg / day, the dose after weight normalization showed a monotonically decreasing trend because the patient's weight increases with age.

[0342] To further evaluate the accuracy of this invention in extracting medication events, epileptic seizure states, and their corresponding dates, the applicant randomly selected 20 inpatient medical records and 50 outpatient / emergency medical records. These samples included longer medical records written by different doctors within the department, as well as medical records written strictly according to the department's standard template. Subsequently, the output results after data processing according to this invention were compared with the gold standard of manual annotation completed jointly by data engineers and clinicians.

[0343] Accuracy assessment metrics include date errors and missed detections. Date errors include correctly identifying an event (drug name or seizure state) but retrieving the wrong date. Missed detections include events that exist in the medical record but were not identified during data governance.

[0344] Please refer to Table 1. In all categories of medication events and seizure states, the number of missed detections (false negatives) was greater than or equal to date errors. The overall accuracy of data management was as follows across different categories: 100% for missed doses and self-discontinuation of medication, 91.7% for initiation of anti-seizure medication, 94.6% for dose reduction / discontinuation, 93.8% for seizure reports, and 97.4% for no seizure reports, demonstrating high overall data management accuracy.

[0345] Table 1. Accuracy Assessment Table for Data Governance

[0346]

[0347] Example 3

[0348] Please see Figure 8 This embodiment proposes an electronic medical record data management device for childhood epilepsy, used to implement the electronic medical record data management method for childhood epilepsy as proposed in Embodiment 1. The device includes:

[0349] The acquisition module 10 acquires electronic medical record data from multiple heterogeneous data sources and constructs multiple keyword dictionaries for identifying and extracting information; the data sources include outpatient and emergency medical records storing unstructured data, inpatient medical records storing unstructured data, and / or pharmacy records storing semi-structured data.

[0350] Processing module 20 is used to obtain medical record data, diagnostic data, electroencephalogram data and imaging data from unstructured data in outpatient and emergency medical records, inpatient medical records, diagnostic records, electroencephalogram records and / or imaging data based on keyword dictionary, and to obtain medication data from semi-structured data in pharmacy records based on keyword dictionary;

[0351] Analysis module 30 is used to generate structured data for clinical analysis of epilepsy based on medical record data, medication data, diagnostic data, electroencephalogram data and imaging data, and to display the patient's epilepsy-related data in the form of visualizations based on the structured data.

[0352] Specifically, the processing module 20 includes:

[0353] The deduplication unit 21 is used to classify data entries with the same specified information in the same emergency room medical record or the same inpatient medical record into the same duplicate entry group, retain the data entry with the latest record time information in each duplicate entry group, and delete the remaining data entries.

[0354] The first identification unit 22 is used to identify the inherent information in each data entry based on a custom function of regular expressions, confirm whether there are any abnormalities in the inherent information of each data entry, and delete the abnormal data entry when there are abnormalities, or supplement and / or correct the inherent information in the abnormal data entry based on data information from other data sources.

[0355] Also includes:

[0356] The second identification unit 23 is used to identify the drug information of each medication entry in the pharmacy record. It uses a string comparison method based on regular expressions to compare the medication information in each medication entry with the medication information keywords in the keyword dictionary. Through the comparison results, it confirms and standardizes the drug name information, deletes medication entries whose identification results meet the first preset condition, and supplements and / or corrects medication entries whose identification results meet the second preset condition.

[0357] The comparison unit 24 is used to compare the medication information in each medication entry with the medication information keywords in the keyword dictionary in turn using a string comparison method based on regular expressions. The comparison results are used to confirm whether there are any abnormalities in the medication information in each medication entry. If there are any abnormalities, the medication entry with abnormalities is deleted, or the medication information in the medication entry with abnormalities is supplemented based on data information from other data sources.

[0358] Standardization unit 25 is used to standardize the administration method, administration frequency, and dosage unit based on each medication item, and to calculate the average daily dose and prescription duration for each medication item. It also distinguishes between emergency medication and long-term medication, and further distinguishes between long-term oral medication and injectable medication used in emergency situations to control epileptic seizures based on the administration method, thereby generating corresponding medication data.

[0359] Example 4

[0360] The present invention also proposes a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the electronic medical record data management method for childhood epilepsy as proposed in Example 1.

[0361] It should be noted that computer-readable media, including both permanent and non-permanent, removable and non-removable media, can store information using any method or technology. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, disk storage, quantum memory, graphene-based storage media or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0362] In summary, this invention proposes a method and apparatus for managing electronic medical record data in children with epilepsy. The proposed solution transforms scattered, chaotic, and difficult-to-use raw electronic medical record data into centralized, organized, and directly applicable structured data for clinical analysis and decision-making in children with epilepsy. This solves the pain point of difficulty in integrating and transforming medical data, improves the intuitiveness and efficiency of epilepsy diagnosis and treatment through visualization, and provides strong data support. At the same time, it can provide analyzable and high-quality data for scientific research.

[0363] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for managing electronic medical record data in children with epilepsy, characterized in that, The method includes: Electronic medical record data from multiple heterogeneous data sources are acquired, and multiple epilepsy-related keyword dictionaries are constructed for identification and / or information extraction. The data sources include outpatient and emergency medical records storing unstructured data, inpatient medical records storing unstructured data, and / or pharmacy records storing semi-structured data. Based on the keyword dictionary, unstructured data in the outpatient and emergency medical records and / or the inpatient medical records are processed to obtain medical record data, and semi-structured data in the pharmacy records are processed based on the keyword dictionary to obtain medication data; Based on the medical record data and medication data, structured data for clinical analysis of epilepsy is generated, and based on the structured data, the patient's epilepsy-related data is displayed in the form of a visualization. The pharmacy records include several medication entries. The average daily dose for each medication entry is calculated based on the normalized drug dose after adjusting for the patient's weight on the query date corresponding to each medication entry. The method for determining the patient's weight on the query date includes: If the outpatient or inpatient medical records contain the patient's weight information, determine the start and end time periods and each record date corresponding to the weight information, filter the weight information, fit a locally weighted regression model with the patient's age as the independent variable and the filtered weight information as the dependent variable, identify and delete abnormal weight values ​​in the weight information by calculating the residuals of the locally weighted regression model, and generate a fitted curve based on the weight values ​​retained in the weight information. If the query date is included in the recording time period and there is a corresponding recording date for the query date, the patient's weight value on the query date is obtained from the fitted curve; if the query date is included in the recording time period but there is no corresponding recording date for the query date, linear interpolation is performed through the fitted curve to obtain the patient's weight value on the query date. If the query date is not included in the recorded time period or the patient's outpatient and inpatient medical records do not contain the patient's weight information, and the patient's age is not greater than a preset age value, the patient's weight value on the query date is determined based on the weight value obtained by fitting the weight data of all children with epilepsy of the same sex as the patient. If the query date is not included in the recorded time period or the patient's outpatient or inpatient medical records do not contain the patient's weight information, and the patient's age is greater than a preset age value, the patient's weight on the query date is determined based on the weight reference value for healthy children.

2. The method for managing electronic medical record data for childhood epilepsy according to claim 1, characterized in that, The outpatient and emergency medical records and / or the inpatient medical records include several data entries, each data entry corresponding to a record time information. The method for managing unstructured data in the outpatient and emergency medical records and / or the inpatient medical records includes: Data entries with the same specified information in the same outpatient or inpatient medical record are grouped into the same duplicate entry group. The data entry with the latest recording time information in each duplicate entry group is retained, and the remaining data entries are deleted. The system uses a custom function based on regular expressions to identify the inherent information in each data entry, confirms whether there are any anomalies in the inherent information of each data entry, and deletes the data entry with the anomaly when anomalies are found, or supplements and / or corrects the inherent information in the data entry with the anomaly based on data information from other data sources.

3. The method for managing electronic medical record data for childhood epilepsy according to claim 2, characterized in that, Each of the aforementioned outpatient / emergency medical records or inpatient medical records corresponds to identification information, which includes name, outpatient number, inpatient number, gender, age, height, and weight. A method for confirming whether any of the inherent information in any of the data entries is abnormal includes: Confirm whether there are any missing inherent information in the data entry and the corresponding record time information of the data entry; Confirm whether the name, outpatient number, and inpatient number in the data entry are consistent with the identification information corresponding to the outpatient or inpatient medical record from which the data entry originates; Confirm whether the name, outpatient number, and inpatient number recorded in the data entry are consistent with other data entries contained in the outpatient or inpatient medical record that is the source of the data entry; And / or, confirm whether the name, outpatient number, and inpatient number recorded in the data entry are consistent with the data information from other data sources.

4. The method for managing electronic medical record data for childhood epilepsy according to claim 2 or 3, characterized in that, Based on the keyword dictionary, unstructured data in the outpatient and / or inpatient medical records is processed to obtain medical record data, including: The free text in each data entry is segmented into short sentences according to punctuation marks. A string matching method based on regular expressions is used to sequentially identify whether each short sentence contains an event description based on the keyword dictionary. Specified data is extracted from the short sentences containing the event description, the terminology of all events is standardized, and the data is summarized to generate the medical record data. The event description includes descriptions of epileptic seizures, epileptic seizure states without seizure descriptions, and drug treatment-related events such as the initiation of anti-epileptic drug use, drug reduction, drug discontinuation, missed doses, or self-discontinuation of drug use.

5. The method for managing electronic medical record data for childhood epilepsy according to claim 1, characterized in that, Based on the keyword dictionary, semi-structured data in the pharmacy records is processed to obtain medication data, including: The drug information of each medication entry in the pharmacy record is identified. The medication information in each medication entry and the medication information keywords in the keyword dictionary are compared sequentially using a string comparison method based on regular expressions. The drug name information is confirmed and standardized through the comparison results. Medication entries that meet the first preset condition are deleted, and medication entries that meet the second preset condition are supplemented and / or corrected. The medication information in each medication entry is compared with the medication information keywords in the keyword dictionary using a string comparison method based on regular expressions. The comparison results are used to determine whether there are any abnormalities in the medication information in each medication entry. If there are any abnormalities, the medication entry with abnormalities is deleted, or the medication information in the medication entry with abnormalities is supplemented based on data information from other data sources. Based on each of the aforementioned medication entries, the administration method, frequency of administration, and dosage unit are standardized, and the average daily dose and prescription duration for each medication entry are calculated. At the same time, emergency medication and long-term medication are distinguished, and based on the administration method, oral medications and injectable medications used in emergency situations to control epileptic seizures are also distinguished, thereby generating the corresponding medication data.

6. The method for managing electronic medical record data for childhood epilepsy according to claim 1, characterized in that, The data source also includes medical record diagnoses, electroencephalogram (EEG) examinations, and imaging examinations. The medical record diagnoses include several diagnostic data entries. The EEG examinations include several EEG examination data in the form of routine EEG, evoked EEG, and / or video EEG. The imaging examinations include several brain CT and brain MRI imaging data. The method further includes: Diagnostic data containing the keyword "epilepsy" is found by string matching based on regular expressions, and the epileptic seizure type, accompanying non-epileptic seizure, status epilepticus, epilepsy syndrome and epilepsy comorbidity corresponding to each diagnostic data containing the keyword "epilepsy" are identified based on the keyword dictionary to obtain diagnostic data. The EEG data are cleaned, and the examination results, background, waveforms, and lobes are identified. The EEG results are classified as normal, abnormal, and borderline. The background is classified as normal, slowed, and poor. The waveforms are classified as epilepsy-related or epilepsy-unrelated. The lobes are classified as frontal, temporal, parietal, occipital, frontotemporal junction, temporoparietal junction, frontoparietal junction, parieto-occipital junction, pituitary, hippocampus, insula, and pineal gland. EEG data are obtained based on the identification and classification results. Clean the image data, identify the examination results and brain lobe regions in the cleaned image data to distinguish the image data as normal or abnormal images, identify the brain lobe regions where abnormal phenomena occur, and obtain image data based on the identification and distinction results. The structured data is generated based on the diagnostic data, the electroencephalogram data, the imaging data, the medical record data, and the medication data.

7. The method for managing electronic medical record data for childhood epilepsy according to claim 1 or 6, characterized in that, Based on the structured data, the patient's epilepsy-related data are displayed in the form of visualizations, including: Based on the structured data, a medication timeline for the patient is constructed in a visualization. The start date and duration of prescriptions for various anti-epileptic drugs corresponding to the patient are marked in the visualization. The medication events and the dates of seizures and seizure-free periods for the patient in the outpatient and inpatient medical records are also marked in the visualization. The medication timelines for emergency and long-term medications are drawn with different colors, and the outpatient and inpatient prescriptions are marked with different symbols.

8. An electronic medical record data management device for childhood epilepsy, characterized in that, The device includes: The acquisition module acquires electronic medical record data from multiple heterogeneous data sources and constructs multiple keyword dictionaries for identifying and extracting information; the data sources include outpatient and emergency medical records storing unstructured data, inpatient medical records storing unstructured data, and / or pharmacy records storing semi-structured data. The processing module is used to process unstructured data in the outpatient and emergency medical records and / or the inpatient medical records based on the keyword dictionary to obtain medical record data, and to process semi-structured data in the pharmacy records based on the keyword dictionary to obtain medication data. The analysis module is used to generate structured data for clinical analysis of epilepsy based on the medical record data and the medication data, and to display the patient's epilepsy-related data in the form of a visualization based on the structured data. The pharmacy records include several medication entries. The average daily dose for each medication entry is calculated based on the normalized drug dose after adjusting for the patient's weight on the query date corresponding to each medication entry. The method for determining the patient's weight on the query date includes: If the outpatient or inpatient medical records contain the patient's weight information, determine the start and end time periods and each record date corresponding to the weight information, filter the weight information, fit a locally weighted regression model with the patient's age as the independent variable and the filtered weight information as the dependent variable, identify and delete abnormal weight values ​​in the weight information by calculating the residuals of the locally weighted regression model, and generate a fitted curve based on the weight values ​​retained in the weight information. If the query date is included in the recording time period and there is a corresponding recording date for the query date, the patient's weight value on the query date is obtained from the fitted curve; if the query date is included in the recording time period but there is no corresponding recording date for the query date, linear interpolation is performed through the fitted curve to obtain the patient's weight value on the query date. If the query date is not included in the recorded time period or the patient's outpatient and inpatient medical records do not contain the patient's weight information, and the patient's age is not greater than a preset age value, the patient's weight value on the query date is determined based on the weight value obtained by fitting the weight data of all children with epilepsy of the same sex as the patient. If the query date is not included in the recorded time period or the patient's outpatient or inpatient medical records do not contain the patient's weight information, and the patient's age is greater than a preset age value, the patient's weight on the query date is determined based on the weight reference value for healthy children.

9. A computer-readable storage medium, characterized in that, It stores a computer program, which, when executed by a processor, implements the electronic medical record data management method for childhood epilepsy as described in any one of claims 1-7.

Images