Prediction of respiratory decompensation in Covid-19 patients using machine learning: The READY trial

Computers in Biology and Medicine, Journal Year: 2020, Volume and Issue: 124, P. 103949 - 103949

Published: Aug. 6, 2020

Currently, physicians are limited in their ability to provide an accurate prognosis for COVID-19 positive patients. Existing scoring systems have been ineffective identifying patient decompensation. Machine learning (ML) may offer alternative strategy. A prospectively validated method predict the need ventilation patients is essential help triage patients, allocate resources, and prevent emergency intubations associated risks. In a multicenter clinical trial, we evaluated performance of machine algorithm prediction invasive mechanical within 24 h initial encounter. We enrolled with diagnosis who were admitted five United States health between March May 4, 2020. 197 REspirAtory Decompensation model covid-19 patients: prospective studY (READY) trial. The had higher diagnostic odds ratio (DOR, 12.58) predicting than comparator early warning system, Modified Early Warning Score (MEWS). also achieved significantly sensitivity (0.90) MEWS, which 0.78, while maintaining specificity (p < 0.05). first trial needs among demonstrated h. This care teams effectively resources. Further, capable accurately 16% more widely used system minimizing false results.

Language: Английский

---

Deep transfer learning artificial intelligence accurately stages COVID-19 lung disease severity on portable chest radiographs

PLoS ONE, Journal Year: 2020, Volume and Issue: 15(7), P. e0236621 - e0236621

Published: July 28, 2020

This study employed deep-learning convolutional neural networks to stage lung disease severity of Coronavirus Disease 2019 (COVID-19) infection on portable chest x-ray (CXR) with radiologist score as ground truth. consisted 131 CXR from 84 COVID-19 patients (51M 55.1±14.9yo; 29F 60.1±14.3yo; 4 missing information). Three expert radiologists scored the left and right separately based degree opacity (0-3) geographic extent (0-4). Deep-learning network (CNN) was used predict scores. Data were split into 80% training 20% testing datasets. Correlation analysis between AI-predicted versus scores analyzed. Comparison made traditional transfer learning. The average 2.52 (range: 0-6) a standard deviation 0.25 (9.9%) across three readers. 3.42 0-8) 0.57 (16.7%) inter-rater agreement yielded Fleiss' Kappa 0.45 for 0.71 score. strongly correlated scores, top model yielding correlation coefficient (R2) 0.90 0.73-0.90 learning 0.83-0.90 learning) mean absolute error 8.5% (ranges: 17.2-21.0% 8.5%-15.5, respectively). Transfer generally performed better. In conclusion, CNN accurately stages infection. approach may prove useful severity, prognosticate, treatment response survival, thereby informing risk management resource allocation.

Language: Английский

---

An artificial intelligence system for predicting the deterioration of COVID-19 patients in the emergency department

npj Digital Medicine, Journal Year: 2021, Volume and Issue: 4(1)

Published: May 12, 2021

Abstract During the coronavirus disease 2019 (COVID-19) pandemic, rapid and accurate triage of patients at emergency department is critical to inform decision-making. We propose a data-driven approach for automatic prediction deterioration risk using deep neural network that learns from chest X-ray images gradient boosting model routine clinical variables. Our AI prognosis system, trained data 3661 patients, achieves an area under receiver operating characteristic curve (AUC) 0.786 (95% CI: 0.745–0.830) when predicting within 96 hours. The extracts informative areas assist clinicians in interpreting predictions performs comparably two radiologists reader study. In order verify performance real setting, we silently deployed preliminary version New York University Langone Health during first wave which produced real-time. summary, our findings demonstrate potential proposed system assisting front-line physicians COVID-19 patients.

Language: Английский

---

DenseNet Convolutional Neural Networks Application for Predicting COVID-19 Using CT Image

SN Computer Science, Journal Year: 2021, Volume and Issue: 2(5)

Published: July 23, 2021

Language: Английский

---

Hatred and trolling detection transliteration framework using hierarchical LSTM in code-mixed social media text

Complex & Intelligent Systems, Journal Year: 2021, Volume and Issue: 9(3), P. 2813 - 2826

Published: Aug. 17, 2021

Abstract The paper describes the usage of self-learning Hierarchical LSTM technique for classifying hatred and trolling contents in social media code-mixed data. LSTM-based learning is a novel architecture inspired from neural models. proposed HLSTM model trained to identify words available contents. systems equipped with predicting mechanism annotating transliteration domain. Hindi–English data are ordered into Hindi, English, labels classification. word embedding character-embedding features used here representation sentence detect words. method developed based on helps recognizing context by mining intention user using that sentence. Wide experiments suggests HLSTM-based classification gives accuracy 97.49% when evaluated against standard parameters like BLSTM, CRF, LR, SVM, Random Forest Decision Tree models especially there some

Language: Английский

---