학술논문
Abstract 14290: Prospective Clinical Validation of the STOPSHOCK Smartphone Application - Artificial Intelligence Model for Prediction of Cardiogenic Shock in Patients With Acute Coronary Syndrome
Document Type
Academic Journal
Author
Bohm, Allan; Jajcay, Nikola; Spartalis, Michael; Gubareva, Ekaterina; Jankova, Jana; Segev, Amitai; Krychtiuk, Konstantin; Tavazzi, Guido; Rathod, Krishnaraj; Maurizi, Kevin; Jarakovic, Milana; Guerra, Federico; Matetzky, Shlomi; Skurk, Carsten; Thevathasan, Tharusan; Huber, Kurt; Dragunova, Patricia; Brucknerova, Jana; Kollarova, Marta; Berta, Imrich; Petrikova, Katarina; Bezak, Branislav
Source
Circulation. Nov 07, 2023 148(Suppl_1 Suppl 1):A14290-A14290
Subject
Language
English
ISSN
0009-7322
Abstract
Background: Cardiogenic shock (CS) complicating acute coronary syndrome (ACS) is a life-threatening condition with mortality reaching 50% despite the use of mechanical circulatory support devices (MCS). It is hypothesized that early implantation of MCS before hemodynamic deterioration could prevent CS. For this purpose, we have developed and externally validated an AI model for CS prediction available as a smartphone application (STOPSHOCK app).Research question: Could the STOPSHOCK app identify ACS patients at high risk of CS development in contemporary clinical practice?Methods: The STOPSHOCK model was derived on a population of 3232 and validated on an external cohort of 5123 patients with high discrimination (c-statistics: 0.84 in external validation). The model was based on variables that were available at first medical contact. A dedicated app was developed to test the performance of the model in the clinical setting. 103 consecutive ACS patients were enrolled in intensive cardiac care units in 8 centers from USA, Europe, and Asia.Results: Heart rate, respiration rate, SpO2, blood glucose level, systolic BP, age, sex, shock index, heart rhythm, type of ACS, history of hypertension, congestive heart failure and hypercholesterolemia were used as input variables. Clinical outcome (presence of CS) was reported at hospital discharge. STOPSHOCK model had a c-statistics of 0.97.Conclusions: Our machine learning model for CS prediction in patients with ACS available at first medical contact showed high discrimination and feasibility in clinical settings as a smartphone app.