T-Dpsom

Manduchi, Laura; Hüser, Matthias; Faltys, Martin; Vogt, Julia E.; Rätsch, Gunnar; Fortuin, Vincent

doi:10.1145/3450439.3451872

Cited by 8 publications

(3 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The hex-map visualization provides a snapshot of the ICU population at any given time and allows for the monitoring of patient states over time with updates, akin to those seen in methodologies like T-DPSOM 57,58 . This dynamic tracking is based on the automated integration of multiple respiratory state dimensions and uses nonlinear dimensionality reduction to provide the position of an individual patient on the map of respiratory health states.…”

Section: Discussionmentioning

confidence: 99%

A comprehensive ML-based Respiratory Monitoring System for Physiological Monitoring & Resource Planning in the ICU

Hüser,

Lyu,

Faltys

et al. 2024

Preprint

Self Cite

View full text Add to dashboard Cite

Respiratory failure (RF) is a frequent occurrence in critically ill patients and is associated with significant morbidity and mortality as well as resource use. To improve the monitoring and management of RF in intensive care unit (ICU) patients, we used machine learning to develop a monitoring system covering the entire management cycle of RF, from early detection and monitoring, to assessment of readiness for extubation and prediction of extubation failure risk. For patients in the ICU in the study cohort, the system predicts 80% of RF events at a precision of 45% with 65% identified 10h before the onset of an RF event. This significantly improves upon a standard clinical baseline based on the SpO2/FiO2 ratio. After a careful analysis of ICU differences, the RF alarm system was externally validated showing similar performance for patients in the external validation cohort. Our system also provides a risk score for extubation failure for patients who are clinically ready to extubate, and we illustrate how such a risk score could be used to extubate patients earlier in certain scenarios. Moreover, we demonstrate that our system, which closely monitors respiratory failure, ventilation need, and extubation readiness for individual patients can also be used for ICU-level ventilator resource planning. In particular, we predict ventilator use 8-16h into the future, corresponding to the next ICU shift, with a mean absolute error of 0.4 ventilators per 10 patients effective ICU capacity.

show abstract

Section: Discussionmentioning

confidence: 99%

A comprehensive ML-based Respiratory Monitoring System for Physiological Monitoring & Resource Planning in the ICU

Hüser,

Lyu,

Faltys

et al. 2024

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…We utilized a deep architecture for probabilistic time-series clustering (Temp-DPSOM, Manduchi et al., 2021 ) combining a variational autoencoder ( Kingma and Welling, 2014 ; Rezende et al., 2014 ), forecasting in the latent space using LSTMs and a self-organizing map (SOM) for clustering input time-series samples. Behavioral trials were described in a 7 dimensional time series sampled at 10 Hz, where the first 6 dimensions corresponded to the maze-center-referenced X and Y coordinates of the center of the body, the relative positions of the nose-end and tail-end of the body, and the area of the animal.…”

Section: Methodsmentioning

confidence: 99%

Natural VTA activity during NREM sleep influences future exploratory behavior

et al. 2022

View full text Add to dashboard Cite

“…Recently, works in computer sciences have shown that deep neural networks such as Deep Autoencoders (DAEs) (Kingma and Welling, 2013;Lecun et al, 2015) and variational autoencoders, have been used in combination Self-Organizing Map (SOM) (Kohonen, 2012) to substantially increase the performance of clustering methods (Forest et al, 2019;Manduchi et al, 2019;Tao et al, 2018). Both, autoencoders and SOM can be considered dimensionality reduction techniques that are based on neural networks, which has as advantage the ability of neural networks to identify complex patterns on high dimensionality data.…”

Section: Introductionmentioning

confidence: 99%

Evaluation of machine learning methods for prospectivity modeling and vectoring of IOCG mineralization

Martins Guerra Prado

View full text Add to dashboard Cite

é bacharel em Geologia (2012) pela Universidade de Brasília (UnB). Obteve mestrado na UnB em 2015, desenvolvendo um estudo sobre análise quantitativa da mineralogia de formações ferríferas por meio de dados espectrais. Em 2014, se tornou pesquisador em geociências do Serviço Geológico do Brasil (SGB), onde atuou no mapeamento geológico do estado de Rondônia por três anos. Atua desde 2019 como coordenador executivo do Centro de Geociências Aplicadas (CGA) do SGB, onde desenvolve pesquisas utilizando técnicas de inteligência artificial para criar soluções na área das geociências. Iniciou seu Ph.D. na Universidade de Campinas (Unicamp), em 2019, com o objetivo de investigar a aplicação de métodos de aprendizado de máquina na exploração de depósitos minerais do tipo IOCG. BIOGRAPHY Elias M. G. Prado has a bachelor's degree in Geology (2012) from the University of Brasília (UnB). He obtained a master's degree from UnB in 2015, developing a study on quantitative analysis of the mineralogy of iron formations through spectral data. In 2014, he became a researcher in geosciences at the Geological Survey of Brazil (SGB), where he worked on the geological mapping of the state of Rondônia for three years. He has served since 2019 as executive coordinator of the SGB's Center for Applied Geosciences (CGA), where he develops research using artificial intelligence techniques to create solutions in the geosciences. He started a Ph.D. at the University of Campinas (Unicamp) in 2019, aiming to investigate the application of machine learning methods in the exploration of IOCG-type mineral deposits. AGRADECIMENTO To my father, Edson, in loving memory. To my mother, Germana, and sister, Cássia, my sincere thanks for the caring, example, and full support. To my wife Flávia and son Caio, thanks for the companionship, encouragement and all the support. I thank Prof. Dr. Carlos Roberto de Souza Filho of the University of Campinas (UNICAMP) for his dedication and guidance during this learning period. To the Prof. E. John Carranza of the University of the Free State (Bloemfontein, South Africa) for the great collaboration in this work. I thank the Geological Survey of Brazil (SGB) for the release for training through internal selection process. I direct this acknowledgment to the head of the Applied Geosciences Center (CGA) of the SGB, Noevaldo Araujo Teixeira, who supports my work so many times To the various geologists and colleagues in Brazil who participate directly or indirectly in the research. To João Motta, Leandro, Rodrigo, Cimara, and Christian for all the support and knowledge exchange.

show abstract

T-Dpsom

Cited by 8 publications

References 31 publications

A comprehensive ML-based Respiratory Monitoring System for Physiological Monitoring & Resource Planning in the ICU

A comprehensive ML-based Respiratory Monitoring System for Physiological Monitoring & Resource Planning in the ICU

Natural VTA activity during NREM sleep influences future exploratory behavior

Evaluation of machine learning methods for prospectivity modeling and vectoring of IOCG mineralization

Contact Info

Product

Resources

About