KR102366206B1 - Apparatus for estimating radiologic report turnaround time on clinical setting and method thereof - Google Patents

Abstract

The present invention relates to an apparatus and method for estimating the reading time of an imaging examination in a clinical situation. According to the present invention, there is provided an apparatus for predicting a report turnaround time (RTAT) of a radiology examination in a clinical situation, comprising: an input unit for selecting a type of a radiological examination to be predicted; and a hospital medical information system. A data collection unit for extracting data on a plurality of independent variables necessary for predicting the reading time required for the selected type of radiology examination, and applying the extracted data on the plurality of independent variables to a pre-established machine learning model There is provided an apparatus for predicting required reading time, comprising: a prediction unit for predicting a required reading time for a type of examination; and an output unit for outputting the predicted required reading time through a user terminal.
According to the present invention, it is possible to effectively predict and provide the required reading time (RTAT), which is expected to be required to read the radiology examination, based on the learning model in consideration of the current clinical situation of the hospital.

Description

Apparatus for estimating radiologic report turnaround time on clinical setting and method thereof

The present invention relates to an apparatus and method for estimating the required reading time of an imaging examination in a clinical situation, and more particularly, to a report turnaround time (RTAT), which is a time required for reading an imaging examination in a current clinical situation. Time) to be predicted and provided through a learning model, and to an apparatus and method for estimating the required reading time of a radiology examination.

The reading report prepared by the radiology doctor in a clinical situation is the crystallization of the radiology department, which is made based on medical knowledge and radiology findings. The training course of radiology majors to train radiologists who can write such readings is very professional and has a system that is differentiated from other departments.

The ability to write a report on radiology examination is a result of such a professional and systematic training process, and is an important medium for communicating information and exchanging opinions with the clinician who requested the examination.

In this process, RTAT is a very important factor, and it is said that it is an essential element for presenting the radiological findings on the patient's condition to the clinician, establishing a rapid treatment plan, and performing a fast and appropriate treatment process. can do.

With the development of medical technology and the sharp rise in the level of imaging tests such as CT/MRI and ultrasound, the results of imaging tests have a great influence on clinical diagnosis, which is important in determining the treatment plan and prognosis of patients. It can be said to be a very important data.

Prompt and appropriate reading time can provide the best treatment experience for both clinicians and patients, but if the reading time delay or non-reading occurs, it can hinder the diagnosis and treatment plan decision of the patient, and even cause unnecessary patient visits. This may lead to civil complaints and legal disputes.

The American College of Radiology publishes and continuously updates ACR practice guidelines for communication of imaging findings with radiology departments, preparation of reading reports, and observance of reading time.

Currently, there are no national or academic guidelines for the reading time of radiology examinations in Korea, but each hospital individually sets and adheres to the limits of the reading time required to improve the quality of care and patient satisfaction. Efforts are being made to do so, and this is partially reflected in various hospital evaluation and accreditation scales.

Although the time required to read radiology examinations is not routinely regulated by law or regulations, it is usually judged within 24 hours from the time of examination in the case of emergency room and ward examinations at tertiary hospitals, and within 7 days in case of outpatient visits or before outpatient visits. This is being used as an evaluation scale and reference material in each hospital's own hospital evaluation and hospital accreditation evaluation.

However, since the reading time of a radiology examination can vary depending on the number of waiting tests, the current status of personnel in the hospital, the type of examination, the difficulty, etc., it is necessary to flexibly determine the various variables that can be collected at the current hospital comprehensively. There is this.

The technology that is the background of the present invention is disclosed in Korean Patent Application Laid-Open No. 2020-0003407 (published on January 09, 2020).

The present invention predicts the required reading time (RTAT) that is expected to be required for image reading of a radiology test in the current clinical situation based on a learning model and provides a predictive result. An object of the present invention is to provide an apparatus and method for predicting .

The present invention provides an apparatus for predicting a report turnaround time (RTAT) of a radiological examination in a clinical situation, comprising: an input unit for selecting a type of a radiological examination to be predicted; A data collection unit for extracting data on a plurality of independent variables necessary for predicting the reading time required for a selected type of radiology examination, and applying the extracted data on the plurality of independent variables to a pre-established machine learning model to apply the corresponding type There is provided an apparatus for predicting required reading time, comprising: a prediction unit for predicting a required reading time related to the examination of

In addition, the hospital medical information system may be an information system in which an electronic medical record system (EMR), a prescription delivery system (OCS), and a medical image storage and transmission system (PACS) are linked.

In addition, the plurality of independent variables include the type of the corresponding test, the reading difficulty, the number of currently unread remaining waiting cases, the average reading time required for the corresponding test within the latest set period, the number of currently available specialists, and available specialists It may include a plurality of factors selected from the comprehensive reading competency index for

In addition, the reading required time prediction apparatus may further include a learning unit for learning the machine learning model based on a plurality of independent variables collected in response to the same type of examination in the past and data on the actual reading required time of the corresponding examination. there is.

In addition, the learning unit performs machine learning analysis of the relationship between a plurality of independent variables (X _i ) collected in response to the same type of test in the past and the required reading time (Y) predicted corresponding thereto by machine learning analysis of each regression coefficient ( β _i ) can be estimated.

Here, i is the index of the independent variable, β _i is the i-th regression coefficient applied to the i-th independent variable, and ε is the error correction coefficient between the predicted reading time Y and the actual reading required time.

In addition, the apparatus for predicting the required reading time may further include a storage unit for storing the type of the selected radiology examination, a plurality of independent variables extracted corresponding thereto, and the predicted reading required time by matching with each other.

In addition, the present invention provides a method for predicting a reading turnaround time using a device for predicting a report turnaround time (RTAT) of a radiology test in a clinical situation, the step of receiving a selection of the type of the radiology test to be predicted and extracting data on a plurality of independent variables necessary for predicting the reading time required for the selected type of radiology examination from a hospital medical information system; There is provided a method for estimating the required reading time by applying the model to predicting the required reading time for a corresponding type of examination, and outputting the predicted reading required time through a user terminal.

In addition, the method of predicting the required reading time may further include learning the machine learning model based on a plurality of independent variables collected in response to the same type of examination in the past and data on the actual reading time of the corresponding examination. there is.

In addition, the method of predicting the required reading time may further include the step of matching the type of the selected radiology examination, a plurality of independent variables extracted corresponding thereto, and the predicted reading required time to each other and storing the matching.

According to the present invention, it is possible to predict and provide the required reading time (RTAT), which is expected to be required to read the radiology examination, based on the learning model in consideration of the current clinical situation of the hospital.

In particular, the present invention can flexibly derive the reading required time that is considered to be the closest to the actual reading in the current clinical situation through a machine learning algorithm based on multiple linear regression.

In addition, it enables medical staff to more efficiently establish future diagnosis and treatment plans for the patient (outpatient treatment date selection, treatment reservation, surgery/procedure plan, etc.) based on the predicted reading time.

1 is a diagram showing the configuration of an apparatus for predicting a reading required time of a radiology examination in a clinical situation according to an embodiment of the present invention.
FIG. 2 is a diagram briefly explaining the function and data flow of each part of FIG. 1 .
3 is a diagram illustrating a state in which the prediction result of the radiology examination reading required time is provided on the EMR/OCR linkage screen according to an embodiment of the present invention.
FIG. 4 is a view for explaining a method of predicting a reading required time of an imaging examination using the apparatus of FIG. 1 .

Then, with reference to the accompanying drawings, embodiments of the present invention will be described in detail so that those of ordinary skill in the art to which the present invention pertains can easily implement them.

The present invention is an apparatus and method for estimating the required reading time of a radiology examination in a clinical situation, and learning the report turnaround time (RTAT) expected to be required for reading an image of a radiology examination in the current clinical situation We propose a technique for effectively predicting based on the model.

1 is a diagram showing the configuration of an apparatus for predicting the reading required time of a radiology examination in a clinical situation according to an embodiment of the present invention, and FIG. 2 is a diagram briefly explaining the function and data flow of each part of FIG. 1 .

1 and 2, the apparatus 100 (hereinafter, the prediction apparatus) for estimating the reading required time of a radiology examination according to an embodiment of the present invention includes an input unit 110, a data collection unit 120, and a prediction unit ( 130 ), an output unit 140 , a learning unit 150 , and a storage unit 160 .

The prediction device 100 receives a selection from the user of a test item to be predicted from the entire list of radiology tests provided through the screen of the user terminal 200, predicts the required reading time (RTAT) for the corresponding test, and predicts the predicted reading required time (RTAT) is displayed on the screen and provided.

The prediction apparatus 100 may select one of a list (list) of imaging examinations listed on the system screen, predict the reading required for the selected examination item, and output the output on the screen.

A hospital medical information system (application program) may be installed and executed in the user terminal 200 . The hospital medical information system may be managed and operated from a separate server. Here, the prediction device 100 may be linked with the hospital medical information system, or may be implemented as software inside the hospital medical information system (application program) and provided as an additional function to the corresponding application program.

The user terminal 200 may correspond to a terminal on the side of hospital personnel (doctors, nurses, staff, etc.) It may include receiving devices. Of course, the user terminal 200 may further include a mobile device, that is, a smart phone, a pad, and the like, which is network-connected to the server and capable of computational inquiry. In this case, in the case of a device with a built-in wireless function (smartphone, laptop, pad, etc.), the function of the hospital medical information system itself and the prediction function according to the present invention may be provided in the form of a mobile application.

The hospital medical information system includes an electronic medical record 20 (Electronic Medical Record; hereinafter EMR), a prescription delivery system 20 (Order Communication System; hereinafter OCS), and a medical image storage and transmission system 30 (Picture Archiving Communication System; hereinafter). PACS) may correspond to an interconnected system, and may include an EMR 10 , an OCS 20 , and a PACS 30 .

The hospital medical information system can collect and manage various types of information by interworking with the EMR (10), OCS (20), and PACS (30), and match and manage related information for each patient, and input and search information. , and may provide a user interface for modification and management.

Hereinafter, the configuration of the apparatus for predicting the required reading time of a radiology examination according to an embodiment of the present invention will be described in more detail.

Referring to FIGS. 1 and 2 , the input unit 110 receives a selection (eg, CT, MRI, ultrasound, etc.) of a type of imaging examination to be predicted from a user. In this case, the input unit 110 may select and receive a desired examination item from the user terminal 200 through an input means such as a mouse or a touch display from among the entire examination list (list) provided through the screen of the user terminal 200 . .

The data collection unit 120 extracts and collects, from the hospital medical information system, data on a plurality of independent variables necessary for predicting the reading required for the selected type of radiology examination. In this case, the data collection unit 120 transmits the collected data to the storage unit 160 , and the storage unit 160 may store the collected data by integrating and sorting the collected data for each independent variable.

The data collection unit 120 extracts a plurality of independent variables corresponding to the corresponding examination from the EMR 10 , the OCS 20 , and the PACS 30 . Here, the data collection unit 120 detects various factors corresponding to the type of the corresponding examination from each of the EMR 10 , the OCS 20 , and the PACS 30 .

Specifically, the plurality of independent variables include the type of test (A), the reading difficulty (B), the number of currently unread remaining waiting cases (C), and the average reading time required for the test within the latest set period (D). , the number of currently available specialists (specialists) (E), and the comprehensive reading competency index (F) of available specialists (specialists) may be included.

Here, the variables 'A' are collected from the OCS 20, the variables 'B' and 'C' are collected from the PACS 30, and the variables 'D', 'E' and 'F' are collected from the EMR 10 . Each can be collected. Each variable is described in detail as follows.

First, the type of the examination (A) refers to the type of radiology examination, and may correspond to CT, MRI, ultrasound, or the like. Of course, since each of the CT scan, the MRI scan, and the ultrasound scan is again classified as a more detailed test item, in an embodiment of the present invention, the type of radiology test may mean a detailed test item within each test group.

Next, the reading difficulty (B) of the test can be divided into a plurality of levels (scores). As a simple example, 'easy' may be divided into 1, 'moderate' may be divided into 2, and 'hard' may be divided into 3.

In addition, the number of currently unread remaining waiting cases C of the corresponding examination may mean the number of remaining unread examinations in a sub-field corresponding to the corresponding examination (modality).

The average reading time required (D) corresponds to the average of the most recent reading times of the test. can be expressed as the average value of all

The comprehensive reading competency index (F) of available experts (specialists) is a comprehensive conversion of the capabilities of the specialists in response to the relevant examination in the hospital. may correspond to the sum of all Here, an individual's reading competency index can be determined by the recent (recently within the same period) reading performance (number) of the specialist, the year of the specialist, position, history of misjudgment, history of omission of reading, etc.

The reason why the above-mentioned kinds of independent variables are considered in the prediction of the required reading time will be described in detail as follows.

The reading time required for a radiology examination is the 'reading waiting time' from the time the image is taken until the radiologist starts reading the actual examination (modality), and the 'reading time', which is the time required for the actual reading (image reading and analysis). It is composed of the sum of '.

Assuming that a specific modality has been performed, this examination is classified and submitted to the reading room of the sub-area (abdomen, chest/heart, nerve head and neck, musculoskeletal, etc.) corresponding to the examination within the radiology department, and each reading room Each examination is read by judging the priority of examination (ward/emergency room examination, examination performed first).

In this process, the number of examinations currently assigned and waiting for the relevant part, the number of radiologists, and the reading competency of those specialists will affect the decrease or increase in the 'reading waiting time'.

After the reading waiting time, the radiologist selects a specific image and starts reading it, and the time until the final reading is provided is the 'reading time'. This has an effect on the type of modality (eg, CT, MRI) and the difficulty of reading, and exams with difficulty in reading require more reading time than other exams.

However, in general, assuming a general difficulty level, certain tests have an average reading time that is predictable to some extent, and this number can also be used as a reference for predicting the reading time. Based on this basis, it can be seen that the above-described plurality of independent variables (A to E) are factors that have a very important influence on the prediction of the reading time of the radiology examination.

In addition, each of these factors corresponds to a mutually independent variable, and the prediction of the reading time for the test can be determined according to the degree to which each of these independent variables affects the reading time, and the weight of each independent variable is Each is related to the regression coefficients used for training the machine learning model.

The prediction unit 130 applies the data related to the plurality of independent variables extracted from the data collection unit 120 to the pre-trained machine learning model to predict the required reading time (RTAT) for the corresponding type of test.

At this time, the prediction unit 130 predicts the required reading time for the corresponding test based on the multi-value linear regression analysis of the machine learning model.

Of course, in order to perform such a prediction, a process of pre-training a machine learning model is involved.

To this end, the learning unit 150 learns a machine learning model based on a plurality of independent variables previously collected in response to a past same type (same type) examination and data regarding an actual reading time of the corresponding examination.

In this case, the actual reading time required for the corresponding examination means the time actually consumed (required) from the time when the corresponding radiology examination is executed until the image reading by a specialist comes out.

The learning unit 150 learns the machine learning model according to Equation 1 below.

where i is the index of the independent variable, p is the type (number) of the independent variable used, β _i is the i-th regression coefficient applied to the i-th independent variable, ε is the reading required time (Y), the predicted dependent variable, and , represents the error correction coefficient between the reading time (actual reading time required) in an actual clinical situation. This ε is an error/residual, which is a correction value for correcting the difference between the predicted dependent variable Y and the actual reading required time.

The learning unit 150 performs machine learning analysis on the relationship between a plurality of independent variables (X _i ) collected in response to the same type of test in the past and the required reading time (Y) predicted corresponding thereto, and performs a regression coefficient ( β _i ) is estimated.

The regression coefficient (β _i ) uses the Ordinary Least Sqare (OLS) and simultaneous input method (Enter) to determine the relationship between the pre-input/stored independent variable (X _i ) and the time taken to read the radiology examination (Y). Therefore, it can be estimated based on machine learning.

In this way, the learning unit 150 learns the machine learning model according to Equation 1 by using the independent variable collected in response to the corresponding test in the past and the actual reading required time corresponding thereto. The learning unit 150 may finally determine a regression coefficient value by adjusting and repeatedly learning each regression coefficient value until the error ε is equal to or less than a set threshold.

In addition, the prediction unit 130 may easily derive the required reading time (RTAT) for the radiology test, which is the current prediction target, through Equation 1 based on each determined regression coefficient (β _i ).

That is, the prediction unit 130 quickly and easily derives the required reading time (Y) for the corresponding radiology examination simply by inputting the data (X _i ) on the plurality of independent variables currently extracted in Equation 1 on which the learning is completed. can do it

Here, the Y value derived from Equation 1 may be in the form of time data. For example, it may be derived in the form of “○hours ○minutes”, “○days ○hours ○minutes”, and the like.

The output unit 140 outputs and provides the derived reading required time Y on the user terminal 200, and may provide it on the EMR/OCS screen as shown in FIG. 3 to be described later. Also, based on the predicted reading time, doctors, nurses, etc. provide information on the patient's outpatient treatment date (notification) for confirmation of test results and consultation, appointment of treatment, schedule for surgery/procedure, treatment method, etc. of the patient. It can establish and inform treatment plans more effectively.

As described above, the present invention flexibly predicts the reading time of the radiology examination according to the current clinical situation and enables the establishment of treatment and treatment plans based on this, so that unnecessary visits due to inaccurate notification and complaints from patients , it is possible to prevent problems such as a decrease in the reliability of the hospital in advance.

Meanwhile, the storage unit 160 may classify, store, and manage the data on a plurality of independent variables collected in response to the corresponding test and the collected time by test type. Of course, the storage unit 160 may match and store the type of the selected radiology examination, the data of independent variables collected corresponding to it, and the reading time result predicted by Equation 1, and utilize this as additional learning data. can

That is, the learning unit 150 calculates the reading required time predicted by Equation 1 in response to the current examination from among the information stored in the storage unit 160 and the reading required time actually observed for the corresponding examination later as additional learning data. can be used to readjust and update each regression coefficient in Equation (1). Of course, for this purpose, the input unit 110 may receive the reading required time actually observed for the test from the user terminal 200 , and the storage unit 160 may match it with related information and store it.

In addition, the learning unit 150 may derive different prediction equations for each type of examination by learning between tests of the same type. Accordingly, the values of the regression coefficients applied to the prediction equation of Equation 1 may vary for each type of examination.

On the other hand, the learning unit 150 and the prediction unit 130 may be included or embedded in the processor as shown in FIG. 2 . Here, the processor embeds an artificial intelligence algorithm and performs an operation based on machine learning. As the artificial intelligence algorithm, an artificial neural network having a structure including an input layer, a hidden layer, and an output layer may be used.

The processor may use a machine learning algorithm based on a highly multi-layered artificial neural network for prior learning of the algorithm. In this way, the processor may predict the reading required time that is expected to be closest to the actual test for the currently predicted test based on each independent variable learned in advance through the artificial intelligence algorithm. The predicted data is stored in the storage unit 160 and can be reused as learning data for predicting the reading duration together with the actual reading duration of the corresponding test, and used as a basis for securing variables for error/residual can be

According to the present invention, electronic medical record (EMR), prescription delivery system (OCS), medical image storage and transmission system (PACS) Through a machine learning algorithm based on multiple linear regression based on multiple linear regression-based machine learning algorithms, various independent variables influencing the reading time of radiology examinations acquired from can be easily derived.

3 is a diagram illustrating a state in which a prediction result of a radiology examination reading required time is provided on an EMR/OCR linkage screen according to an embodiment of the present invention.

In the case of FIG. 3, the execution screen of the hospital medical information system in a form in which EMR and OCR are linked (combined) is exemplified. Here, when 'Abdomen (kidney) MRI' is selected from the user terminal 200 from the list of radiology examinations listed on the screen, the prediction apparatus 100 predicts the required reading time (RTAT) for the corresponding examination and predicts the result ( 'The required reading time (RTAT) for the test is 26 hours and 16 minutes [2 hours and 16 minutes per day].') is immediately displayed as a pop-up, enabling intuitive confirmation.

FIG. 4 is a view for explaining a method of predicting a reading required time of an imaging examination using the apparatus of FIG. 1 .

First, the learning unit 150 uses a processor based on an artificial intelligence algorithm to convert data about a plurality of independent variables collected in response to the corresponding radiology examination in the past and the actual reading time of the corresponding radiology examination as a learning data set. Using, the machine learning model according to Equation 1 is learned (S410).

The learning unit 150 estimates each of the regression coefficients applied to each independent variable through the machine learning process for Equation 1, and repeatedly learns and estimates until the residual falls within an error range. In this case, the more historical data used, the higher the reliability of the model. The learning unit 150 may build an individual learning model for each type of imaging examination (CT, MRI, ultrasound, etc.).

After the learning is completed, in response to the type of radiology examination designated by the user, the prediction value of the required reading time for the type of imaging examination may be fed back.

Specifically, for this purpose, the input unit 110 receives a selection from the user terminal 200 of the type of imaging examination to be predicted as shown in FIG. 3 ( S420 ).

Then, the data collection unit 120 extracts data for a plurality of independent variables necessary for predicting the reading time of the corresponding type of radiology examination from the EMR, OCW, and PACS ( S430 ).

Here, the independent variables are the type of test, the reading difficulty, the number of currently unread remaining waiting cases, the average reading time of the test within the latest set period, the number of currently available experts, and comprehensive reading of available experts Competency index, etc. may be included.

The prediction unit 130 inputs the extracted data on the plurality of independent variables to the machine learning model according to Equation 1 (S440), and obtains an output value (Y) corresponding thereto. The required reading time (Y) for the inspection is predicted (S450).

Thereafter, the output unit 140 outputs the predicted reading required time through the user terminal 200 (S460). In this case, the prediction result of the required reading time may be provided in the form of a message, text, or the like as shown in FIG. 3 , or may be provided in the form of an image or the like.

As described above, according to the present invention, it is possible to predict the reading time (RTAT) of the corresponding radiology examination to be predicted in connection with the electronic medical record (EMR), the prescription delivery system (OCS), and the medical image storage and transmission system (PACS). It is possible to effectively predict and provide the RTAT of the radiology examination by extracting the data of various independent variables for the purpose and applying it to a pre-trained machine learning model.

Therefore, the present invention can effectively predict and provide the reading required time that is expected to be required to read the radiology examination in consideration of the current clinical situation of the hospital based on the learning model.

In addition, through a machine learning algorithm based on multiple linear regression, it is possible to flexibly derive the reading required time that is considered to be the closest to the actual reading in the current clinical situation according to the field situation.

In addition, it enables medical staff to more efficiently establish future diagnosis and treatment plans for the patient (outpatient treatment date selection, treatment reservation, surgery/procedure plan, etc.) based on the predicted reading time.

Although the present invention has been described with reference to the embodiment shown in the drawings, which is only exemplary, those of ordinary skill in the art will understand that various modifications and equivalent other embodiments are possible therefrom. Accordingly, the true technical protection scope of the present invention should be determined by the technical spirit of the appended claims.

10: EMR 20: OCS
30: PACS 100: Reading time prediction device
110: input unit 120: data collection unit
130: prediction unit 140: output unit
150: learning unit 160: storage unit
200: user terminal

Claims (12)

An apparatus for predicting a report turnaround time (RTAT) of a radiology examination in a clinical situation, the apparatus comprising:
A learning unit that learns a machine learning model based on multiple linear regression analysis by using multiple independent variables collected in response to a similar test in the past and data on the actual reading time of the test:
an input unit for selecting a type of imaging examination to be predicted;
a data collection unit for extracting data on a plurality of independent variables necessary for predicting a reading required time of the selected type of radiology examination from a hospital medical information system;
a predictor for predicting a reading time required for a corresponding type of test by applying the extracted data on the plurality of independent variables to the previously built machine learning model; and
and an output unit for outputting the predicted required reading time through a user terminal,
The plurality of independent variables are
Includes the type of test, the difficulty of reading, the number of remaining unread cases, the average reading time required for the test within the latest set period, the number of currently available experts, and a comprehensive reading competency index for available experts reading time prediction device. The method according to claim 1,
The hospital medical information system,
An electronic medical record system (EMR), a prescription delivery system (OCS), and a medical image storage and transmission system (PACS) are linked information systems to predict the reading time required. delete delete The method according to claim 1,
The learning unit,
It is a method of estimating each regression coefficient (β _i ) of the following equation by machine learning analysis of the relationship between a plurality of independent variables (X _i ) collected in response to the same type of examination in the past and the correspondingly predicted reading time (Y). A device for predicting reading time for radiology examinations.

Here, i is the index of the independent variable, β _i is the i-th regression coefficient applied to the i-th independent variable, and ε is the error correction coefficient between the predicted reading time Y and the actual reading required time. The method according to claim 1,
The apparatus for predicting the required reading time of a radiology examination further comprising a storage unit for storing the type of the selected radiology examination, a plurality of independent variables extracted corresponding thereto, and the predicted reading required time by matching with each other. In a method for predicting read turnaround time using a device for predicting report turnaround time (RTAT) of a radiology test in a clinical situation, the method comprising:
Learning a machine learning model based on multiple linear regression analysis by using a plurality of independent variables collected in response to a similar test in the past and data on an actual reading time of the test;
receiving a selection of a type of imaging examination to be predicted;
extracting data on a plurality of independent variables necessary for predicting a reading required for the selected type of radiology examination from a hospital medical information system;
estimating the required reading time for a corresponding type of test by applying the extracted data on the plurality of independent variables to the previously built machine learning model; and
outputting the predicted reading required time through a user terminal,
The plurality of independent variables are
Includes the type of test, the difficulty of reading, the number of currently unread remaining waiting cases, the average reading time required for the test within the latest set period, the number of currently available experts, and a comprehensive reading competency index for available experts A method of estimating the required reading time. 8. The method of claim 7,
The hospital medical information system,
An electronic medical record system (EMR), prescription delivery system (OCS), and a medical image storage and transmission system (PACS) are linked information system, a method of predicting the reading time required. delete delete 8. The method of claim 7,
The learning step is
It is a method of estimating each regression coefficient (β _i ) of the following equation by machine learning analysis of the relationship between a plurality of independent variables (X _i ) collected in response to the same type of examination in the past and the correspondingly predicted reading time (Y). How to estimate the reading time:

Here, i is the index of the independent variable, β _i is the i-th regression coefficient applied to the i-th independent variable, and ε is the error correction coefficient between the predicted reading time Y and the actual reading required time. 8. The method of claim 7,
The method further comprising the step of matching and storing the type of the selected radiology examination, a plurality of independent variables extracted corresponding thereto, and the predicted reading required time.

Images