This year, there is a growing interest in the birth of AI (artificial intelligence) medical devices developed by Korean companies.
AI medical devices refer to products that analyze, diagnose, and predict medical information collected from patients, such as patient medical records, biometric information, and medical images.
It analyzes the medical information collected and provides the information (quantitative value for the specific site) applied to the diagnosis and treatment of the disease, or automatically diagnoses or predicts the possibility of the presence or absence of the disease, and provides it to the medical staff.
Currently, there are four medical devices in Korea that are commercialized using AI technology.
JLK Inspection, Buno, Lounette and Silicon Sapien are in clinical trials to receive approval for medical devices.
JLK Inspection and Viewo were approved in September 2017, and LoNit and Silicon Sapien were approved for clinical trials in October of the same year.
These companies are validating artificial intelligence diagnostic algorithms based on MRI, CT, and X-ray images through clinical trials.
The JLK Inspection is based on MRI, which is based on the medical imaging analysis software, the Bone Scale X-ray image-based bone age measurement (medical image analysis software) Medical image detection assistant software) and Silicon Sapien is developing coronary artery prediction system (medical imaging diagnostic assistant software) based on coronary CT image.
Artificial intelligence-based infarction diagnostic software, developed by JK Leke Inspection, aims to provide medical staff with information on the causes of cerebral infarction. The company said that it will make diagnosis and proper prescription for each patient. Research is being conducted at the Ilsan Hospital of Dongguk University and the Seoul National University Hospital of Bundang.
Vuno is conducting clinical trials of 'VUNOmed-BoneAge' to measure the bone age of growing children. The accuracy of bone age measurement is evaluated by using X-ray image for growing men and women in Korea University Hospital. In the case of bone age measurement, a lot is done at a university hospital, but it is a field that can not be practiced well at a clinic.
VNO is developing the VUNOmed-BoneAge, which learns bone age measurement data from university hospitals, at the opening of the clinic. In addition, early detection technology for lung disease is under development.
Lunit is developing medical image detection assistant software called 'Lunit sight for Chest Radiography'. The safety and efficacy of pulmonary nodule diagnosis using chest X-ray imaging are evaluated at Seoul National University Hospital and Boramae Hospital.
The company is developing the world's first cloud-based real-time X-ray reading service and opened its 'Louvre Insight' in November last year for simple testing and research purposes. According to the Lunet side, the accuracy of the readings can be improved up to 20%.
Silicon Sapien is developing HeartMedi 1.0, which measures the degree of stenosis through fractionated flow reserve (FFR) measured based on coronary CT images. It was developed by the Kangwon National University Project Team and was founded in 2014 by Silicon Sapien. They are recruiting patients at Seoul National University Hospital in Bundang and conducting clinical trials.
In the case of coronary artery disease, medical personnel perform myocardial fractionation blood flow reserve, an invasive test to determine the degree of stenosis to determine stenting. 'HeartMedi 1.0' is a non-invasive test that uses coronary CT to confirm the degree of stenosis. It is expected that the inconvenience that sufferers with coronary artery disease should suffer.
AI medical devices being developed by Korean companies are being conducted as retrospective clinical trials to confirm the validity of software using existing medical data except for Silicon Sapien.
Retrospective clinical trials are shorter than prospective clinical trials in which patients must be recruited. It is known that it usually takes two to three months for data collection and three months for analysis. In about six months, clinical trials of AI medical devices are concluded.
The Food and Drug Administration will review and give final approval within 85 days of receipt of a medical device license application.
It is expected that the first AI medical device will be born in the first half of this year as soon as J-ELK Inspection and BUENO approved the clinical trial plan in Oct. last month by Lounet and Silicon Sapien.
Of course, due to problems such as how much data can be secured in this process, the duration of clinical trials may be long, and precise timing can not be predicted because supplementary instructions may be issued for insufficient data during the period of review of the drug store. However, there is no doubt that this year's first AI medical device will be born.
So can AI medical devices be used right away in real medical practice?
Industry experts and experts should keep an eye on this.
The KFDA is working to develop guidelines for AI medical device approval screening and clinical trial validation, and Ryu Young-jin, the chief of the KFDA, has said that it will be able to quickly release advanced medical devices into the market through quick approval through New Year's Day. Evaluation, payroll, and practical use in the clinical field.
Choi Yoon-sup, CEO of Digital Healthcare Partners (DHP), a specialist in accelerating the start-up of healthcare, said, "AI medical devices under development can not be used without physicians in the field and assist physicians. Therefore, it should be observed whether it will be introduced to hospitals after permission. It was developed in line with the needs of doctors, but it has to be well integrated into the clinic and the treatment process. "
"We need to demonstrate the value of AI medical devices without interfering with existing practices and procedures. For this purpose, convenience of treatment, clarification of treatment, efficiency of treatment, and safety should be improved. This year, AI medical devices will be licensed and introduced. The need for AI medical devices and the needs of doctors may be different for hospitals. The spread to other hospitals can be expected when the needs of hospitals and doctors are well-matched. "
The number of start-ups to develop AI medical devices in Korea is about 10, and the number of large-scale companies including Samsung Electronics, which declared development, is increased.
According to the Ministry of Food and Drug Administration, no clinical trials have yet been received or approved, except for four sites that have received and approved the current protocol.
Although it is not possible to predict how much it will be used in actual medical field, it is expected that the development of advanced medical devices such as AI medical devices will become active.
Many global companies, including Google, IBM, Microsoft, and Johnson & Johnson, are developing and launching products that combine AI technology. Google Streams, Deep Mind Health, and IBM Watson. We are in the midst of product development with leading brands such as BUNO and ROUNET, which are recognized overseas.
The KFDA is also in the position to actively support product development. The KFDA has already organized a specialist consultation body last year to collect opinions from industry and academia, and announced guidelines on the approval of AI technology-based medical device licensing and guidelines for evaluating clinical trial efficacy. We will continue to do this in the future.
"We already have discussions with the industry and published the Guideline for Guideline for Approval of Medical Device Applicable to AI Technology and Guideline for Validity of Clinical Trial Evaluation. "This year, we are holding a briefing session for related industries and holding a briefing session if requested by companies, medical institutions, educational institutions, academia, etc. This year, I will try hard. "