Jesús Sampedro-Gómez1, P Ignacio Dorado-Díaz1, Víctor Vicente-Palacios2, Antonio Sánchez-Puente3, Manuel Jiménez-Navarro4, J Alberto San Roman5, Purificación Galindo-Villardón6, Pedro L Sanchez7, Francisco Fernández-Avilés8. 1. Department of Cardiology, Hospital Universitario de Salamanca-IBSAL, Universidad de Salamanca, Salamanca, Spain; CIBERCV, Instituto de Salud Carlos III, Madrid, Spain. 2. Philips Healthcare, Salamanca, Spain. 3. Department of Cardiology, Hospital Universitario de Salamanca-IBSAL, Universidad de Salamanca, Salamanca, Spain. 4. CIBERCV, Instituto de Salud Carlos III, Madrid, Spain; UGC Área del Corazón, Hospital Virgen de la Victoria-IBIMA, Universidad de Málaga, Málaga, Spain. 5. CIBERCV, Instituto de Salud Carlos III, Madrid, Spain; Instituto de Ciencias del Corazón, Hospital Clínico Universitario de Valladolid, Valladolid, Spain. 6. Department of Statistics, Universidad de Salamanca, Salamanca, Spain. 7. Department of Cardiology, Hospital Universitario de Salamanca-IBSAL, Universidad de Salamanca, Salamanca, Spain; CIBERCV, Instituto de Salud Carlos III, Madrid, Spain. Electronic address: pedrolsanchez@secardiologia.es. 8. CIBERCV, Instituto de Salud Carlos III, Madrid, Spain; Department of Cardiology, Hospital General Universitario Gregorio Marañón-IiSGM, Universidad Complutense de Madrid, Madrid, Spain.
Abstract
BACKGROUND:Machine learning (ML) has arrived in medicine to deliver individually adapted medical care. This study sought to use ML to discriminate stent restenosis (SR) compared with existing predictive scores of SR. To develop an easily applicable model, we performed our predictions without any additional variables other than those obtained in daily practice. METHODS: The dataset, obtained from the Grupo de Análisis de la Cardiopatía Isquémica Aguda (GRACIA)-3 trial, consisted of 263 patients with demographic, clinical, and angiographic characteristics; 23 (9%) of them presented with SR at 12 months after stent implantation. A methodology to work with small imbalanced datasets, based in cross-validation and the precision/recall (PR) plots, was used, and state-of-the-art ML classifiers were trained. RESULTS: Our best performing model (0.46, area under the PR curve [AUC-PR]) was developed with an extremely randomized trees classifier, which showed better performance than chance alone (0.09 AUC-PR, corresponding to the 9% of patients presenting SR in our dataset) and 3 existing scores; Prevention of Restenosis With Tranilast and its Outcomes (PRESTO)-1 (0.31 AUC-PR), PRESTO-2 (0.27 AUC-PR), and Evaluation of Drug-Eluting Stents and Ischemic Events (EVENT) (0.18 AUC-PR). The most important variables ranked according to their contribution to the predictions were diabetes, ≥2 vessel-coronary disease, post-percutaneous coronary intervention thrombolysis in myocardial infarction (PCI TIMI)-flow, abnormal platelets, post-PCI thrombus, and abnormal cholesterol. To counteract the lack of external validation for our study, we deployed our ML algorithm in an open source calculator, in which the model would stratify patients of high and low risk as an example tool to determine generalizability of prediction models from small imbalanced sample size. CONCLUSIONS: Applied immediately after stent implantation, a ML model better differentiates those patients who will present with SR over current discriminators.
RCT Entities:
BACKGROUND: Machine learning (ML) has arrived in medicine to deliver individually adapted medical care. This study sought to use ML to discriminate stent restenosis (SR) compared with existing predictive scores of SR. To develop an easily applicable model, we performed our predictions without any additional variables other than those obtained in daily practice. METHODS: The dataset, obtained from the Grupo de Análisis de la Cardiopatía Isquémica Aguda (GRACIA)-3 trial, consisted of 263 patients with demographic, clinical, and angiographic characteristics; 23 (9%) of them presented with SR at 12 months after stent implantation. A methodology to work with small imbalanced datasets, based in cross-validation and the precision/recall (PR) plots, was used, and state-of-the-art ML classifiers were trained. RESULTS: Our best performing model (0.46, area under the PR curve [AUC-PR]) was developed with an extremely randomized trees classifier, which showed better performance than chance alone (0.09 AUC-PR, corresponding to the 9% of patients presenting SR in our dataset) and 3 existing scores; Prevention of Restenosis With Tranilast and its Outcomes (PRESTO)-1 (0.31 AUC-PR), PRESTO-2 (0.27 AUC-PR), and Evaluation of Drug-Eluting Stents and Ischemic Events (EVENT) (0.18 AUC-PR). The most important variables ranked according to their contribution to the predictions were diabetes, ≥2 vessel-coronary disease, post-percutaneous coronary intervention thrombolysis in myocardial infarction (PCI TIMI)-flow, abnormal platelets, post-PCI thrombus, and abnormal cholesterol. To counteract the lack of external validation for our study, we deployed our ML algorithm in an open source calculator, in which the model would stratify patients of high and low risk as an example tool to determine generalizability of prediction models from small imbalanced sample size. CONCLUSIONS: Applied immediately after stent implantation, a ML model better differentiates those patients who will present with SR over current discriminators.
Authors: Pablo Juan-Salvadores; Cesar Veiga; Víctor Alfonso Jiménez Díaz; Alba Guitián González; Cristina Iglesia Carreño; Cristina Martínez Reglero; José Antonio Baz Alonso; Francisco Caamaño Isorna; Andrés Iñiguez Romo Journal: Diagnostics (Basel) Date: 2022-02-06
Authors: Walid Ben Ali; Ahmad Pesaranghader; Robert Avram; Pavel Overtchouk; Nils Perrin; Stéphane Laffite; Raymond Cartier; Reda Ibrahim; Thomas Modine; Julie G Hussin Journal: Front Cardiovasc Med Date: 2021-12-08