Obstetric Medical AI Predicts Signs of Fetal Hypoxia and Maternal Cardiomyopathy in Advance

Pregnancy and childbirth are special areas that require simultaneous care for two lives: the mother and the fetus. Recently, research is actively being conducted in medical fields to utilize Artificial Intelligence (AI) technology to predict emergency situations that may occur during delivery in advance and assist medical staff in their decision-making. Professor Lee Seung-mi of the Department of Obstetrics and Gynecology at Seoul National University Hospital explained through a Seoul National University Hospital TV video how AI can be utilized as an assistant to protect the safety of mothers and fetuses in the field of obstetrics.

Can Fetal Hypoxia and Maternal Cardiomyopathy be Predicted by AI?

One of the most urgent moments in obstetric care is when the fetus's condition deteriorates rapidly. Professor Lee Seung-mi explained that if a fetus falls into hypoxia during labor, physical changes occur, and if these changes can be analyzed in advance through AI, responses such as emergency surgery can be expedited. This is because even a difference of a few minutes can have a decisive impact on the survival and occurrence of complications for the fetus. In the video, Professor Lee added, "Since hypoxia in a fetus can cause various complications if it lasts too long, it is important to deliver the baby quickly from that state," and noted that if AI can predict it 5 to 10 minutes in advance, more rapid surgical preparation will be possible.

'Peripartum cardiomyopathy,' which threatens the mother's health, is also a target for AI prediction. Peripartum cardiomyopathy is a disease where the heart's pumping function decreases sharply around the time of delivery, and if it is not recovered at an appropriate time, it can be fatal to the mother's health. Mentioning research using electrocardiogram (ECG) data, Professor Lee stated that signs of decreased heart function can be captured through subtle changes in the ECG. In particular, she suggested that this technology has the potential to be implemented through wearable devices such as smartwatches in the future. Professor Lee mentioned that research results confirming heart function problems through ECG data showed high performance, and offered the outlook that 24-hour monitoring would theoretically be possible.

Changes in Clinical Settings, from Ultrasound Interpretation to Medical Record Summarization

AI is already providing practical help in medical settings. In the case of ultrasound examinations, which are essential in obstetrics, the examination time has been significantly reduced as AI technology is applied. While it previously took about 30 minutes to measure ventricular size or major structures through precision ultrasound, commercialization has progressed to the point where examinations can be completed in 10 to 15 minutes with the assistance of AI. This is because AI assists with ultrasound images and helps in the process of finding and measuring the size of ventricles or the location of specific structures.

Additionally, technologies such as 'Kmed.ai', a Korean medical-specialized AI model that operates in a closed network environment to protect patient information, are being developed. This model is based on a Large Language Model (LLM) with high performance equivalent to the national medical licensing examination level. Through this, medical staff can immediately check guidelines such as the criteria for gestational diabetes or receive help in summarizing vast amounts of discharge summaries and medical records. This performs the role of an 'assistant' to help medical staff focus more on patient care. In particular, technologies to recognize and organize external medical records as images or to recognize and organize records from a doctor's handwritten charts are in the development stage, which is expected to increase the work efficiency of medical staff.

AI as a 'Mistake-Preventing Assistant' Rather Than a Doctor Replacement

Regarding concerns that the role of doctors will be reduced due to the advancement of AI technology, Professor Lee Seung-mi drew a clear line. Since the subject who makes the final decision on medical actions and bears the resulting responsibility is ultimately the doctor, she holds the position that it is impossible for AI to completely replace doctors. In particular, obstetrics is a field where technology must be introduced very conservatively and sufficient evidence must be secured, as it is directly linked to the life and safety of patients. Professor Lee emphasized, "There are opinions that do not see the medical community changing bit by bit for patient safety, but because lives are at stake, it cannot be easily replaced."

Professor Lee emphasized that the ultimate goal of AI is 'diagnosis' and 'prediction.' To prevent human error that may occur when medical staff are fatigued or short-staffed, the core is to perform a role like a 'secretary' that rings an alarm in advance when signs of a patient's condition worsening are detected. In other words, AI is not expected to be an entity that replaces doctors, but a reliable assistant that helps medical staff respond one step ahead by notifying them of maternal bleeding or changes in the fetus's condition in advance.