Gravio Blog
April 29, 2022

[Case Study] Kodama Foundation Inc, Shinsei Hospital

Automatic detection of unauthorised patients going out or wandering around the hospital by integrated use of face recognition camera AI and IoT sensors. The monitoring system provides security and peace of mind to both patients and hospital staff.
[Case Study] Kodama Foundation Inc, Shinsei Hospital

Background

Kagoshima City's Shinsei Hospital, a general incorporated foundation of the Kodama Hoso-kai, introduced Gravio Enterprise AI Edition to prevent various problems caused by inpatients suffering from dementia or delirious patients after surgery from wandering around the hospital and going out without permission. AI facial recognition through cameras can now automatically detect wandering patients and going out without permission. In addition, a motion sensor provided by Gravio was introduced to strengthen monitoring of wandering in the hospital at night, a door open/close sensor was introduced to detect the opening and closing of hospital room doors, and a CO₂ sensor was introduced to provide room ventilation at the appropriate time.

Challenges

  • Hospital staff have to manually prevent hospitalised patients from going out without permission or wandering around the hospital.
  • RFID-based behaviour monitoring solutions were difficult to introduce due to their high cost and labour.
  • A low-cost solution that could be implemented by small and medium-sized medical institutions was required.
  • A solution that could be built without programming skills was required.
  • When staff members enter the hospital at night, it takes a long time to enter the hospital due to the time-consuming process of identification and unlocking the door at the staff station.

Solution

  • Introducing the Gravio Enterprise AI Edition, a facial recognition AI and IoT platform.
  • The face recognition camera AI automatically detects patients who pass in front of the cameras installed in the hospital.
  • Patient detection is notified to Patlite installed at staff stations and notified nurses' through LINE messaging application.
  • Additional Gravio motion sensors, door open/close sensors, and CO₂ sensors were also installed.
  • Linked to electronic locks, the system uses a facial recognition camera AI to unlock the locks.

Effects

  • Unauthorised patients leaving or wandering in the hospital can now be automatically detected by camera AI and IoT sensors, significantly reducing the workload on staff.
  • Provides patients with a safe and comfortable space by measuring CO₂ concentration in waiting rooms, which tend to be dense, and indicating the location where patients and staff can see it, enabling ventilation at the appropriate time.
  • Low cost and short time to implement the solution.
  • Enables solution construction with software configuration and adjustment work without programming.
  • Smooth entry into the building at night by eliminating the need to confirm entry into the building.

Overview

Interview With User

Could you give us a brief description of Shinsei Hospital's business?

Kumagai: Our hospital is a care-mix hospital located in Kagoshima City, and we are able to meet a wide range of medical needs in the community, from regular outpatient to inpatient care. It has a general ward with 28 beds and a convalescent ward with 37 beds. Unusually for a hospital of this size, it is equipped with surgical facilities and doctors who can perform surgical procedures, making it a one-stop provider of all types of patient care. I served as director of the hospital until July 2020, and since leaving that position, I have been involved in many negotiations with medical associations and other organisations outside the hospital.

How did you introduce a "wandering detection solution" using Gravio Enterprise AI Edition with Face Recognition Camera AI?

Kumagai: We have many inpatients with dementia and patients admitted for surgical procedures, and some patients with dementia or in a delirious state after surgery try to leave the hospital or wander around the hospital. Although the rule is to apply to the nurse for permission before going out for safety reasons, in reality there are many patients who try to go out without permission, and I believe that many hospitals, not only our hospital, are suffering from the same problem. We consulted with Densetsu Kogyo to see if there was a good way to reduce these unauthorised absences.

Yanagimoto: Like Shinsei Hospital, Densetsu Kogyo is based in Kagoshima City and is a one-stop provider of electrical equipment and security camera installation and maintenance services to customers in the Kyushu and Okinawa areas. When Shinsei Hospital first approached us about this project, we wondered if we could track and detect the activities of patients by attaching RFID tags (IC tags) to their slippers, pyjamas, and other items they wear. However, we came to the conclusion that this was not very realistic considering the cost and the man-hours required to manage the RFID tags. After much consideration, we finally came up with a face recognition AI solution that utilises cameras.

There are many camera-based facial recognition AI solutions out there, what made you choose Gravio Enterprise AI Edition?

Yanagimoto: At first, we compared more than 10 solutions and finally narrowed it down to three. The primary reason we ultimately chose Gravio Enterprise AI Edition was that it allows us to process patient image data captured by cameras only in the edge computer, without sending it to the cloud. Patient images are personal information, so we thought that Gravio Enterprise AI Edition would be preferable because it completes the AI image inference process inside the edge computer and does not transmit the image data outside the computer.

Was "ease of implementation" a consideration?

Yanagimoto: AI face recognition solutions from major manufacturers are extremely expensive, and it is very difficult for small and medium-sized facilities and medical institutions to implement them. We were also attracted by the fact that instead of building a system from scratch, we could use pre-trained image inference models and preset programs to quickly and efficiently build the system. Furthermore, we do not have many people with programming skills, but Gravio Enterprise AI Edition does not require programming and can be built with software configuration and adjustment.

What was your impression of the company Asteria?

Yanagimoto: Actually, we had considered another small venture company's product as a candidate for selection, but venture companies are risky in terms of business continuity. In this respect, Asteria has a wealth of experience and a high degree of trust. We also have a wealth of expertise in security camera installation and network design and construction, so we decided that Gravio Enterprise AI Edition would be a solution that could take advantage of our strengths.

When did you start working on Gravio Enterprise AI Edition?

Yanagimoto: First, we conducted a technical verification within our company in August 2021, and after the Shinsei Hospital actually confirmed the results, they gave us the go-ahead for full-scale implementation. Then, in September 2021, we installed the cameras and Gravio-related devices in two separate installations and network construction work, and the system began full-scale operation at the end of September. First, after registering the facial photos of patients who might wander, we installed cameras at the main entrance and service entrance, and constructed a system in which the face recognition camera AI can detect when a patient passes through these locations.

In addition, Gravio motion sensors were installed at two stairwells in the hospital to automatically detect whether patients are wandering around the hospital at night.

Another attractive feature of Gravio as a wandering detection method is that it can use a combination of face recognition camera AI and IoT sensors in the right places at the right time.

Wandering Detection Solution - AI and IoT can be combined in the right place at the right time

How do you inform the hospital staff about the detection results?

Kumagai: When the face recognition camera AI detects a patient, the patrol lights installed at staff stations flash and emit a warning sound, and a notification message and a photo of the wandering patient captured by the camera are instantly displayed on the nurses' smartphone LINE app. With this system in place, nurses and other hospital staff no longer have to worry about wandering patients with dementia going out without permission. Before, we had to be constantly on the lookout for wandering patients, but now that we have this solution in place, we are alerted by a patrol light if something goes wrong, which contributes to the efficiency of our operations.

Have you successfully detected a wandering patient with this system?

Kumagai: Yes. We still have patients with dementia in the hospital, and when one of them tried to go out without permission during the night, the sensor reacted and we were able to prevent it at an early stage. If we had not introduced this solution, we might not have noticed the unauthorised absence.

Please tell us about any difficulties or challenges you encountered in implementing Gravio Enterprise AI Edition.

Yanagimoto: Gravio Enterprise AI Edition is a no-code design that can be built without programming, so even our staff was able to build it smoothly. The LINE application is already equipped with a function for sending text messages, but we added our own function for attaching a photo of a wandering dementia patient captured by the camera, which we believe further improved usability. In addition, the recognition rate of the AI changes depending on the angle of the face, sunlight, and lighting conditions, so tuning these settings was both interesting and difficult. Incidentally, such tuning is still ongoing.

I understand that you are now installing a third camera not only at the entrance and exit, but also in the waiting room?

Yanagimoto: Yes. We are currently testing and operating a camera AI to detect the number of people in the waiting room, thinking that it would be possible to automatically determine the congestion status of the waiting room.

Kumagai: In addition, Gravio CO₂ sensors and temperature and humidity sensors are installed in the waiting and reception rooms, and the measured values are displayed on the "Gravio LED matrix. This allows staff to quickly notice when the number of people staying in a room increases, and to ventilate the room, thereby contributing to infection control measures. In the future, we are also considering installing Gravio open/close sensors on the doors of patient rooms where there is a risk of wandering, so that doors are automatically detected when they are opened at night. We are currently in the process of installing these open/close sensors in conference rooms to verify their practicality.

A combination of AI and IoT can be used at the same place for specific purposes

So you are not only using camera AI for facial recognition, but also Gravio's various IoT capabilities?

Kumagai: Yes, I believe that by making more effective use of Gravio's functions, we can further improve the quality of services provided to our patients and the operational efficiency of our staff. I also believe that a system such as Gravio's will be effective not only in a small medical institution like our hospital, but also in a large hospital where many staff members and patients come and go. For example, if nighttime access for hospital staff and management of access to each facility within the hospital could be done using Gravio Enterprise AI Edition's facial recognition instead of ID cards and passwords, it would be a very simple and strong security measure. We have already asked Densetsu to enable emergency doctors and technicians to automatically open and enter the building at night using facial recognition. Since the time of entry is also recorded, we are thinking that in the future we can register all staff members and use this system for "work style reform”.

Yanagimoto: We also feel that Gravio has great potential, and in November 2021, we concluded a distributor agreement with Asteria. With this agreement, we would like to develop a wandering detection solution based on the "wandering detection solution" introduced at Shinsei Hospital that can also be introduced at small- and medium-sized medical institutions and serviced residences in the Kyushu area, using facial recognition based on camera AI. We also believe that the 3-dense detection solution using CO₂ sensors that we introduced for Shinsei Hospital will be useful for a variety of customers, including hospitals.

Overall Solution Diagram

Messages From The People In Charge

Kodama Foundation Inc, Shinsei Hospital, Former Director, Teruo Kumagai, Director of Kagoshima City Medical Association

The introduction of the Gravio Enterprise AI Edition with Face Recognition Camera AI not only provides safety and security to patients, but also contributes to operational efficiency as nurses no longer have to pay more attention to unauthorised wandering dementia patients than before. In addition, doctors and laboratory technicians who are called in suddenly at night are able to open the door and enter the building automatically by facial recognition. This will also record the time they arrive at the office, and we are considering registering all staff members so that we can utilise this system for "work style reform" in the future. We believe that we can further improve the quality of services provided to patients by effectively utilising other functions of Gravio, such as CO₂ sensors.

System Engineer, Densetsu Kogyo Co, Mr. Takahisa Yanagimoto, Gravio Distributor

Gravio's no-code tool allows us to build the system without programming and with software configuration and adjustment work. We have always had a wealth of expertise in security camera installation and network design and construction, so in that sense, I think this solution allows us to take advantage of our strengths. In the future, we would like to expand our services to small and medium-sized medical institutions and serviced residences in the Kyushu area.

Get in touch
Latest Posts
[Case Study] Kodama Foundation Inc, Shinsei Hospital
Automatic detection of unauthorised patients going out or wandering around the hospital by integrated use of face recognition camera AI and IoT sensors. The monitoring system provides security and peace of mind to both patients and hospital staff.
Friday, May 13, 2022
Read More
The Four Ds: The Future of Business IoT
The future is 4D - How businesses thrive in the future applying the 4D principle
Monday, April 18, 2022
Read More