Number
Numerical value
Bounding Box
Image
The “Digital Number Dataset Construction Project” aims to collect and process data from four types of digital devices: thermometers, scales, blood pressure monitors, and blood glucose meters. The project's goal is to gather AI training data that enables the development of a model capable of recognizing these digital devices and interpreting the information displayed on them.
Datumo provides free, high-quality training data for smarter artificial intelligence. This dataset was constructed as part of an “AI Dataset Sponsorship Program” hosted by Datumo, in collaboration with TILDE. TILDE is a startup made up of members who aspire to create an 'endless wave of medical data.' Their goal is to offer technology that enables easier and more efficient utilization of healthcare data. They provide a variety of service platforms to hospitals, businesses, and the general public, all with the purpose of fostering a healthier society.
One of the challenges of this project was to collect as many images from various devices as possible. The user demographics of the Cash Mission were diverse in terms of age, gender, and region, so we could collect images from various devices in a short period of time. In particular, we successfully collected data from less popular devices like blood glucose meters and blood pressure monitors. We believe this project prominently highlighted the advantages of Cash Mission.
"The cornerstone of telemedicine is a system that can continuously monitor the patient's condition by managing data effectively. However, it's always been a challenge to consistently collect data from devices like blood pressure monitors, glucometers, scales, and thermometers, even though they're crucial for telemedicine.
Addressing this, TILDE and Datumo have been developing a technology capable of converting image data from portable medical devices into digital data anytime, anywhere. To support this, we've built an image dataset for blood pressure monitors, glucometers, scales, and thermometers.
This unique dataset we've created is quite valuable and hard to come by. It's set to be a key component for the artificial intelligence technologies that will drive telemedicine forward."
- 1,598 thermometer images and corresponding json files
- 1,640 scale images and corresponding json files
- 2,512 glucometer images and corresponding json files
- 14,500 blood pressure monitor images and corresponding json files
In this project, Datumo's crowd-sourcing platform, Cash Mission, was used for all image collection and numerical data processing. With the participation of about 6,000 users, we were able to quickly collect and process images and numerical data from a wide variety of devices.
셀Crowd workers on Datumo's crowd-sourcing platform 'Cash Mission(App)' directly participated in tasks such as Collect Digital Numeric Display Devices, Inspect Digital Numeric Display Devices, and Medical Device Number Labeling + OCR Missio' to collect and process part of the data.
A guide created by the specialist guide team on 'Cash Mission(Web)' to assist crowd workers in understanding their missions
Sample Data
- Thermometer
{"category": "Thermometer", "Temperature": "36.2", "tem_other1": null, "tem_other2": null, "tem_other3": null}- Scale
{"category": "Scale", "Weight": "64.0", "wei_other1": null, "wei_other2": null, "wei_other3": null}- Blood_glucose
{"category": "Blood_glucose", "Glucose": "79", "date": null, "hour": "07:42:00", "temperature": null, "Day-AVG": null, "other1": null, "other2": null, "other3": null}- Sphygmomanometer
{"coordinates": {"x1": 0.3682997904479176, "y1": 0.18161180476730987, "x2": 0.9480834715794988, "y2": 0.7707150964812713},
"category": "sphygmomanometer", "sys": "101", "dia": "63", "pulse": "77",
"hour": "14:43:00", "average": null, "other2": null, "other3": "135"}- Customized personal medical database using OCR technology
- Service for recording blood pressure/blood sugar/body temperature for health management
CC BY-SA
Reusers are allowed to distribute, remix, adapt, and build upon the material in any medium or format, even commercially, so long as attribution is given to the creator. If you remix, adapt, or build upon the material, you must license the modified material under identical terms.
https://creativecommons.org/licenses/by-sa/3.0/deed.en