[SBA x IT Donga Joint Planning] The Seoul Metropolitan Government and the Seoul Business Agency (SBA) have established startup hubs (centers) in Chang-dong, Seongsu, and Dongjak, Seoul. They support startups through stage-by-stage programs from initial startup to growth phases. Promising startups with high growth potential that have shown remarkable activity by 2025 are introduced.

Saramul Ihaehada is developing a portable medical device based on artificial intelligence (AI) that predicts sudden cardiac arrest (hereafter referred to as cardiac arrest), called the '24-hour Heart Keeper' (hereafter referred to as Heart Keeper). Heart Keeper measures heart rate and respiration in a non-contact manner using radar sensors, and based on this, calculates heart rate variability (HRV) and respiratory variability (RV) to predict cardiac arrest.

Saramul Ihaehada plans to officially launch Heart Keeper after obtaining the second-class medical device certification from the Ministry of Food and Drug Safety (MFDS) in the second half of 2026. They also plan to enter the U.S. telemedicine market and develop non-contact blood pressure measurement and real-time monitoring medical devices for hospitalized patients.

With the goal of shifting the medical paradigm to disease prevention, an interview was conducted with Jeong Chang-jin, CEO of Saramul Ihaehada, about non-contact cardiac arrest prediction technology and Heart Keeper.


Founded to Develop Cardiac Arrest Prediction Solution

IT Donga: Hello, CEO Jeong Chang-jin. Please introduce yourself.

Jeong Chang-jin: Hello, I am Jeong Chang-jin of Saramul Ihaehada. I have developed medical information systems for 20 years and worked as an executive at a medical device development company. I mainly handled electronic medical records (EMR). Then, in July 2021, I founded Saramul Ihaehada.

IT Donga: What motivated you to start Saramul Ihaehada?

Jeong Chang-jin: Cardiac arrest is a condition where all heart activity is suddenly lost due to irregular heartbeats, causing breathing to stop. If cardiopulmonary resuscitation is not received within the golden time of 4 minutes after cardiac arrest occurs, it can lead to death due to brain damage. The problem is that most do not receive appropriate measures within the golden time. In fact, the mortality rate from cardiac arrest is quite high at about 91%.

I believed that predicting cardiac arrest at least 1 hour to a maximum of 24 hours in advance could drastically reduce the mortality rate. So, since 2009, I started contemplating cardiac arrest prediction methods with Professor Lee Sang-hyung of the Department of Neurosurgery at Seoul National University College of Medicine.

During this time, I came across a paper stating that various precursor symptoms such as chest pain, shortness of breath, dizziness, fatigue, and palpitations are experienced hours to days before cardiac arrest occurs. I also encountered technology that measures heart rate and respiration in a non-contact manner using radar sensors. This led me to think that a cardiac arrest prediction solution could be developed, which prompted the founding of Saramul Ihaehada.

IT Donga: It took quite a long time from the initial contemplation in 2009 to the actual founding.

Jeong Chang-jin: Various attempts were made during that time, but with the technology available then, it was not possible to properly predict cardiac arrest. Then, upon encountering the non-contact method based on radar sensors, we determined that we could commercialize the product we envisioned. Thus, we began preparing for the startup in earnest.


IT Donga: Please introduce Saramul Ihaehada. The company name is impressive; what does it mean?

Jeong Chang-jin: Saramul Ihaehada means 'understanding the individual characteristics of people to predict diseases considering their unique physiological characteristics and changes.' It embodies our intention to model the normal state unique to each individual rather than applying the same standard to everyone and to predict diseases by comprehensively analyzing the direction of changes and instability of various indicators.

We are developing Heart Keeper, an AI-based portable smart medical device that predicts cardiac arrest. Through this, we aim to play a key role in the field of personalized healthcare services, such as early detection of cardiac arrest, arrhythmia prevention, and emergency response to cardiac arrest occurrences.

Non-contact Measurement of Vital Signs, AI-based Cardiac Arrest Prediction 'Heart Keeper'

IT Donga: Please introduce Heart Keeper.

Jeong Chang-jin: To predict cardiac arrest, it is necessary to check vital signs such as heart rate, respiration, heart rate variability, and respiratory variability. Heart rate and respiration are the most important indicators of heart health. Typically, the heart beats faster when inhaling and slower when exhaling, but there are cases where the heart rate increases while respiration decreases. Cardiac arrest can occur at this time. Therefore, heart rate and respiration must be measured simultaneously.

Heart rate variability refers to the range of heart rate fluctuations, and respiratory variability indicates the range of inhalation and exhalation. By confirming the synchronization state of heart rate and respiration through heart rate variability and respiratory variability, the health status of the heart can be determined.

Heart Keeper consists of a hardware medical device that measures heart rate and respiration in a non-contact manner and an AI-based software medical device (SaMD) that calculates heart rate variability and respiratory variability to predict cardiac arrest.


We applied radar sensors for collecting vital signals for non-contact measurement. Radar has the property of returning after hitting an object, allowing it to recognize the shape and position of objects. Heart Keeper directs radar towards the human thorax and detects thoracic phase changes to measure respiration and heart rate. Our self-developed AI algorithm extracts heart rate variability and respiratory variability data and informs the risk of cardiac arrest. Since Heart Keeper uses a non-contact method, there is no need to attach separate sensors or devices to the body, and continuous 24-hour measurement is possible.

We are the only ones with a solution that measures heart rate and respiration in a non-contact manner and extracts heart rate variability and respiratory variability to predict cardiac arrest. While smartwatches or electrocardiographs measure heart rate, they do not measure respiration. Existing cardiac arrest prediction solutions are based on EMR data written by medical staff, making it difficult to measure patient status in real-time. We have improved these limitations.

For reference, Heart Keeper also includes an electrocardiogram (ECG) sensor. Since cardiac arrest is directly related to human life, a dual safety device was prepared. If the risk of cardiac arrest or arrhythmia is predicted while monitoring heart health in a non-contact manner for 24 hours, the ECG sensor can be used to measure the ECG more precisely.

IT Donga: Accuracy seems important for Heart Keeper.

Jeong Chang-jin: We have enhanced completeness and accuracy through verification with co-developer Professor Lee Sang-hyung and Professor Yoon Chi-soon of the Department of Thoracic Surgery at Incheon St. Mary's Hospital, who joined as an advisor in 2021. As a result, we received a test report with an accuracy of over 93%, a recall rate of 88%, and a precision of 86% from the cardiac arrest prediction model test through the Korea Laboratory Accreditation Scheme (KOLAS) last July. It has been recognized as having sufficiently high accuracy for commercialization.

IT Donga: When is Heart Keeper expected to be released?

Jeong Chang-jin: The main functions of Heart Keeper have been completed, and we plan to add the PARIS algorithm and obtain the second-class medical device certification from the MFDS next year before launching it. The official release is expected in the second half of 2026.

IT Donga: What is the PARIS algorithm?

Jeong Chang-jin: When predicting cardiac arrest, if judged based on the average heart rate of general high-risk patients, sensitivity to individual physiological characteristics and changes may decrease. This is because heart rate and respiration patterns vary from person to person. To increase the accuracy of cardiac arrest prediction, it is necessary to model the normal state of individuals and comprehensively analyze the direction of changes and instability of various indicators. Therefore, we developed the PARIS algorithm, a personalized AI model for cardiac arrest prediction. For reference, PARIS stands for Personalized Arrest Risk via an Interwoven Sequence-model, a multi-layered time-series fusion-based personalized cardiac arrest risk inference model.


IT Donga: You are currently receiving support from the Seoul Startup Center Dongjak. What kind of support have you received?

Jeong Chang-jin: We moved into the Seoul Startup Center Dongjak in October 2024, and the pleasant and clean work environment and various rest spaces have greatly helped improve work efficiency. Additionally, through programs linked with SBA, we received support such as creating brand toons for product promotion and evaluating the value of company-owned patents. We received advice on attracting investment, business direction, contractual relationships with investors, and shareholder agreements from experts such as accelerators, venture capitalists, and lawyers. There is a lot of education on aspects that early startups should not overlook, which is very helpful in running the company.

Goal to Shift to Prevention-focused Medical Paradigm

IT Donga: Finally, please tell us about the future plans and goals of Saramul Ihaehada.

Jeong Chang-jin: In the long term, we plan to expand the scope of disease prediction based on non-contact measurement technology. First, we aim to develop non-contact blood pressure measurement technology. This will allow continuous monitoring of blood pressure status without the patient being conscious, aiding in managing chronic diseases such as hypertension. We will also develop an automatic vital sign measurement medical device for hospitalized patients. We plan to enhance severity management efficiency by measuring heart rate, respiration, heart rate variability, respiratory variability, blood pressure, and oxygen saturation in a non-contact manner in real-time. We plan to enter the U.S. market in the second half of 2026. After establishing a branch, we aim to obtain U.S. FDA approval and enter the U.S. telemedicine market.

Our goal is to analyze human behavior and living spaces to predict and prevent diseases. Through this, we aim to shift the medical paradigm from treatment to prevention and from hospitals to living spaces. We will contribute to saving many lives by preventing severe diseases such as cardiac arrest.

IT Donga Han Man-hyeok Reporter (mh@itdonga.com)

Popular News