Clinical imaging · 2026
Clinical integration of an inferior vena cava filter alert system using an artificial intelligence application.
СтатьяНазвание и аннотация: Перевод в очереди
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Hardy Josiah, Ahmed Omar, Malinow Ian, Parsa Shirin, Lim Maria, Herskovits Edward, Kim Hyun, Fang Adam
Аннотация
OBJECTIVE: Timely follow-up and removal of inferior vena cava (IVC) filters is important for preventing filter-related complications. This study evaluated the performance and clinical impact of integrating the Filter Alert System (FAS), a natural language processing tool for identifying patients with an inferior vena cava (IVC) filter.
METHODS: The FAS was developed using the NLP feature of an artificial intelligence (AI) software application, which scans computed tomography (CT) radiology reports. The software is linked to an online notification system, which then identifies reports that contain keywords. An initial test study was conducted to validate the FAS over a 2-month period, and a prospective study was conducted wherein the FAS was clinically integrated over a 9-month period. The percentage of patients eligible for filter retrieval who were scheduled for clinic, IVC filter retrieval rates, and complication rates were compared 9 months before (pre-FAS) and after integration of the FAS (post-FAS) using a Fisher's test.
RESULTS: The FAS accuracy was 99.7%; it achieved high sensitivity (85.7%), specificity (99.9%), and positive predictive value (94.7%). After clinical integration, significantly more eligible patients were scheduled for clinic evaluation (58.0% vs 21.6%, respectively) and subsequent filter retrieval (44.0% vs 21.6%, respectively) in the post-FAS versus pre-FAS group with similar rates of IVC filter-related complications. Among patients with known filter placement dates, the mean (range) filter dwell time was similar between the pre-FAS and post-FAS groups (38.9 [2.3-190.7] months vs 65.6 [3.6-239.8] months).
CONCLUSIONS: FAS can be clinically integrated for promoting active surveillance of IVC filter patients.
METHODS: The FAS was developed using the NLP feature of an artificial intelligence (AI) software application, which scans computed tomography (CT) radiology reports. The software is linked to an online notification system, which then identifies reports that contain keywords. An initial test study was conducted to validate the FAS over a 2-month period, and a prospective study was conducted wherein the FAS was clinically integrated over a 9-month period. The percentage of patients eligible for filter retrieval who were scheduled for clinic, IVC filter retrieval rates, and complication rates were compared 9 months before (pre-FAS) and after integration of the FAS (post-FAS) using a Fisher's test.
RESULTS: The FAS accuracy was 99.7%; it achieved high sensitivity (85.7%), specificity (99.9%), and positive predictive value (94.7%). After clinical integration, significantly more eligible patients were scheduled for clinic evaluation (58.0% vs 21.6%, respectively) and subsequent filter retrieval (44.0% vs 21.6%, respectively) in the post-FAS versus pre-FAS group with similar rates of IVC filter-related complications. Among patients with known filter placement dates, the mean (range) filter dwell time was similar between the pre-FAS and post-FAS groups (38.9 [2.3-190.7] months vs 65.6 [3.6-239.8] months).
CONCLUSIONS: FAS can be clinically integrated for promoting active surveillance of IVC filter patients.