KR101883433B1 - System and method for tailored intervetion through prediction of inpatient falls - Google Patents

Abstract

The present invention provides a customized intervention guidance system that enables evaluation of the risk of falling of an inpatient patient based on clinical data of the inpatient and notifies the result to the medical staff, And a method for providing a customized arbitration guidance service using the system. By using the system of the present invention, it is possible to accurately and effectively predict the likelihood of a fall of an inpatient, and to utilize the arbitration guidance according to an individual patient condition.

Description

FIELD OF THE INVENTION [0001] The present invention relates to a system and method for customized mediation by predicting the risk of in-

The present invention relates to a system for providing nursing services for in-patient patients, and more particularly, to a system for providing nursing services for in-patient patients, more particularly, And a method for providing a personalized mediation service.

Patients admitted to the hospital are very vulnerable to falls due to unfamiliar circumstances, acute illness, surgery, bed rest, medication, treatment and treatment, and various tube and catheter insertion. Especially, as elderly patients are increasing due to aging, many accidents due to falls occur in hospitalized patients. Fall injuries lead to bleeding, fractures, permanent functional impairment, and serious death from bruises, as well as longer treatment periods and increased medical costs due to injuries.

Inpatient falls are known to be typical preventable injuries occurring in medical institutions. Falls falling with inpatient falls and injuries are included in the main criteria for evaluating service quality of domestic and overseas medical institutions. Therefore, if the fall can be accurately predicted from preventable damage, it can be a great help to prevent falls and subsequent sequelae.

With regard to the prevention of fall of the inpatient, there is provided a sensor or a photographing means for detecting the movement or movement of the inpatient in the vicinity of the inpatient, and when an abnormal symptom is detected in the behavior of the patient by these sensors or photographing means A method of notifying a nurse is known (for example, Korean Patent Publication No. 2016-0027411). However, if the patient's behavior is captured and the fall of the hospitalized patient is detected, the patient's sudden abnormal symptoms can not be responded promptly. Thus, in the event of a fall in actual hospitalization, it is not free from the physical injury of the hospitalized patient and the claim from the inpatient caregivers. In addition, when the behavior of the inpatient is reported using the image processing device, etc., it may cause a privacy invasion problem of the inpatient. In addition, there is a high dissatisfaction with the interruption and interruption of the nurses due to the high false positive signal of the sensor in the practical work, and it is difficult to apply in the case of a multi-patient room such as a 6-person room. It is showing.

SUMMARY OF THE INVENTION The present invention has been made in order to solve the problems of the prior art described above, and it is an object of the present invention to provide a system capable of efficiently providing information on the inpatient patient to the medical staff, System to provide personalized mediation services to individual patients.

Another object of the present invention is to provide a system capable of providing highly accurate information on the risk of falling while minimizing the privacy infringement of the inpatient patient, and to provide personalized arbitration service to all hospitalized patients as well as multi-patient rooms using this system To provide a way to

According to an aspect of the present invention, there is provided a customized arbitration service system for predicting a fall risk of an inpatient patient, the system comprising: a database for storing clinical data of an inpatient patient stored in a clinical database, A data conversion module that corresponds to the concept represented in the fall risk prediction algorithm of the fall risk prediction algorithm and converts the data into data that can be input to the fall risk prediction algorithm; A data conversion module for setting the category and the category for causing fall based on the clinical data of the inpatient patient converted by the data conversion module and storing the data that can be input to the fall risk prediction algorithm by the data conversion module A fall prediction module that receives clinical data of the converted inpatient and calculates a risk risk probability of the inpatient based on a link between the respective concepts and concepts; Providing a fall risk level for each inpatient based on a fall risk probability of the inpatient calculated in the fall prediction module; and providing a fall risk level for each in-patient patient based on the fall risk probability of the in- A customized intervention module that presents services to individual patients; And a fall event reporting module for generating a fall incident report generated in the inpatient using the data used in calculating the fall risk probability, based on the fall risk of the inpatient.

For example, the fall causality category set in the fall prediction module may include the demographic information of the inpatient, hospital information, nursing condition at the time of admission, doctor prescription information including dosage, nursing demand information, nursing assessment, nursing diagnosis , Nursing intervention and outcome information.

In one exemplary embodiment, the data conversion module is configured to perform an algorithm for predicting falls risk through standard term mapping of clinical data of the inpatient in a predetermined period of the clinical data of the inpatient patient stored in the clinical database And can be converted into data that can be entered into the fall risk prediction algorithm.

For example, the data conversion module may include: a data extracting unit for selecting and extracting data related to prediction of a fall risk of an inpatient from clinical data of an inpatient patient stored in the clinical database; A standard term mapping unit for mapping standard terms to correspond to concepts expressed in an algorithm for predicting fall occurrence of an inpatient; and a standard term mapping unit for mapping the data mapped in standard terms by the standard term mapping unit, Into a data type defined by an algorithm for prediction.

The fall prediction module may be configured to generate a node based on the scale of the set concept and apply a dynamic Bayesian network that links the nodes to form a link, A fall prediction setting unit for setting knowledge related to a fall prediction model by constructing a conditional probability table; a fall prediction unit for calculating a fall risk probability for each of the inpatients using the knowledge related to the fall prediction model set in the fall prediction setting unit; Wherein the clinical data of the inpatient patient is converted into data that can be input to the fall risk prediction algorithm by the data conversion module in accordance with the period, Judges the concept, and determines an image corresponding to the determined concept Conditional probability may include according to the period from the table on the basis of the clinical data of the inputted inpatient fall prediction inference unit for calculating a probability for the fall risk hospital patients, respectively.

According to another aspect of the present invention, there is provided a method for providing a nursing intervention service through predicting a fall risk of an inpatient patient, the method comprising: a data conversion module for classifying clinical data of an inpatient patient extracted from a clinical database into an inferior risk prediction algorithm Corresponding to the represented concept and transformed into data that can be input to the fall risk prediction algorithm; The fall prediction module is configured to set a category and a category of the fall occurrence cause based on the converted clinical data of the inpatient patient, and the data that can be input to the fall risk prediction algorithm by the data conversion module Receiving the clinical data of the inpatient who has been converted into the inpatient, and calculating the risk risk probability of the inpatient based on the link between the concepts and the concepts; Based on the fall risk probability of the inpatient patient calculated in the fall prediction module, a fall risk level is provided for each inpatient, and the fall risk for each inpatient is determined by the customized intervention module, A step of presenting a customized arbitration service required for the inpatient by referring to the data value; And a step of generating a fall incident report based on the entered nursing record and the inpatient data used for calculating the fall risk probability of the inpatient by the fall incident reporting module, This study suggests a method of providing customized intervention services through prediction of fall risk of hospitalized patients.

The present invention proposes a system and method for inferring the risk of falling of an inpatient based on clinical data of an inpatient and providing a customized intervention guidance service for each inpatient based on the data used for reasoning results and reasoning .

According to the system and method of the present invention, it is possible to objectively and efficiently predict, evaluate, and calculate the risk of a fall in hospitalized patients based on clinical data of hospitalized patients, to inquire about the possibility of a fall in hospitalized patients based on the calculated evaluation data, Can be confirmed.

There is no possibility of infringing on the privacy of the hospitalized patients compared with the fall detection method based on the movement or the behavior of the hospitalized patients simply based on the objective electronic medical record data on the inpatient. In addition, we provide customized intervention guidance to provide medical staff with information on the possibility of falling inferiority of reliable hospitalized patients using objective medical record data while minimizing user interruption, and to provide individualized nursing intervention services to all inpatients .

1 is a schematic diagram schematically illustrating a nursing intervention guidance system through prediction of a fall risk of an inpatient in accordance with an exemplary embodiment of the present invention.
2 is a schematic diagram that schematically illustrates a module that configures an apparatus for predicting a fall risk of an inpatient in accordance with an exemplary embodiment of the present invention.
3 is a diagram schematically illustrating an example of a table in which clinical data of an inpatient patient is mapped to standard terms in connection with predicting the risk of falling of an inpatient in accordance with an exemplary embodiment of the present invention.
Figure 4A is an illustration of an example of transforming clinical data of an inpatient patient mapped in standard terms into data suitable for a fall risk prediction algorithm in connection with predicting fall risk of an inpatient in accordance with an exemplary embodiment of the present invention .
Figure 4B is a schematic diagram that schematically illustrates the data flow until the inpatient clinical data is converted to data suitable for the fall risk prediction algorithm in accordance with an exemplary embodiment of the present invention.
FIG. 5 is a diagram for explaining a Bayesian network model for obtaining inference results by converting various categories and concepts defined in the fall prediction module according to an exemplary embodiment of the present invention into a scale or a variable.
FIG. 6 is a diagram illustrating an example of a screen presented through a speculation result inquiry unit according to an exemplary embodiment of the present invention.
FIG. 7 is a view showing an example of a screen presented through the customization arbitration guide according to the exemplary embodiment of the present invention; FIG.
8 is a view showing an example of a screen presented through the fall occurrence report module according to an exemplary embodiment of the present invention.
FIG. 9 is a graph showing ROC (receiver operating characteristics) curves for fall occurrence prediction results deduced in the fall prediction module according to an exemplary embodiment of the present invention. FIG.

Hereinafter, the present invention will be described in more detail with reference to the accompanying drawings where necessary.

1 is a schematic diagram schematically illustrating a customized intervention guidance system through prediction of a fall risk of an inpatient in accordance with an exemplary embodiment of the present invention. 1, the nursing intervention service system 100 according to the present invention includes a clinical database 110, a database 110 connected to the clinical database 110, and various data and / Or a personalized mediation device 200 that transmits, receives, and processes information.

For example, the clinical database 110 may be a storage means for recording clinical data of an inpatient, for example, an electronic medical record system including an electronic nursing recording system (ENRS). EMRS) or electronic health record system (EHRS).

The custom arbitration device 200 may be any device or device that is accessible to the clinical database 110 and may be, for example, administered by hospital practitioners and connected to an electronic medical record system (EMRS) A mobile device, a medical device, or the like, capable of inquiring or inputting various information about the inpatient, including data.

For example, the custom mediation device 200 may include data such as clinical data of the inpatient, data or information extracted from the clinical data, the probability of a fall occurrence risk calculated therefrom, and a personalized mediation record including a fall incident record Input means such as a keyboard or a touch panel for inputting information; Output means such as a printer or a display for inquiring and outputting the data or information; A memory for storing information such as a fall risk prediction algorithm and information on a fall prediction probability and / or a fall occurrence report history of an inpatient calculated by the algorithm, as well as an appropriate application for implementing a fall risk prediction algorithm It may be a computing device that is mounted.

In FIG. 1, a custom arbitration device 200 equipped with an algorithm for predicting a fall occurrence of an inpatient patient is connected to the electronic medical record system (EMRS) in which a clinical database of an inpatient patient is recorded and stored. However, (200) may be implemented in the same system area as the EMRS.

In one exemplary embodiment, as shown in FIG. 1, the custom arbitration device 200 may include a plurality of modules and a database (DB). As used herein, the term " module " may refer to a logical unit or set of hardware and / or software capable of executing the program code, It does not.

The term " database (DB) " herein may refer to a functional and structural combination of software and hardware that stores information corresponding to each DB. The DB may be implemented as at least one table, and may further include a separate DBMS (Database Management System) for searching, storing, and managing information stored in the DB. In addition, it can be implemented in various ways such as a linked-list, a tree, and a relational DB, and includes all data storage media and data structures capable of storing information corresponding to the DB.

In one exemplary embodiment, the custom arbitration device 200, which may be networked with the clinical database 110, includes a data transformation module 210, a fall prediction module 220, a custom arbitration guidance module 230, A reporting module 240, a database module 250, and a controller 260 such as a CPU and a scheduler for controlling data transmission and processing by these modules.

The data conversion module 210 corresponds the clinical data of the inpatient patient stored in the clinical database 110 to the concept used in the algorithm for predicting the fall risk of the inpatient through standard term mapping, . If necessary, the data conversion module 210 extracts only recent clinical data from the clinical data recorded and stored in the clinical database 110 at regular intervals, for example, 24 hours or 48 hours, It is possible to cope with the concept of predicting the fall risk of the patient and convert it into data suitable for input into the fall prediction risk algorithm.

The fall prediction module 220 is mapped to the standard terms by the data conversion module 210 and based on the clinical data of the inpatient patient converted into a form of data suitable for input into the fall prediction risk algorithm, Predict, infer, and calculate the risk of falling risk. To this end, the fall prediction module 220 establishes and defines a category and a concept for each category that are known to be related to fall occurrence. Also, the fall prediction module 220 calculates the fall risk probability of the inpatient in each concept and category based on the clinical data mapped to the standard term and converted into the data format suitable for the fall risk prediction algorithm can do.

The custom arbitration guidance module 230 presents the fall risk and the risk level for each inpatient in the case where there is a request for inquiry by a medical staff based on the probability of a fall risk of the inpatient calculated in the fall prediction module 220 . Also, the custom arbitration guidance module 230 may provide an essential arbitration arbitration service required to prevent fall of the inpatient in question based on the fall risk level for the inpatient calculated in the fall prediction module 220 to the custom arbitration device 200 ), And can be displayed. The content of the customized arbitration presented is provided in an individualized manner according to the clinical context or condition of the individual patient using the data of the clinical database 100 and the data conversion module 210.

The fall event reporting module 240 guides the nursing record for the fall in relation to the fall incident on the inpatient and records the fall incident report based on the entered nursing record and the inpatient data used to calculate the fall risk probability of the in- Respectively. At this time, the data set of the data conversion module 210 used in deducing the latest fall occurrence probability, the risk level, and the fall prediction probability stored in the database module 250 can be used.

The database module 250 stores the data sets used in the custom arbitration device 200 and the fall prediction results. The stored data may be used by the custom arbitration guidance module 230 and the fall event reporting module 240 via the controller 260.

The control unit 260 performs an interface with the clinical database 110 and controls the input, output, transmission, reception, and processing of data performed in the custom arbitration device 200 as a whole. For example, the control unit 260 may control the input of data, such as numbers, in the custom arbitration device 200, or input of appropriate information when selecting from a plurality of data / information, ) Controls the flow of data / information, such as the transmission and reception of input data / information in various modules.

According to the present invention, it is possible to objectively predict the risk of falling for each inpatient on the basis of clinical data on the inpatient, and to provide a personalized nursing intervention guide for each patient. Conventionally, fall risk assessment to prevent falls in hospitalized patients was to repeatedly apply 5 to 9 items. However, the tools for assessing falls in the traditional inpatient population rely heavily on nurses' subjective judgment, depending on the abstract concept. Because there is a difference in the clinical judgment of experienced nurses and less experienced nurses, there is a large variation in the content and frequency of nursing intervention services to prevent falls occurring in the same high-risk patients, The incidence also varies. Based on a review of the traditional fall risk assessment tools published by the Cochrane Collaboration in 2012, which provides an academic basis for evidence-based healthcare practices, the clinical efficacy of traditional fall risk assessment tools is very poor . The nursing academy, including the American Nursing Association, acknowledges that the practical effect of the fall risk assessment tool is "moderate". The prediction accuracy of the evaluation tool for the fall inpatient is greatly lowered as well as a large difference according to the type of the medical institution or the characteristics of the patient group, and thus there is a limitation in providing an effective nursing service for the inpatient by predicting the actual fall occurrence .

However, according to the system of the present invention, it is possible to estimate the risk of falling for each in-patient patient by fully taking into account the categories and concepts closely related to the occurrence of the inpatient based on the clinical data of the in- Based on this, it is possible to provide customized intervention services based on scientifically known grounds regarding the level of fall risk that can occur to each inpatient and the individual patient status. According to an exemplary embodiment of the present invention, the customization of the present invention, which is configured to predict and calculate the risk of a fall in hospitalized patient based on clinical data of an inpatient patient, and to provide nursing services for an inpatient patient objectively feasible The configuration and function of the custom arbitration device 200 constituting the arbitration service system 100 will be described in more detail.

2 is a schematic diagram that schematically illustrates a module that configures a custom arbitration device 200 for predicting fall risk of an inpatient in accordance with an exemplary embodiment of the present invention. 2, a data transformation module 210, a fall prediction module 220, a custom arbitration guide module 230, and a database module 220, which constitute a custom arbitration device 200 according to an exemplary embodiment of the present invention, (250) may be divided into a number of functional parts to implement the functions implemented in each module.

The data conversion module 210, which maps the clinical data of the inpatient patient to the standard term for predicting the fall risk and converts it into a form suitable for the fall risk prediction algorithm, includes a data extraction unit 212, a standard term mapping unit 214, And a data conversion unit 216. [

The data extracting unit 212 extracts data related to the occurrence of inpatient falls from the inpatient clinical database 110 (see FIG. 1), for example, clinical data of the inpatient patient recorded and stored in the electronic medical record system (EMRS) Clinical data is selected and extracted. In one exemplary embodiment, the data extracting unit 212 retrieves data necessary for predicting occurrence of an inpatient's fall from the inpatient's medical care information and nursing information stored in the electronic medical record system (EMRS) Can be extracted.

An independent fall prediction service scheduler that arbitrates data extraction, transmission, reception and processing between the clinical database (EMRS) 110 and the data conversion module 210, with respect to the clinical data of the inpatient patient extracted by the data extraction unit 212 (Not shown). The scheduler may be implemented in the clinical database 110 or in a separate program form. The data extracting unit 212 extracts the clinical data of the inpatient patient recorded and stored in the clinical database 110 for a predetermined period of time, for example, 24 hours or 48 hours set by the fall prediction service scheduler, And can be configured to selectively extract clinical data. In other words, the data extracting unit 212 sets a period corresponding to a predetermined period according to a set value defined in the fall prediction scheduler when searching clinical data of an inpatient, Day) or within 48 hours of data collection and extraction with timestamp.

Thus, the reason why the fall prediction service scheduler extracts and collects only the clinical data within a certain period from the clinical database of the hospitalized patient reflects the flow of the nursing work that repeatedly observes and measures the frequent state changes of the hospitalized patients. It is preferable to extract only recent clinical data based on a predetermined period because recent clinical data among clinical data of hospitalized patients are meaningful in the inpatient fall risk evaluation result of the fall prediction module 220 which will be described later.

For example, for all patients on the day of admission, or for hospitalized patients who are determined to have a high-risk of falling, clinical data up to the past 24 hours can be extracted based on the current time of the system, - For inpatients determined to be low-risk, clinical data up to a minimum of 24 hours in the past based on the current time of the system can be extracted. At this time, if there is overlapping measured clinical data within each set period, all the clinical data can be extracted in the order of time stamp, and if there is no clinical data, the missing data can be processed.

Further, when the data extracting unit 212 starts extracting the clinical data of the hospitalized patient, it may follow a specific set value defined in the fall prediction service scheduler. For example, a time-driven approach such as 0:00 (midnight) or nurse shift (such as 7:00, 15:00, 22:00) can be adopted based on the current time of the system. Alternatively, an event-driven method may be used to extract and collect clinical data of the hospitalized patient, depending on whether the hospitalized patient has undergone surgery or a high-risk procedure, or if there is a specific event, such as motorized (nursing unit change) have.

At this time, the clinical data of the inpatient patient extracted through the data extracting unit 212, for example, the clinical data of the inpatient patient extracted from the electronic medical record system (EMRS), is stored in the tuple (Tuple) . Therefore, it is necessary to classify the clinical data of the inpatient patient extracted by the data extracting unit 212 into standard terms corresponding to the fall risk prediction algorithm. The standard term mapping unit 214 performs a standard term mapping of the clinical data of the inpatient patient extracted through the data extracting unit 212 to generate a standard term map . For example, the clinical data of the inpatient can be generated as a mapping table through the standard term mapping unit 214.

3 shows an example of a mapping table that must be defined in advance in the standard term mapping unit 214 according to an exemplary embodiment of the present invention. This mapping table may be configured for a single medical institution or may be configured for multiple medical institutions. With this mapping table, the inpatient clinical data can be matched to the fall risk prediction algorithm. On the other hand, Table 1 below shows an exemplary structure of a standard term mapping table of variables related to a fall risk prediction algorithm.

According to the exemplary embodiment, as shown in FIG. 3 and Table 1, the mapping table generated by the standard term mapping unit 214 so that the clinical data corresponds to the standard terminology is the International Health Terminology Development Organization Standard Development Organization (IHTSDO), version 2016 (http://browser.ihtsdotools.org/) of the systematized nomenclature of medicine - clinical terms (http://browser.ihtsdotools.org/) recognized by the US Department of Health and Human Services as a terminology system for medical information standards and / Based on the International Nursing Practice Classification System (ICNP®) version 2015 (http: //www.icn/ch/ICNP-Browser-New.html) developed for electronic nursing records by local EMRS And the fall risk prediction algorithm. As such, the standard terminology system helps to understand correct conceptual mapping and system implementation work through presentation version, English notation, Korean notation, concept code and axis or type information representing concept structure. Local EMR is divided into medical institution name, EMR system name and version, so system and version change can be tracked.

In an exemplary embodiment, the data of the mapping table generated to correspond to the fall risk prediction algorithm may be mapped 1: 1 with the clinical data of the inpatient recorded in the EMR. In another embodiment, the data in the mapping table may be mapped to 1: n (where n is an integer greater than or equal to 2) with the clinical data of the inpatient. Further, the data of the mapping table may be generated secondarily from the clinical data recorded in one or more EMRs by calculation or conversion rules. In this case, the data of the mapping table generated by the standard term mapping unit 214 and the data finally converted by the data conversion unit 216, which will be described later, should be referred to together.

In addition, the relationship between SNOMED CT and ICNP® suggests that SNOMED CT includes ICNP® as a more comprehensive terminology, but as a discrete term system, the time differences that occur in the process of continually updating and the ICNP® structure type , Judgment, etc.) and hierarchical information to complement the nursing context, so that the clinical data can be accurately mapped when actually implementing the fall risk prediction algorithm.

In the mapping table exemplarily shown in FIG. 3, a category and a concept are classified into categories applicable to the fall risk prediction algorithm constituting the fall prediction module 220, which will be described later in detail, , A concept or a factor constituting a node. Thus, the SNOMED CT and ICNP® terms mapped to each concept are terms of granularity that best represent the concept, but if the terminology for expressing sufficient detail is not in the terminology, ) Terms.

In the exemplary embodiment, when the age of the patient is 38 years and the number of days of hospitalization (Admdays) is 7 days among the clinical data of the inpatient recorded in the EMR, the standard term mapping unit 214 stores the standard term according to the SNOMED CT standard , The standard ICNP standard term, the local EMR system data information, and the fall risk prediction algorithm to generate the standard term data, and the inpatient age data "[current chronological age (424144002), 38] stay (308540004), 7] ".

The data conversion unit 216 converts the clinical data of the inpatient patient mapped to the standard term for the fall risk prediction algorithm by the standard term mapping unit 214 into the fall risk prediction algorithm, ), The clinical data of the inpatient who has been mapped in standard terms is converted into the data types and formats defined in the fall risk prediction algorithm. Conversion rules are determined according to the clinical characteristics and meaning of the clinical data of hospitalized patients mapped in standard terms. For example, they are mostly expressed as "If-then conditional expressions", "Boolean expressions" and "operations". However, more complex expressions may be used for clinical data for inpatients with different measurements. In particular, the definition of the data conversion unit 216 may have a large effect on the prediction accuracy because the data conversion rule having the highest prediction accuracy is set through the optimization process of the fall prediction model.

FIG. 4A shows the clinical data of an in-patient patient mapped to a standard term by the standard term mapping unit 214 according to an exemplary embodiment of the present invention, as input data for a fall prediction of an in-patient patient by the data conversion unit 216 And schematically shows the process of conversion. Figure 4B also shows a schematic diagram of a data flow up to the time the clinical data of an inpatient is converted to data suitable for a fall risk prediction algorithm, in accordance with an exemplary embodiment of the present invention.

As shown in FIG. 4A, for example, when the age of the inpatient is 38 years and the number of days in the hospital is 7 days, a classification value of 3 is given to the age of the inpatient concerned with the fall prediction, and 2 Quot; is given. In this way, for the fall risk prediction algorithm of the inpatient patient schematically shown in Table 1 and FIG. 3, the standard term mapping unit 214 performs the hospitalization defined in the mapping table according to the category and concept related to the clinical data of the in- The patient's data may be converted into appropriate data for input to the inpatient's fall risk prediction algorithm in the data converter 216. [ Based on the clinical data of the inpatient patient who has been converted into data suitable for the fall risk prediction algorithm of the inpatient in the data conversion module 210, the fall prediction module 220 estimates the risk of falling, i.e., Reasoning, recording. For example, as shown in FIG. 2, the fall prediction module 220 may include a fall prediction setting unit 222, a fall prediction reason unit 224, and the like.

The fall prediction setting unit 222 stores the knowledge related to the fall prediction model of the inpatient and constructs the conditional probability data on the fall risk of the inpatient based on the knowledge. Table 2 below shows an example of categories and concepts (nodes) associated with fall prediction of an inpatient patient stored in the fall prediction setup section 222 according to an exemplary embodiment of the present invention. 5 is a block diagram illustrating a Bayesian network model for obtaining inferred results by converting various categories and concepts defined in the fall prediction setting unit 222 constituting the fall prediction module according to an exemplary embodiment of the present invention into a scale or a variable Fig.

As shown in Table 2, in relation to the risk of a fall in hospitalized patient, the fall prediction setting unit 222 largely includes demographic information, hospitalization information, hospital nursing condition, nursing information, risk assessment tool (nursing demand information) , Nursing diagnosis, nursing assessment, nursing intervention, and medication. In addition, although the concept illustrated in Table 2 is set in each category for each of these categories, the present invention is not limited thereto. As described above, in relation to the risk of a fall in hospitalized patient, the fall prediction setting unit 222 defines and stores a category related to the fall prediction model of the hospitalized patient and a concept belonging to each category.

The fall prediction setting unit 222 receives the clinical data of the inpatient patient converted by the data conversion unit 216 in accordance with the fall risk prediction algorithm. Next, as shown in FIG. 5, the clinical data of the inpatient patient converted in accordance with the fall risk prediction algorithm is compared with the relationship between the concept node and the concept node, which is defined, set and stored in advance in association with the fall prediction model (E.g., an arrow). The mapped data creates different conditional probability tables for each node on the combination of the parent node (the node giving the arrow) and the child node (the node receiving the arrow).

At this time, a certain scale is assigned to the concept classified by each category, and a concept mode can be created based on such a scale. In addition, conditional probability data on the risk of falling of hospitalized patients can be constructed by applying a dynamic Bayesian network that links the concept nodes to links.

In an alternative embodiment, fall prediction setter 222 may use decision tree, fuzzy logic, deep-run (), and so on, using the same data set, in addition to the dynamic Bayesian network algorithm shown schematically in FIG. machine-learning algorithm such as learning can be used, or a hybrid algorithm using two or more algorithms can be used.

There are a variety of ways to express knowledge for fall prediction, for example using data mining tools like SAS®, the enterprise miner for SPSS®, the PASW modeler and Clementine, open source Inka, Rapid Miner, etc. It can be expressed visually. Tools such as R, tensor flow, Azure, Torch, etc. can also be used as a deep-running tool.

The fall prediction inference unit 224 directly calculates the fall risk of the inpatient using the knowledge related to the fall prediction model defined in the fall prediction setting unit 222. [ In an exemplary embodiment, the fall prediction inferring unit 224 calculates a falling risk probability of an inpatient based on the latest clinical data for the inpatient in accordance with a predetermined cycle. That is, when the clinical data of the inpatient in the recent 24 hours or 48 hours is mapped to the standard term through the data conversion module 210 and converted into data suitable for the algorithm for predicting fall of the inpatient, The clinical data corresponding to the concept set in the fall prediction setting unit 222 among the clinical data of the inpatient patients mapped by the standard term by the standard term 210 is inputted and the concepts corresponding to the corresponding clinical data and the respective concepts are included in each cycle Judged the category to be classified.

Then, the data of the conditional probability table corresponding to each category determined through the dynamic Bayesian network that has been performed in the fall prediction setting unit 222 is extracted, and the latest clinical data is used to calculate the fall The risk probability can be calculated. Through such a method, the fall prediction reasoning unit 224 can calculate the probability of a fall occurrence risk for each inpatient and a certain period. This probability value and the data input value used for inference can either be stored in the clinical database 110 (see FIG. 1) in the EMRS server or stored in an inference database 252 that can configure the database module 250.

For example, a fall risk prediction result for each inpatient patient derived through the fall prediction inference unit 224 may be a probability indicating a fall probability for each patient. This value can be calculated in the range of 0 to 1, where 0 means 0% probability of falling, and 1 means 100% probability of falling. The probability of falling occurrence risk probability for each inpatient predicted by the fall prediction inferring unit 224 is determined by constructing a clinical database 110 (see FIG. 1) such as an EMRS server and / or a custom arbitration device 200 together with a time stamp Is stored in an inference database 252 that constitutes a separate database module 250 that performs the functions described herein. For example, the fall prediction risk probability data recorded and stored in the clinical database 110 and / or the inference database 252 may include patient information (e.g., registration number, name, sex, age, The inpatient nursing unit), the occurrence date of the fall occurrence risk inference, the clinical data of the hospitalized patient converted through the data conversion unit 216 fed with the fall risk prediction algorithm constituting the fall prediction module 220, 1). ≪ / RTI >

The data stored in the reasoning database 252 constituting the custom arbitration device 200 may be used in the custom arbitration guidance module 230 such as the speculation result inquiry unit 232 and the custom arbitration guide unit 234 described later. In addition, the data stored in the inference database 252 may be utilized for periodically evaluating, updating, and learning the fall prediction module together with actual fall event data generated and stored in the fall event reporting module 240.

For example, when a dynamic Bayesian network model is adopted as an algorithm for predicting falls in inpatients, nursing informatics experts intervene in evaluation, update, and learning of fall prediction module to reevaluate and optimize model variables, factors and structures Can be performed. On the other hand, when machine learning such as deep-running is adopted for predicting the fall of the inpatient, whenever an actual fall accident occurs to the hospitalized patient, the new occurrence can be automatically performed by reflecting the accident occurrence data.

The custom arbitration module 230 may include a speculation result inquiry part 232 and a custom arbitration guide part 234. The inference result inquiry unit 232 informs the user, that is, the nurse, of the prediction result of the risk of falling occurrence for each inpatient obtained through the fall prediction module 220. [ 6 is a diagram showing an example of a screen presented through the inference result inquiry unit 232 according to the exemplary embodiment of the present invention. As shown in FIG. 6, the reasoning result inquiry unit 232 is provided to the user by the nursing service unit, the patient, and the inference result generation time for the one-unit or user-set period.

For example, the inference result inquiry unit 232 basically provides the user with the date and probability value of the risk of the inpatient occurrence of the inpatient and basically provides the probability value of each risk of falling occurrence according to the cutoff Risk-level information can be given. The threshold value can be set according to the policy of each medical institution, and the fall risk level can be divided into two (high-risk, low-risk) or three (high-risk, intermediate risk, low-risk) . The fall risk level allows the user to quickly grasp the risk of falling of the inpatient through a visualization expression such as color or icon according to severity of the fall occurrence, It can support business.

Optionally, the speculation result inquiry unit 232 may provide a list of fall prevention service priority persons by inquiring only the high-risk level patient list, or by specifying the inquiry condition only such that the patient list is above the fall prediction probability value designated by the user . ≪ / RTI > In addition, hyperlinks per patient and instance provide transformed data, raw data, or additional reference / description data from the inpatient clinical database on which the speculation results are based to help users understand And may provide user-specific learning opportunities.

The customization arbitration guide 234 accesses the nursing record according to the risk of falling occurrence using the data of the inpatient patient falling in the inference database 252 and the converted data and the raw data from the inpatient clinical database And provides a nursing intervention prompt to the user. The Nursing Interview Prompt prompts the necessary nursing interventions, such as the patient assessment to be performed at the relevant risk level, the patient / guardian training and surrounding environment, output of training data, and communication between the medical team and the caregiver / It is presented according to the clinical context. For example, it is inappropriate to educate patients with urinary catheterization on the day of surgery, and training of intravenous poles and line caution is not appropriate for patients who do not have continuous intravenous injections, such as fluid or blood injections. Is different. 7 is a diagram showing an example of a prompt screen presented through the customization arbitration guide 234 according to the exemplary embodiment of the present invention.

The contents of the arbitration provided through the customized arbitration guidance unit 234 are configured according to the practical guidelines in the medical institution reflecting the international nursing practice guidelines and the medical institution status, It can also be implemented as a set of risk nursing statements. Preferably, the customization arbitration guide 234 reminds the customization arbitration content in an active form when the user accesses the patient record, and allows the user to quickly enter the content.

For example, the Nursing Intervention Template is automatically popped up once per nursing staff in the case of fallen high-risk level patients, indicating that the patient who is currently performing the record is a fallen-high-risk patient and presents the nursing intervention to be provided . After that, it provides a function to easily activate the template through the separate fall nursing button in the nursing journal. If the fall risk level of the patient is changed from high-risk to low-risk, the automatic pop-up may be implemented to be released. On the other hand, if there is an event that affects the fall risk level change such as high risk procedure, surgery, or electric power, the fall prevention nursing guidance prompt can be automatically popped up to facilitate detailed patient evaluation.

For example, the contents of the nursing intervention template can be linked to the standard nursing statements in the nursing record. Therefore, the contents of the nursing intervention entered as a template can be reflected in the nursing journal input together with the "risk of falling risk" or "fall risk" nursing diagnosis, and can be viewed together with other nursing statements on the nursing journal inquiry screen. If necessary, existing fall risk assessment tools such as "Morse Fall Scale", "STRATIFY", and "Hendrich II" can also be included in the nursing intervention guidance to help prevent missing re-assessments.

The fall event reporting module 240 generates a fall incident report of the medical institution based on the entered nursing record and the inpatient data used to calculate the fall risk probability of the inpatient, . 8 is a view showing an example of a screen presented through the fall occurrence report module according to an exemplary embodiment of the present invention.

The original fall incident report is an administrative quality document prepared after the fall of the hospitalized patient. It is written separately in the existing medical institution irrespective of the nursing record. The fall event reporting module 240 of the system of the present invention associates the fall event report with the fall record of the nursing journal so that a draft can be automatically generated. Because the fall record is aimed at early detection of damage through immediate evaluation and assessment of fall patients and prevention of recurrence of additional fall, it is important that the date and time of occurrence, the situation of detection, the content of immediate physical assessment, It consists of contents.

Similarly, falls report is information gathered for medical institution service quality management purpose such as occurrence frequency, location, type, degree of damage, risk of fall at hospitalization, measurement method, risk of falling, and additional measures in medical institutions. Should be entered into the nursing record and the incident report. Thus, much of the content overlaps, and the fall event reporting module 240 of the present system automatically generates a report draft using the fall record of the nursing journal. The draft generated by the fall event reporting module 240 may need to be modified or supplemented from outside the system.

The linkage between the fall event report generated and generated by the fall event reporting module 240 and the nursing record is essential for preventing the omission of records or missing event reports frequently occurring in the event of a fall and improving the quality of the preventive nursing intervention guidance service can do. For example, the fall risk level for a patient who has a fall record in the nursing journal and for which a fall incident reporting draft has been generated is immediately changed to high-risk. These patients will remain at high-risk levels by the day of discharge. Data such as a fall incident report generated and generated by the fall event reporting module 240 may be recorded and stored in a fall database 254 constituting the database module 240.

Using the above-described customized intervention guidance service system, clinical data based on the clinical data recorded in the inpatient clinical database, in particular the recent clinical database, is converted into clinical data suitable for the fall risk prediction algorithm, The probability of a fall occurrence risk can be objectively predicted for each inpatient by using a concept node classified by a related category and a networked algorithm linked by a link between the concept node and the concept node. The timely provision of customized intervention services, which is essential for hospitalized patients at high risk of falling, can dramatically reduce the incidence of hospitalized patients.

Indeed, the likelihood of falls occurring for inpatients was assessed using the customized mediation service provision system described with reference to FIGS. 1-8. We randomly selected 9,507 (81,451 hospitalization) data from patients who were hospitalized at Seoul National University Bundang Hospital from September 1, 2014 to September 1, 2015 for more than one day at the neurological center, hematologic tumor and cardiovascular center After defining the concept node (corresponding to the categories and concepts set in Table 3) and links that constitute the dynamic Bayesian network, the fall occurrence prediction model was learned. We compared 4,800 data (40,728 hospital days) that were not used to construct a prediction model of the fall occurrence algorithm with actual fall incident records. The results of the c-statistics (statistics) evaluating the fall prediction power in the fall prediction module of the present invention are shown in Table 3, and FIG. 9 is a graph showing the results of the fall occurrence prediction inferred in the fall prediction module according to the exemplary embodiment of the present invention FIG. 5 is a graph showing a receiver operating characteristic (ROC)

As shown in Table 3, the sensitivity was 94.9%, the specificity 82.62%, the positive predictive value 84.45%, and the negative predictive value 94.22% at the drop probability 20%. Also, as shown in FIG. 9, the area of the area under curve (AUC) of the receiver-operating characteristic curve (ROC) showing the overall performance of the prediction is 0.9573, which is quite high.

Although the present invention has been described based on the exemplary embodiments and examples of the present invention, the present invention is not limited to the technical ideas described in the above embodiments and examples. Those skilled in the art will appreciate that various modifications and changes may be made without departing from the scope of the present invention. It will be apparent, however, that such modifications and variations are all within the scope of the present invention.

100: Custom Mediation Service System
110: clinical database (EMRS, electronic medical record system)
200: Custom Mediation Device
210: Data conversion module 212: Data extraction unit
214: Standard term mapping unit 216: Data conversion unit
220: Fall prediction module 222: Fall prediction setting module
224: falling prediction reasoning part 230: customized arbitration guidance module
232: inference result inquiry unit 234: custom arbitration information unit
240: fall event reporting module 250: database module
252: heuristic database 254: fallback database

Claims (6)

As a customized arbitration service system that predicts the fall risk of hospitalized patients,
In order to link the local patient data stored in the clinical database of the individual medical institution with the fall risk prediction algorithm, the local data of the inpatient patients stored in the clinical database are represented in the algorithm for predicting the fall risk of the inpatient through standard term mapping And converting the inpatient's local data mapped to the standard term into data that can be entered into a fall risk prediction algorithm;
The data conversion module sets a category and a category category for causing falls based on the local data of the inpatient patient converted into data that can be input to the fall risk prediction algorithm by the data conversion module, A fall prediction module that receives local data of an inpatient patient converted into data that can be input to a fall risk prediction algorithm and calculates a fall risk probability of an inpatient based on a link between the set concepts and concepts;
Providing a fall risk level for each inpatient based on the fall risk probability of the inpatient patient calculated in the fall prediction module and based on the data used for the fall risk and the risk prediction for each in- A customized intervention module that presents customized intervention services required for the inpatient in accordance with the Nursing Practice Guidelines; And
The nursing record for a fall occurring in an inpatient is guided, and the probability of occurrence and risk level of the inpatient falling in the inpatient stored in the clinical database and the locality of the inpatient patient converted into the data that can be input to the fall risk prediction algorithm by the data conversion module Based on the data, a fall incident report generated by the inpatient in association with the fall record of the nursing journal,
A Personalized Arbitration Service System by Predicting the Fall Risk of Inpatients.
The method according to claim 1,
The fall causality category set in the fall prediction module includes the inpatient demographic information, hospital admission information, nursing information at the time of hospitalization, nursing information, risk assessment tool, nursing diagnosis, nursing assessment, custom mediation and medication information A customized intervention service system based on prediction of the risk of falling inpatients
3. The method of claim 2,
Wherein the data conversion module maps the local data of the inpatient patient updated in a predetermined cycle among the local data of the inpatient patient stored in the clinical database to a concept expressed in an algorithm for prediction of fall risk through standard term mapping , A personalized arbitration service system that converts the risk data of hospitalized patients into data that can be input to the fall risk prediction algorithm.
4. The method according to any one of claims 1 to 3,
Wherein the data conversion module comprises: a data extracting unit for selecting and extracting data related to prediction of fall risk of an inpatient from local data of an inpatient patient stored in the clinical database; The concept expressed in the algorithm for predicting fall incidence of hospitalized patients through systematic development organization, systematic nomenclature of medicine (SNOMED CT) and standard term mapping according to international nursing practice classification system And a data conversion unit for converting the data mapped by the standard term mapping unit by the standard term mapping unit into a data type defined by an algorithm for predicting fall occurrence of a hospitalized patient, Customized arbitration service system through risk prediction.
The method of claim 3,
Wherein the fall prediction module comprises:
A falling Bayesian network by constructing a conditional probability table for the risk of falling of a hospitalized patient by applying a dynamic Bayesian network that creates a node based on the above- A fall prediction setting unit for setting knowledge related to the model,
A fall prediction inference unit for calculating a fall risk probability for each inpatient using knowledge related to a fall prediction model set by the fall prediction setting unit, And determining a concept corresponding to the local data through the dynamic Bayesian network for each cycle, and determining the concept corresponding to the determined concept from the conditional probability tables corresponding to the determined concept, And a fall prediction inferring unit for calculating a fall risk probability for each of the inpatient patients in accordance with the fall risk prediction of the inpatient.
A method for providing a customized mediation service through predicting the risk of falling of an inpatient,
The data conversion module allows the local data of the inpatient patient stored in the clinical database to be correlated to the inpatient's patient data through standard term mapping so that the local data of the inpatient stored in the clinical database of the respective medical institution can be associated with the fall risk prediction algorithm. Corresponding to the concept represented in the fall risk prediction algorithm, and converting the local data of the inpatient mapped to the standard term into data that can be input to the fall risk prediction algorithm;
By the fall prediction module, based on the local data of the inpatient patient converted into data that can be input to the fall risk prediction algorithm by the data conversion module, a category and a category concept for causing falls are set A data conversion module for receiving local data of the inpatient patient converted into data that can be input to the fall risk prediction algorithm and calculating a risk risk probability of the inpatient based on the link between the respective concepts and concepts step;
According to the customized intervention module, the fall risk level for each inpatient is provided based on the fall risk probability of the inpatient calculated in the fall prediction module, and the fall risk and fall risk calculation for each in- Providing a customized mediation service required for the inpatient in accordance with the fall nursing practice guidelines based on the data used; And
The fall event reporting module allows the nursing record of the fall occurring in the hospitalized patient to be guided, the probability of occurrence and risk level of the inpatient fall of the inpatient stored in the clinical database, and the data that can be input to the fall risk prediction algorithm by the data conversion module Based on the local data of the inpatient who has been converted into the nurse log, the fall incident report generated in the inpatient is generated in connection with the fall record of the nursing journal
A method for providing a customized mediation service by predicting a fall risk of an inpatient patient.

Images