The medical community is paying attention to whether locally developed artificial intelligence (AI)-based medical devices will be commercialized this year.

AI medical devices refer to products that analyze medical information such as patients’ medical records, body measurements, and medical images to help physicians diagnose or predict a disease.

More than 10 start-up companies are developing AI medical devices in Korea, and four are planning to commercialize their products sooner or later.

JLK Inspection, Vuno, Lunit, and SiliconSapiens are conducting clinical trials to get their products obtain approval as medical devices. JLK Inspection and Vuno got the regulator’s nod for clinical trials in September, and Lunit and SiliconSapiens, in October, respectively.

Through the trials, the four are testing AI diagnosis algorithms, based on magnetic resonance imaging (MRI) scans, computed tomography (CT) scans, and X-ray images.

JLK Inspection is working on software to diagnose cerebral infarction through MRI scans, and Vuno, on software to measure bone age through bone X-ray images. Lunit is developing software to diagnose solitary pulmonary nodule (SPN) through chest X-rays, and SiliconSapiens, a system that predicts a coronary artery disease based on CT scans of the coronary arteries.

JLK Inspection’s AI-based software aims to give physicians information about the cause of the cerebral infarction of a patient, enabling customized diagnosis and prescription. The firm’s clinical trials are taking place at Dongguk University Ilsan Hospital and Seoul National University Bundang Hospital.

Vuno is on a clinical trial called “VUNOmed-BoneAge” to measure the bone age of children in growth. At hospitals under Korea University Medical Center, the study is testing the accuracy of the bone age measurement on men and women growing up, based on X-rays.

Lunit’s study, “Lunit sight for Chest Radiography,” is testing software that aids detection of medical images. At Seoul National University Hospital and Boramae Hospital, the trial is testing the safety and efficacy of the software to diagnose solitary pulmonary nodule (SPN) through chest X-rays.

Lunit said it was developing the world’s first cloud-based, real-time X-ray reading service. For a simple test and research, the company opened an online page “Lunit Insight” in November.

SiliconSapiens is working on “HeartMedi 1.0” to assess the degree of stenosis of a diseased coronary artery through fractional flow reserve (FFR), measured by CT scans on the coronary arteries. Kangwon National University’s project group developed the device, and the group later founded SiliconSapiens in 2014. The company recruited patients at Seoul National University Bundang Hospital for the clinical trial.

On a patient suffering from a coronary artery disease, physicians conduct FFR, an invasive technique, to learn how severe the patient’s stenosis is and to determine the need for stenting. “HeartMedi 1.0” is a non-invasive inspection, using CT scans on the coronary arteries to check the severity of the stenosis. The company expects the technology will reduce inconvenience for coronary artery disease patients.

Except for SiliconSapiens, the three developers of AI medical devices are conducting retrospective clinical trials that check the effectiveness of the software using existing medical data.

Retrospective clinical trials have shorter study period than prospective clinical trials, which have to recruit patients in advance. The former needs two to three months for data collection and an additional three months for data analysis. In six months, clinical trials on AI medical devices could be completed.

The Ministry of Food and Drug Safety reviews a request for approval for a medical device within 85 days from the day of the request submission and gives the final approval. As the four companies obtained approvals for clinical trials in September and October last year, locally developed AI medical devices can be commercialized in the first half of this year at the earliest, observers said.

The ministry said it planned to support the development of such medical products aggressively. It set up an expert consultation group last year to gather opinions from business and academic circles and announced guidelines for approval and assessment of AI medical devices. The ministry also announced guidelines to evaluate the efficacy of clinical trials on AI medical products.

“The government conducts a briefing for related businesses every year. On a request from a company, a hospital, an educational institution, or an academic association, we also directly open a meeting with them. This year, the ministry will make extra efforts to approve advanced medical devices quickly,” said Kang Young-kyu, a section chief at the ministry’s High-tech Medical Devices Division.

lhs@docdocdoc.co.kr

<© Korea Biomedical Review, All rights reserved.>