TW202227033A - Medicine cabinet system with image recognition training module capable of reducing probability of false judgment by human eyes and increasing accuracy for medicine picking - Google Patents

Abstract

The present invention includes a processing portion, a plurality of medicine chests, a medical order database, an image capture device, a medicine image database, and an image recognition training module. The plurality of medicine chests are respectively placed with medicine and respectively connected to the processing portion. The medical order database is provided with a plurality of medical order codes. The image recognition training module may obtain the medicine images from different angles of medicine in advance and store the images into the medicine image database with corresponding medicine codes. When corresponding medical order code is input to the processing portion, the processing portion controls the corresponding medicine chest to open and take out medicine and the image capture device captures the image of medicine, so that the medicine may be picked up after double checking of medicine images and medical order codes. Thus, the present invention utilizes the image recognition training module to enhance the recognition accuracy, reduce the probability of false judgment by human eyes through image recognition for medicine picking, increase the accuracy of medicine picking after double checking through medical order database, and manually check again to successfully complete the medicine picking operation.

Description

Medicine cabinet system with image recognition training module

The invention relates to a medicine cabinet system with an image recognition training module, in particular to a medicine cabinet system with an image recognition training module, which can improve the recognition accuracy, reduce the probability of misjudgment by the human eye after taking medicine through image recognition, and a doctor's order database 2 Sub-comparison improves the accuracy of taking medicine, and manual re-checking can successfully complete medicine cabinet system with image recognition training module.

The traditional medicine box system (or called smart medicine cabinet) usually has a plurality of medicine boxes, and each medicine box is connected to a control device and can be opened or locked independently. After receiving a prescription doctor's order, input the prescription doctor's order to the control device, the prescription doctor's order (it can also be said that the control device can search for the corresponding medicine box) find the corresponding medicine box, and open the specific medicine box for taking medicine. However, after taking out the medicine, the right or wrong of the medicine (if it is necessary to take two medicines and one bottle of one medicine at a time, but due to busyness or negligence, two bottles of the same medicine are mistakenly taken) and the quantity are judged by the human eye. It is checked by the controller when the shift is over. In this way, after taking the medicine, it may be directly injected or given to the patient, and it may be too late by the time of counting. In view of this, it is necessary to develop a technology that can solve the above-mentioned conventional shortcomings.

The purpose of the present invention is to provide a medicine cabinet system with an image recognition training module, which also has an image recognition training module that can improve the recognition accuracy, that can reduce the probability of misjudgment by the human eye after taking medicine through image recognition, and a database of medical orders. The secondary comparison improves the accuracy of medicine taking, and the manual re-checking can successfully complete the medicine taking and so on. In particular, the problem to be solved by the present invention is that so far, there is no system that can reduce the misjudgment rate of the human eye and improve the accuracy of medicine taking by cooperating with the doctor's order system to check. The technical means to solve the above problems is to provide a medicine cabinet system with an image recognition training module, which includes: a processing part, which has an input interface and a display part; A plurality of medicine boxes are respectively used for being arranged in a medicine cabinet, each medicine box is used for placing at least one medicine, and is respectively connected to the processing part; A doctor's order database, which is linked to the processing unit, and has plural doctor's order codes and corresponding plural drug details; one of the plural doctor's order codes and the corresponding plural drug details corresponds to at least one drug; an image capture device connected to the processing unit; A drug image database is connected to the processing unit and the image capture device, and has plural drug codes and corresponding plural drug images, and one of the plural drug codes and the corresponding plural drug images is for There should be at least one drug; An image recognition training module is set corresponding to the image capturing device, and the image recognition training module has a rolling part, a slope part and a collecting part; the rolling part is the medicine, at least one transparent roller One of the medicines is built in and fixed; Thereby, the rolling part can roll on the slope part to reach the collecting part; the image capturing device is used for recording video towards the slope part during the rolling process, so as to establish the plurality of medicines in the medicine image database In the image, when the medical order code is input to the processing unit through the input interface, the processing unit compares the input medical order code with the medical order database, and when it matches one of the plural medical order codes in the medical order database, It controls the opening of the corresponding medicine box for taking out the corresponding medicine, and for the image capture device to capture an image and input it to the processing unit, and the processing unit firstly compares the input image and the medicine image The multiple drug images in the database are compared, which is one of the multiple drug codes and the corresponding multiple drug images; then, the processing unit compares the image with the doctor order database. If one of the details of the medicines in the doctor's order code should be multiple, the medicine will be taken. If there is no correspondence, the medicine will be taken after the manual comparison is correct. The above objects and advantages of the present invention can be easily understood from the detailed description and accompanying drawings of the following selected embodiments. Hereinafter, the present invention will be described in detail with the following examples and accompanying drawings:

Referring to Figures 1A, 1B, 2, 3A, 3B, 4A, 4B and 5, the present invention is a medicine cabinet system with an image recognition training module. Medicine cabinet system, including: A processing unit 10 has an input interface 11 and a display unit 12 . The plurality of medicine boxes 20 are respectively used to be arranged in a medicine cabinet 91 . Each of the medicine boxes 20 is used for placing at least one medicine 92 and is connected to the processing part 10 respectively. A medical order database 30 is connected to the processing unit 10 and has plural medical order codes and corresponding plural drug details; one of the plural medical order codes and the corresponding plural drug details corresponds to at least one drug 92 . An image capturing device 40 is connected to the processing unit 10 . A drug image database 50 is connected to the processing unit 10 and the image capture device 40, and has plural drug codes and corresponding plural drug images, and one of the plural drug codes and the corresponding plural drug images , which corresponds to at least one drug92. An image recognition training module 60 is provided corresponding to the image capturing device 40, and the image recognition training module 60 has a rolling part, a slope part 61 and a collecting part 62; the rolling part is the medicine 92 (refer to FIG. 3A ), and at least one transparent roller M is built-in and fixed to one of the medicines 92 (refer to FIG. 4A ). In this way, the rolling part can roll (at least one circle) on the slope part 61 to reach the collecting part 62; the image capturing device 40 is used for recording video towards the slope part 61 during the rolling process to establish the For the plurality of drug images in the drug image database 50, when a doctor's order code is input to the processing unit 10 through the input interface 11 (actually, it may be an action of scanning a prescription signature), the processing unit 10 compares the inputted doctor's order code with the doctor's order code. The doctor's order database 30 is compared, and when one of the plurality of doctor's order codes in the doctor's order database 30 is matched, the corresponding medicine box 20 is controlled to be opened for taking out the corresponding medicine 92 and supplying the The image capturing device 40 (eg, Fig. 6A, Fig. 6B, Fig. 7A, and Fig. 7B) captures an image 93 and inputs it to the processing unit 10. The processing unit 10 first combines the inputted image 93 with the drug image database 50 of the plural drug images are compared, which is one of the plural drug codes and the corresponding plural drug images. Next, the processing unit 10 compares the image 93 with the doctor's order database 30. If it corresponds to one of the medicine details of the multiple doctor's order codes, the medicine is taken. If there is no correspondence, another manual comparison is made. After correct, complete the medicine. In practice, the processing unit 10 can be a drug management computer of a related medical institution. The input interface 11 can be at least one of a mouse, a keyboard, a scanner, a voice control input device, and a touch input device. The display portion 12 can be a screen. The image capturing device 40 can be at least one of a video recorder, a video camera, and a camera. The image capturing device 40 can capture images of the scrolling portion (eg, as shown in Figures 1A, 3A, and 4A) during the scrolling process, or capture images of the medicine 92 taken out from the medicine box 20 . The drug image database 50 may have a deep learning module (which may be a neural network-like learning module, or a related artificial intelligence learning module, which are known technologies, and will not be described in detail), and may have the following Two identification modes: [a] The first identification mode: the image capturing device 40 records a video toward the slope portion 61 to obtain a plurality of images, and transmits the plurality of images to the drug image database 50 , and the drug image database 50 passes through the depth The learning module identifies the medicine 92 corresponding to the scroll portion, and then organizes a plurality of medicine images, wherein the plurality of medicine images are images corresponding to different (360-degree) angles of the medicine 92 respectively. [b] The second identification mode: the image capturing device 40 takes a picture of the corresponding medicine 92 taken out of the medicine box 20 (eg, the 6A, 6B, 7A, 7B, 8A and 8B pictures) , and transmit the image 93 to the drug image database 50 , and the drug image database 50 compares the corresponding at least one drug image through the deep learning module. Of course, the aforementioned second identification mode is only for enhancing the identification of the medicine 92 taken out from the medicine box 20 , and does not necessarily need to be executed. The slope portion 61 has a slope θ, which is between 5 degrees and 30 degrees. The present invention is applied to the management of drugs in related medical institutions, and is mainly based on controlled drugs. First, establish the doctor's order database 30 for different controlled drugs according to their doctor's order codes and drug details. For example, referring to Figure 2, suppose a controlled drug's doctor's order code is set to 10135, and its corresponding details are A (drug) *1 (stick, tablet), and then assume that a controlled drug is set to have its doctor's order code 10136, and its corresponding detail is B (medicine)*2 (stick, tablet). Further, images from different angles are captured for each stored controlled drug in advance (the image recognition training module can be used for this), and the drug image database 50 is established accordingly. For example, as shown in Figure 5, suppose a controlled drug (for example: bottled injection) is set to its drug code as 0001, and the corresponding images of the drug at various angles (for example: 0~360 degrees, one per degree) Data may be bottle-shaped outlines (at most different labels for different angles). Suppose that a controlled drug (for example: a single lozenge package) is set to its drug code as 00XX, the corresponding image data of the drug from various angles may be the image of the lozenge seen from the front, and the image of the lozenge not seen from the back. As for the process of taking the medicine, inputting the doctor's order code to the processing unit 10 through the input interface 11 may actually be the action of scanning the prescription signature, or directly clicking the doctor's order that the doctor has uploaded to the system. Then the processing unit 10 compares the inputted medical order code with the medical order database 30, and controls the corresponding medicine box 20 to open for taking out the corresponding medical order code when it matches one of the plurality of medical order codes. Medicines 92. The point is that, after taking out the medicine 92, the image 93 must be captured by the image capture device 40 (for example, Figs. 6A, 6B, 7A and 7B) and input to the processing unit 10. The processing unit 10 First, compare the input image 93 with the drug image database 50 (this part can be further compared by using artificial intelligence systems, all kinds of artificial intelligence have been practiced for a long time, so I won't go into details. It corresponds to one of the plural drug codes and the corresponding plural drug images. Moreover, the processing unit 10 must compare the image 93 with the doctor's order database 30. If it corresponds to one of the medicine details of the plural doctor's order codes, the medicine will be taken. If there is no correspondence, the medicine will be taken manually. After the comparison is correct, the medicine is completed. The advantages and effects of the present invention are as follows: [1] Image recognition training module can improve recognition accuracy. The present invention has an image recognition training module that can capture images of the drug at different angles (360 degrees) to establish the drug image database, and can recognize the drug no matter how the drug box 20 is placed. It can reduce the adjustment time of the placement position, and secondly, it can improve the recognition accuracy. Therefore, the image recognition training module can improve the recognition accuracy. [2] Taking medicine through image recognition can reduce the probability of misjudgment by the human eye. The part to be specially explained here is that when ordinary people are busy, their eyes may make misjudgments visually. Taking related medical staff as an example, they may be dazzled and take the wrong medicine for a while while the medical staff are busy. The present invention is aimed at this part, As long as the image of the drug is captured, it can be automatically (artificial intelligence) identification through the drug image database, which greatly reduces the probability of misjudgment by the human eye. Therefore, taking medicine through image recognition can reduce the probability of misjudgment by the human eye. [3] The secondary comparison of the doctor's order database improves the accuracy of taking medicine. Even if it has been identified correctly by the drug image database, the present invention still sets a second checkpoint, that is, if the doctor order database is used for comparison, if it corresponds to one of the drug details of the plural number of doctor order codes, it is desirable to medicine. Therefore, the secondary comparison of the doctor's order database improves the accuracy of taking medicines. [4] The medicine can be successfully completed by manual re-checking. If the doctor's order database cannot be successfully checked for some reason, the final check will be done manually. At this time, since two automatic checks have been passed, the error rate has been greatly reduced. Therefore, the medicine can be successfully collected by manual checking again. The above is only to describe the present invention in detail by means of preferred embodiments, and any simple modifications and changes made to the embodiments do not depart from the spirit and scope of the present invention.

10: Processing Department 11: Input interface 12: Display part 20: Pill Box 30: Database of medical orders 40: Image capture device 50: Drug Image Database          60: Image recognition training module 61: Ramp section 62: Collection Department 91: Medicine cabinet 92: Medicines 93: Video M: transparent roller θ: slope

FIG. 1A is a schematic diagram of the first embodiment of the present invention FIG. 1B is a schematic diagram of the second embodiment of the present invention Figure 2 is a schematic diagram of another aspect of Figure 1B Fig. 3A is a schematic diagram of the first embodiment of the image training module of the present invention Figure 3B is a schematic diagram of the captured image of Figure 3A Fig. 4A is a schematic diagram of the second embodiment of the image training module of the present invention Fig. 4B is a schematic diagram of the captured image of Fig. 4A Fig. 5 is a schematic diagram of the drug image database of the present invention Fig. 6A is a schematic diagram of the first embodiment of the present invention for taking pictures of medicines taken out Fig. 6B is a schematic diagram of the image of Fig. 6A Fig. 7A is a schematic diagram of the second embodiment of the present invention for taking pictures of medicines taken out Figure 7B is a schematic diagram of the image in Figure 7A Fig. 8A is a schematic diagram of the third embodiment of the present invention for taking pictures of medicines taken out Fig. 8B is a schematic diagram of another embodiment of Fig. 8A

10: Processing Department

11: Input interface

12: Display part

20: Pill Box

30: Database of doctor's orders

40: Image capture device

50: Drug Image Database

60: Image recognition training module

61: Ramp

62: Collection Department

91: Medicine Cabinet

92: Medicines

Claims (8)

A medicine cabinet system with an image recognition training module, comprising: a processing part, which has an input interface and a display part; A plurality of medicine boxes are respectively used for being arranged in a medicine cabinet, each medicine box is used for placing at least one medicine, and is respectively connected to the processing part; A doctor's order database, which is linked to the processing unit, and has plural doctor's order codes and corresponding plural drug details; one of the plural doctor's order codes and the corresponding plural drug details corresponds to at least one drug; an image capture device connected to the processing unit; A drug image database is connected to the processing unit and the image capture device, and has plural drug codes and corresponding plural drug images, and one of the plural drug codes and the corresponding plural drug images is for There should be at least one drug; An image recognition training module is set corresponding to the image capturing device, and the image recognition training module has a rolling part, a slope part and a collecting part; the rolling part is the medicine, at least one transparent roller Built-in and fixed one of the medicines; Thereby, the rolling part can roll on the slope part to reach the collecting part; the image capturing device is used for recording video towards the slope part during the rolling process, so as to establish the plurality of medicines in the medicine image database In the image, when the medical order code is input to the processing unit through the input interface, the processing unit compares the input medical order code with the medical order database, and when it matches one of the plural medical order codes in the medical order database, It controls the opening of the corresponding medicine box for taking out the corresponding medicine, and for the image capture device to capture an image and input it to the processing unit, and the processing unit firstly compares the input image and the medicine image The multiple drug images in the database are compared, which is one of the multiple drug codes and the corresponding multiple drug images; then, the processing unit compares the image with the doctor order database. If one of the details of the medicines in the doctor's order code should be multiple, the medicine will be taken. If there is no correspondence, the medicine will be taken after the manual comparison is correct. The medicine cabinet system with an image recognition training module as claimed in claim 1, wherein the slope portion has a slope, which is between 5 degrees and 30 degrees. The medicine cabinet system with an image recognition training module as described in claim 1, wherein: The drug image database has a deep learning module, which is a neural network-like learning module; Thereby, when the image capturing device performs video recording toward the slope portion to obtain plural images, and transmits the plural images to the drug image database, the drug image database recognizes the corresponding drug image through the deep learning module. The medicine in the rolling part is further arranged into a plurality of medicine images, and the plurality of medicine images are images corresponding to different angles of the medicine. The medicine cabinet system with an image recognition training module as described in claim 1, wherein: The drug image database has a deep learning module, which is a neural network-like learning module; Thereby, when the image capturing device takes a picture of the corresponding medicine taken out from the medicine box, and transmits the image to the medicine image database, the medicine image database is compared through the deep learning module. corresponding to the at least one drug image. The medicine cabinet system with an image recognition training module according to claim 1, wherein the processing unit is a medicine management computer. The medicine cabinet system with an image recognition training module according to claim 1, wherein the input interface is at least one of a mouse, a keyboard, a scanner, a voice control input device, and a touch input device. The medicine cabinet system with an image recognition training module according to claim 1, wherein the display part is a screen. The medicine cabinet system with an image recognition training module according to claim 1, wherein the image capture device is at least one of a video recorder, a video camera, and a camera.

Images