42 AI prediction models at Hallym University Medical Center can predict patient conditions in advance
Nov 01, 2024
In the rapidly changing technology world, artificial intelligence (AI) has emerged as a remarkable innovation. It has been introduced and used in all occupational groups that have innovated in various industries, and medical institutions are no exception. Currently, AI is actively applied in the healthcare field and examples that are widely known as cases are AI chatbots for counseling or treatment, and diagnostic fields centered on image recognition. These solutions were trained to acquire and read vast amounts of learning data and were introduced into the medical field. It increases patient convenience and diagnostic accuracy and assists doctors to receive help in diagnosis and reading.
Beyond diagnosis and reading, one of the special applications of AI is predictive modeling. It uses the power of AI to make treatment decisions and predict the patient's future. AI models are spreading beyond diagnostic assistance to disease prediction and management areas. Accordingly, medical institutions are also introducing AI technology and applying it to the medical field.
Hallym University Medical Center (Director Kim Yong-sun) has been rapidly developing a medical AI prediction model since 2020. The reason is to quickly prevent safety accidents that may occur in patients to save patients' lives and increase patient safety. The Information Management Bureau under the Medical Center conceived the development of an AI prediction model by prioritizing diseases and accidents that easily occur in hospitals with 'patient safety' as its top priority.
First, in order to secure new AI deep learning and machine learning technologies, internal instructors were nurtured, and then a curriculum was created to conduct internal training for developers of the Information Management Bureau. Since then, developers have developed AI prediction models by conducting action learning in education, and they have been mounted on the digital comprehensive medical information system (Refomax, Reformax) to create the current prediction model.
Hallym University Medical Center's 42 AI prediction models are the largest number of developments and applications among domestic medical institutions, and the average prediction rate of 42 prediction models is 87%. The high prediction rate is due to the use of an average of 10 years' worth of patient data in Electronic Medical Record (EMR). In addition, learning variables (data) such as medical department, age, gender, medical day, and diagnostic code were analyzed and processed and applied to optimized machine learning algorithms.
The developed AI prediction model is used for emergency patients, outpatients, and inpatients such as ▲fall and bath window prediction model ▲ arteriovenous vascular stenosis prediction model for dialysis patients ▲ venitis occurrence prediction model ▲ diabetes complications prediction model ▲ CRE·CPE infection prediction model ▲ emergency room patients' bedsores occurrence prediction model ▲ delirium occurrence prediction model.
◇'How did you know and come here beforehand?'…Prediction of Patient Status
AI prediction model calculates and presents the likelihood of occurrence in real time whenever medical staff inquire patient information in the prescription delivery system (OCS). The AI calculates the likelihood of occurrence based on patient information that changes every moment, and classifies patients into high, medium, and low-risk groups according to this prediction.
For example, aspiration pneumonia is a disease that occurs when pathogenic bacteria contained in secretions in the stomach or oral cavity enter the lungs through the bronchial tubes rather than the esophagus. Previously, they had no choice but to deal with only clinical situations that increased the risk of developing aspiration pneumonia in hospitalized patients, such as repeated cerebral infarction, dementia, and unconsciousness. However, the predictive model identifies the risk of developing aspiration pneumonia in hospitalized patients in real time and can cope with it before the disease occurs, allowing patients to receive treatment more safely.
As an actual AI prediction model patient experience case, guardian Kim Mi-sook (59 years old, pseudonym) once said, "When my husband, who was hospitalized with pneumonia, had dysphagia (swallowing disorder), a nurse in the ward rushed to me at once, inserted an airway, provided oral suction quickly, and kindly provided education and explanation for aspiration prevention. 'How did you know that you came here beforehand? I thought so, but AI predicted and told me first, so I later found out that the nurses were doing it"I felt grateful to the medical staff because the AI prediction model was amazing and it seemed to save lives thanks to it."
Jang Kyung-hee, head of the nursing team at Hallym University Chuncheon Sacred Heart Hospital, said "Because it is possible to grasp the possibility of aspiration pneumonia in real time, customized management is possible for high-risk patients such as the elderly."In addition to aspiration pneumonia through this AI, we can also do activities to prevent various emergencies that may occur due to swallowing disorders or aspiration.".
◇ Increase the efficiency of medical staff by introducing predictive models
In addition, the work efficiency of medical staff has increased. Until now, it was not easy for a small number of medical staff to monitor hundreds of patients at the same time before the predictive model was developed, but now the predictive accuracy of the predictive model is high, so just being able to grasp the risk before the patient's condition deteriorates will inevitably increase the efficiency of medical staff's work.
Since the introduction of the actual AI prediction model, it has been confirmed that the likelihood of severe cases in patients has decreased and the work efficiency of medical staff has improved. According to a satisfaction survey of nurses working in ward 108 of Hallim University's Gangnam Sacred Heart Hospital in August, 97% of all respondents said they were satisfied with the introduction of AI prediction models. The reasons for satisfaction were that the patient's condition could be identified in real time, so that customized patient management was possible 24 hours a day for ', '365 days, and 'the severe incidence of patients was lowered, and the work convenience was higher than before, so there were many work efficiency and patient management enhancements.
Lim Eun-joo, head of the nursing department at Hallym University's Gangnam Sacred Heart Hospital, said, "Patients and their guardians have also become more alert as they frequently encountered the previously vaguely accepted risk of safety accidents." The introduction of AI prediction models has led to a reduction in the occurrence of actual patient safety accidents, contributing to the establishment of a safe hospital culture."
As such, the introduction of the AI prediction model led to satisfaction for all patients, guardians, and medical staff. As patient safety has been strengthened, VOCs (customer sounds) have also decreased. Hallym University Medical Center plans to develop four more AI prediction models and apply them to the treatment site within this year. In the future, we plan to do our best to strengthen patient safety and make it a hospital where patients can trust and attend.
Beyond diagnosis and reading, one of the special applications of AI is predictive modeling. It uses the power of AI to make treatment decisions and predict the patient's future. AI models are spreading beyond diagnostic assistance to disease prediction and management areas. Accordingly, medical institutions are also introducing AI technology and applying it to the medical field.
Hallym University Medical Center (Director Kim Yong-sun) has been rapidly developing a medical AI prediction model since 2020. The reason is to quickly prevent safety accidents that may occur in patients to save patients' lives and increase patient safety. The Information Management Bureau under the Medical Center conceived the development of an AI prediction model by prioritizing diseases and accidents that easily occur in hospitals with 'patient safety' as its top priority.
First, in order to secure new AI deep learning and machine learning technologies, internal instructors were nurtured, and then a curriculum was created to conduct internal training for developers of the Information Management Bureau. Since then, developers have developed AI prediction models by conducting action learning in education, and they have been mounted on the digital comprehensive medical information system (Refomax, Reformax) to create the current prediction model.
Hallym University Medical Center's 42 AI prediction models are the largest number of developments and applications among domestic medical institutions, and the average prediction rate of 42 prediction models is 87%. The high prediction rate is due to the use of an average of 10 years' worth of patient data in Electronic Medical Record (EMR). In addition, learning variables (data) such as medical department, age, gender, medical day, and diagnostic code were analyzed and processed and applied to optimized machine learning algorithms.
The developed AI prediction model is used for emergency patients, outpatients, and inpatients such as ▲fall and bath window prediction model ▲ arteriovenous vascular stenosis prediction model for dialysis patients ▲ venitis occurrence prediction model ▲ diabetes complications prediction model ▲ CRE·CPE infection prediction model ▲ emergency room patients' bedsores occurrence prediction model ▲ delirium occurrence prediction model.
◇'How did you know and come here beforehand?'…Prediction of Patient Status
AI prediction model calculates and presents the likelihood of occurrence in real time whenever medical staff inquire patient information in the prescription delivery system (OCS). The AI calculates the likelihood of occurrence based on patient information that changes every moment, and classifies patients into high, medium, and low-risk groups according to this prediction.
For example, aspiration pneumonia is a disease that occurs when pathogenic bacteria contained in secretions in the stomach or oral cavity enter the lungs through the bronchial tubes rather than the esophagus. Previously, they had no choice but to deal with only clinical situations that increased the risk of developing aspiration pneumonia in hospitalized patients, such as repeated cerebral infarction, dementia, and unconsciousness. However, the predictive model identifies the risk of developing aspiration pneumonia in hospitalized patients in real time and can cope with it before the disease occurs, allowing patients to receive treatment more safely.
As an actual AI prediction model patient experience case, guardian Kim Mi-sook (59 years old, pseudonym) once said, "When my husband, who was hospitalized with pneumonia, had dysphagia (swallowing disorder), a nurse in the ward rushed to me at once, inserted an airway, provided oral suction quickly, and kindly provided education and explanation for aspiration prevention. 'How did you know that you came here beforehand? I thought so, but AI predicted and told me first, so I later found out that the nurses were doing it"I felt grateful to the medical staff because the AI prediction model was amazing and it seemed to save lives thanks to it."
Jang Kyung-hee, head of the nursing team at Hallym University Chuncheon Sacred Heart Hospital, said "Because it is possible to grasp the possibility of aspiration pneumonia in real time, customized management is possible for high-risk patients such as the elderly."In addition to aspiration pneumonia through this AI, we can also do activities to prevent various emergencies that may occur due to swallowing disorders or aspiration.".
◇ Increase the efficiency of medical staff by introducing predictive models
In addition, the work efficiency of medical staff has increased. Until now, it was not easy for a small number of medical staff to monitor hundreds of patients at the same time before the predictive model was developed, but now the predictive accuracy of the predictive model is high, so just being able to grasp the risk before the patient's condition deteriorates will inevitably increase the efficiency of medical staff's work.
Since the introduction of the actual AI prediction model, it has been confirmed that the likelihood of severe cases in patients has decreased and the work efficiency of medical staff has improved. According to a satisfaction survey of nurses working in ward 108 of Hallim University's Gangnam Sacred Heart Hospital in August, 97% of all respondents said they were satisfied with the introduction of AI prediction models. The reasons for satisfaction were that the patient's condition could be identified in real time, so that customized patient management was possible 24 hours a day for ', '365 days, and 'the severe incidence of patients was lowered, and the work convenience was higher than before, so there were many work efficiency and patient management enhancements.
Lim Eun-joo, head of the nursing department at Hallym University's Gangnam Sacred Heart Hospital, said, "Patients and their guardians have also become more alert as they frequently encountered the previously vaguely accepted risk of safety accidents." The introduction of AI prediction models has led to a reduction in the occurrence of actual patient safety accidents, contributing to the establishment of a safe hospital culture."
As such, the introduction of the AI prediction model led to satisfaction for all patients, guardians, and medical staff. As patient safety has been strengthened, VOCs (customer sounds) have also decreased. Hallym University Medical Center plans to develop four more AI prediction models and apply them to the treatment site within this year. In the future, we plan to do our best to strengthen patient safety and make it a hospital where patients can trust and attend.
|
bellho@sportschosun.com
