News & Events

  • Title : Hallym’s AI model predicts possibility of falls and bedsores in real time
No. :
Date :
Hit :


  • Hallym University Medical Center (HUMC) has developed an artificial intelligence (AI) model for the first time in Korea that can predict the possibility of inpatients falling or having bedsores in real-time.

    Falls and bedsores are the most important criteria that hospitals manage for patient safety along with in-hospital infection. Since these safety issues could affect the recovery and prognosis of patients in the course of treatment, hospitals can improve the quality of medical care reducing costs by preventing accidents.

    In order to develop this AI model, HUMC has analyzed and processed 160,000 cases of falls over the last 5 years and 280,000 cases of bedsores over the last 10 years, and applied them to an optimized machine learning algorithm.

    The ‘AI model for prediction of the risk of falls’ reflect more than 20 factors including patients’ basic information, drugs that can cause falling, the use of anticoagulant drug, osteoporosis, walking habits, and cognitive disorder.

    The ‘AI model for prediction of the risk of bedsores’ was also developed after a machine learning process based on more than 20 factors such as recognition of sense, humidity, level of mobility, nutritional status, diet, and underlying diseases.

    Although the existing falls and bedsores prediction tools allowed medical staff to identify the level of risk in only three stages of high, middle, and low at certain points such as at the time of hospitalization or the time after the surgery, the biggest feature of the AI model developed by HUMC this time is that it is possible to get the results in real-time.

    “Whenever the nurses at the ward check the patient information, AI models calculate the possibility of falls and bedsores in real-time and show the result value to them,” explained Manager Kang-Il Lee of Medical Information Team. “It is meaningful in the sense of immediately checking the risk of falls and bedsores that change in real-time depending on what medical treatment is used,” he added.

    All five hospitals of HUMC are using this AI model. When the bedsore prediction rate exceeds 70% at general ward or 90% at ICUs, nurses take care of their patients more intensively. In the case of falls, preventive nursing programs are operated from a lower predicted value than that of bedsores.

    “As the risk of falls and bedsores can be identified in real-time, it is now possible to provide customized intensive management for patients in high risk groups,” said Unit Manager Hae-Jung Cho. “It also seems to help reduce safety accidents by making patients and their families behave more carefully after seeing the numerical risk of safety accidents,” she added.

    By Hyun Ho Choi, Int’l Cooperation Team, HUMC (