Pediatric Intensive Care Unit Length of Stay Prediction by Machine Learning DOI Creative Commons
Hammad A. Ganatra,

Samir Latifi,

Orkun Baloğlu

et al.

Bioengineering, Journal Year: 2024, Volume and Issue: 11(10), P. 962 - 962

Published: Sept. 26, 2024

Purpose: To develop and validate machine learning models for predicting the length of stay (LOS) in Pediatric Intensive Care Unit (PICU) using data from Virtual Systems (VPS) database. Methods: A retrospective study was conducted utilizing (ML) algorithms to analyze predict PICU LOS based on historical patient VPS The included over 100 North American PICUs spanning years 2015–2020. After excluding entries with missing variables those indicating recovery cardiac surgery, dataset comprised 123,354 encounters. Various ML models, including Support Vector Machine, Stochastic Gradient Descent Classifier, K-Nearest Neighbors, Decision Tree, Boosting, CatBoost, Recurrent Neural Networks (RNNs), were evaluated their accuracy at thresholds 24 h, 36 48 72 5 days, 7 days. Results: RNN demonstrated highest accuracy, particularly h thresholds, rates between 70 73%. These results far outperform traditional statistical existing prediction methods that report only around 50%, which is effectively unusable practical setting. also exhibited balanced performance sensitivity (up 74%) specificity 82%) these thresholds. Conclusions: RNNs, show moderate effectiveness slightly 70%, outperforming previously reported human predictions. This suggests potential utility enhancing resource staffing management PICUs. However, further improvements through training specialized databases can potentially achieve better clinical applicability.

Language: Английский

Artificial intelligence for breast cancer detection and its health technology assessment: A scoping review DOI Creative Commons

Anisie Uwimana,

Giorgio Gnecco, Massimo Riccaboni

et al.

Computers in Biology and Medicine, Journal Year: 2024, Volume and Issue: 184, P. 109391 - 109391

Published: Nov. 22, 2024

Language: Английский

Citations

4
Impact of an artificial intelligence decision support system among radiologists with different levels of experience in breast ultrasound: A prospective study in a tertiary center DOI
Giovanni Irmici, Andrea Cozzi,

Cathérine Depretto

et al.

European Journal of Radiology, Journal Year: 2025, Volume and Issue: 185, P. 112012 - 112012

Published: Feb. 26, 2025

Language: Английский

Citations

0
Artificial Intelligence - A Primer for Diagnosis and Interpretation of Breast Cancer DOI Creative Commons

Anand Mohan Jha,

Abikesh Prasada Kumar Mahapatra,

John Abraham

et al.

International Journal of Trends in OncoScience, Journal Year: 2024, Volume and Issue: unknown, P. 27 - 36

Published: Jan. 5, 2024

Breast Cancer (BC) is a major universal health problem. Early detection and precise diagnosis are vital for enlightening outcomes. Artificial Intelligence (AI) technologies can potentially revolutionize the field of BC by providing quantitative representations medical images to assist in segmentation, diagnosis, prognosis. AI improve image quality, detect segment breast lesions, classify cancer predict its behavior, integrate data from multiple sources clinical It lead more personalized effective treatment patients. Challenges faced real-life solicitations include curation, model interpretability, run-through guidelines. However, implementation expected deliver guidance patient-tailored management. global problem; early crucial improving Imaging key screening, effectiveness assessment tool. irresistible number creates heavy capacity radiologists delays reporting. has potential imaging efficiency accuracy. recognize, segment, diagnose tumor lesions automatically analyze on molecular level. could strategies. AI-assisted still stages development, research needed validate effectiveness. Therefore, promising new technology that progress BC, BC. More bring this practice.

Language: Английский

Citations

2
Pediatric Intensive Care Unit Length of Stay Prediction by Machine Learning DOI Creative Commons
Hammad A. Ganatra,

Samir Latifi,

Orkun Baloğlu

et al.

Bioengineering, Journal Year: 2024, Volume and Issue: 11(10), P. 962 - 962

Published: Sept. 26, 2024

Purpose: To develop and validate machine learning models for predicting the length of stay (LOS) in Pediatric Intensive Care Unit (PICU) using data from Virtual Systems (VPS) database. Methods: A retrospective study was conducted utilizing (ML) algorithms to analyze predict PICU LOS based on historical patient VPS The included over 100 North American PICUs spanning years 2015–2020. After excluding entries with missing variables those indicating recovery cardiac surgery, dataset comprised 123,354 encounters. Various ML models, including Support Vector Machine, Stochastic Gradient Descent Classifier, K-Nearest Neighbors, Decision Tree, Boosting, CatBoost, Recurrent Neural Networks (RNNs), were evaluated their accuracy at thresholds 24 h, 36 48 72 5 days, 7 days. Results: RNN demonstrated highest accuracy, particularly h thresholds, rates between 70 73%. These results far outperform traditional statistical existing prediction methods that report only around 50%, which is effectively unusable practical setting. also exhibited balanced performance sensitivity (up 74%) specificity 82%) these thresholds. Conclusions: RNNs, show moderate effectiveness slightly 70%, outperforming previously reported human predictions. This suggests potential utility enhancing resource staffing management PICUs. However, further improvements through training specialized databases can potentially achieve better clinical applicability.

Language: Английский

Citations

1