scholarly journals Learning to Predict in-hospital mortality risk in intensive care unit with attention-based Temporal convolution network

2020 ◽  
Author(s):  
Yu-wen Chen ◽  
Yu-jie Li ◽  
Zhi-yong Yang ◽  
Kun-hua Zhong ◽  
Li-ge Zhang ◽  
...  
Keyword(s):  
Time Series ◽  
Mortality Risk ◽  
Time Series Data ◽  
Risk Model ◽  
Series Data ◽  
Dynamic Prediction ◽  
Physiological Variables ◽  
Prediction Of Mortality ◽  
Low Sensitivity ◽  
Mimic Iii

Abstract Background: Dynamic prediction of patients’ mortality risk in ICU with time series data is limited due to the high dimensionality, uncertainty with sampling intervals, and other issues. New deep learning method, temporal convolution network (TCN), makes it possible to deal with complex clinical time series data in ICU. We aimed to develop and validate it to predict mortality risk using time series data from MIMIC III dataset. Methods: 21139 records of ICU stays were analyzed and in total 17 physiological variables from the MIMIC III dataset were used to predict mortality risk. Then we compared the model performances of attention-based TCN with traditional artificial intelligence (AI) method. Results: Area Under Receiver Operating Characteristic (AUCROC) and Area Under Precision-Recall curve (AUC-PR) of attention-based TCN for predicting the mortality risk 48h after ICU admission were 0.837(0.824 -0.850) and 0.454. The sensitivity and specificity of attention-based TCN were 67.1% and 82.6%, compared to the traditional AI method yield low sensitivity (<50%). Conclusions: Attention-based TCN model achieved better performance in prediction of mortality risk with time series data than traditional AI methods and conventional score-based models. Attention-based TCN mortality risk model has the potential for helping decision-making in critical patients.

scholarly journals Dynamic predict in-hospital mortality risk in intensive care unit with a new deep learning of artificial intelligence

2020 ◽  
Author(s):  
Yu-wen Chen ◽  
Yu-jie Li ◽  
Zhi-yong Yang ◽  
Kun-hua Zhong ◽  
Li-ge Zhang ◽  
...  
Keyword(s):  
Artificial Intelligence ◽  
Time Series ◽  
Deep Learning ◽  
Mortality Risk ◽  
Time Series Data ◽  
Risk Model ◽  
Series Data ◽  
Physiological Variables ◽  
Prediction Of Mortality ◽  
Mimic Iii

Abstract Background Dynamic prediction of patients’ mortality risk in ICU with time series data is limited due to the high dimensionality, uncertainty with sampling intervals, and other issues. New deep learning method, temporal convolution network (TCN), makes it possible to deal with complex clinical time series data in ICU. We aimed to develop and validate it to predict mortality risk using time series data from MIMIC III dataset. Methods Finally, 21139 records of ICU stays were analyzed and in total 17 physiological variables from the MIMIC III dataset were used to predict mortality risk. Then we compared the model performances of attention-based TCN with traditional artificial intelligence (AI) method. Results The Area Under Receiver Operating Characteristic (AUCROC) and Area Under Precision-Recall curve (AUC-PR) of attention-based TCN for predicting the mortality risk 48 h after ICU admission were 0.837(0.824–0.850) and 0.454. The sensitivity and specificity of attention-based TCN were 67.1% and 82.6%, compared to the traditional AI method yield low sensitivity (< 50%). Conclusions Attention-based TCN model achieved better performance in prediction of mortality risk with time series data than traditional AI methods and conventional score-based models. Attention-based TCN mortality risk model has the potential for helping decision-making in critical patients.

scholarly journals Fractal Complexity of Daily Physical Activity Patterns Differs With Age Over the Life Span and Is Associated With Mortality in Older Adults

2018 ◽  
Vol 74 (9) ◽  
pp. 1461-1467 ◽  
Author(s):  
David A Raichlen ◽  
Yann C Klimentidis ◽  
Chiu-Hsieh Hsu ◽  
Gene E Alexander
Keyword(s):  
Physical Activity ◽  
Older Adults ◽  
Time Series ◽  
Time Scales ◽  
Mortality Risk ◽  
Time Series Data ◽  
Series Data ◽  
Fluctuation Analysis ◽  
Wide Range ◽  
Wearable Accelerometers

Abstract Background Accelerometers are included in a wide range of devices that monitor and track physical activity for health-related applications. However, the clinical utility of the information embedded in their rich time-series data has been greatly understudied and has yet to be fully realized. Here, we examine the potential for fractal complexity of actigraphy data to serve as a clinical biomarker for mortality risk. Methods We use detrended fluctuation analysis (DFA) to analyze actigraphy data from the National Health and Nutrition Examination Survey (NHANES; n = 11,694). The DFA method measures fractal complexity (signal self-affinity across time-scales) as correlations between the amplitude of signal fluctuations in time-series data across a range of time-scales. The slope, α, relating the fluctuation amplitudes to the time-scales over which they were measured describes the complexity of the signal. Results Fractal complexity of physical activity (α) decreased significantly with age (p = 1.29E−6) and was lower in women compared with men (p = 1.79E−4). Higher levels of moderate-to-vigorous physical activity in older adults and in women were associated with greater fractal complexity. In adults aged 50–79 years, lower fractal complexity of activity (α) was associated with greater mortality (hazard ratio = 0.64; 95% confidence interval = 0.49–0.82) after adjusting for age, exercise engagement, chronic diseases, and other covariates associated with mortality. Conclusions Wearable accelerometers can provide a noninvasive biomarker of physiological aging and mortality risk after adjusting for other factors strongly associated with mortality. Thus, this fractal analysis of accelerometer signals provides a novel clinical application for wearable accelerometers, advancing efforts for remote monitoring of physiological health by clinicians.

scholarly journals Using patient biomarker time series to determine mortality risk in hospitalised COVID-19 patients: a comparative analysis across two New York hospitals

2021 ◽  
Author(s):  
Ben Lambert ◽  
Isaac J. Stopard ◽  
Amir Momeni-Boroujeni ◽  
Rachelle Mendoza ◽  
Alejandro Zuretti
Keyword(s):  
New York ◽  
Time Series ◽  
Mortality Risk ◽  
Time Series Data ◽  
Medical Center ◽  
New York State ◽  
Prognostic Models ◽  
Series Data ◽  
Clinical Measurements ◽  
The Individual

AbstractA large range of prognostic models for determining the risk of COVID-19 patient mortality exist, but these typically restrict the set of biomarkers considered to measurements available at patient admission. Additionally, many of these models are trained and tested on patient cohorts from a single hospital, raising questions about the generalisability of results. We used a Bayesian Markov model to analyse time series data of biomarker measurements taken throughout the duration of a COVID-19 patient’s hospitalisation for n = 1540 patients from two hospitals in New York: State University of New York (SUNY) Downstate Health Sciences University and Maimonides Medical Center. In doing so, we quantified the mortality risk associated with both static (e.g. demographic and patient history variables) and dynamic factors (e.g. changes in biomarkers) throughout hospitalisation. By using our model to make predictions across the hospitals, we assessed how predictive factors generalised between the two cohorts. The individual dynamics of the measurements and their associated mortality risk were remarkably consistent across the hospitals. The model accuracy in predicting patient outcome (death or discharge) was 72.3% (predicting SUNY; posterior median accuracy) and 71.4% (predicting Maimonides) respectively. Model sensitivity was higher for detecting patients who would go on to be discharged (79.2%) versus those who died (61.0%). Our results indicate the utility of including dynamic clinical measurements when assessing patient mortality risk but also highlight the difficulty of identifying high risk patients.

Heat-mortality risk and the population concentration of metropolitan areas in Japan: a nationwide time-series study

2020 ◽  
Author(s):  
Whanhee Lee ◽  
Kristie L Ebi ◽  
Yoonhee Kim ◽  
Masahiro Hashizume ◽  
Yasushi Honda ◽  
...  
Keyword(s):  
Time Series ◽  
Mortality Risk ◽  
Time Series Data ◽  
Metropolitan Areas ◽  
Population Increase ◽  
Medical Services ◽  
Series Data ◽  
Population Concentration ◽  
Series Study

Abstract Background The complex role of urbanisation in heat-mortality risk has not been fully studied. Japan has experienced a rapid population increase and densification in metropolitan areas since the 2000s; we investigated the effects of population concentration in metropolitan areas on heat-mortality risk using nationwide data. Methods We collected time-series data for mortality and weather variables for all 47 prefectures in Japan (1980–2015). The prefectures were classified into three sub-areas based on population size: lowest (&lt;1 500 000), intermediate (1 500 000 to 3 000 000), and highest (&gt;3 000 000; i.e. metropolitan areas). Regional indicators associated with the population concentration of metropolitan areas were obtained. Results Since the 2000s, the population concentration intensified in the metropolitan areas, with the highest heat-mortality risk in prefectures with the highest population. Higher population density and apartment % as well as lower forest area and medical services were associated with higher heat-mortality risk; these associations have generally become stronger since the 2000s. Conclusions Population concentration in metropolitan areas intensified interregional disparities in demography, living environments, and medical services in Japan; these disparities were associated with higher heat-mortality risk. Our results can contribute to policies to reduce vulnerability to high temperatures.

Graphical Exploratory Data Analysis for Categorical Longitudinal and Time Series Data

2013 ◽  
Author(s):  
Stephen J. Tueller ◽  
Richard A. Van Dorn ◽  
Georgiy Bobashev ◽  
Barry Eggleston
Keyword(s):  
Time Series ◽  
Data Analysis ◽  
Exploratory Data Analysis ◽  
Time Series Data ◽  
Series Data ◽  
Exploratory Data

Faktor-Faktor Yang Mempengaruhi Nilai Tukar Dollar Amerika Serikat Terhadap Rupiah Tahun 2000–2013

2017 ◽  
Vol 1 (2) ◽  
pp. 177-191
Author(s):  
Rizki Rahma Kusumadewi ◽  
Wahyu Widayat
Keyword(s):  
Time Series ◽  
Exchange Rate ◽  
Money Supply ◽  
Time Series Data ◽  
The United States ◽  
Economic Conditions ◽  
Series Data ◽  
Arch Model ◽  
United States Dollar ◽  
The Exchange Rate

Exchange rate is one tool to measure a country’s economic conditions. The growth of a stable currency value indicates that the country has a relatively good economic conditions or stable. This study has the purpose to analyze the factors that affect the exchange rate of the Indonesian Rupiah against the United States Dollar in the period of 2000-2013. The data used in this study is a secondary data which are time series data, made up of exports, imports, inflation, the BI rate, Gross Domestic Product (GDP), and the money supply (M1) in the quarter base, from first quarter on 2000 to fourth quarter on 2013. Regression model time series data used the ARCH-GARCH with ARCH model selection indicates that the variables that significantly influence the exchange rate are exports, inflation, the central bank rate and the money supply (M1). Whereas import and GDP did not give any influence.

An Anomaly Detection Method with Exemplar Subsequence for Time Series Data

2016 ◽  
Vol 136 (3) ◽  
pp. 363-372
Author(s):  
Takaaki Nakamura ◽  
Makoto Imamura ◽  
Masashi Tatedoko ◽  
Norio Hirai
Keyword(s):  
Time Series ◽  
Anomaly Detection ◽  
Time Series Data ◽  
Detection Method ◽  
Series Data

The Comparison of Land Subsidence between East and West Side of Bangkok, Thailand

2020 ◽  
Vol 17 (3) ◽  
pp. 1
Author(s):  
Angkana Pumpuang ◽  
Anuphao Aobpaet
Keyword(s):  
Time Series ◽  
Land Subsidence ◽  
Time Series Data ◽  
Series Data ◽  
Industrial Estates ◽  
Vertical Displacements ◽  
Land Deformation ◽  
Extensive Farming ◽  
Chao Phraya River ◽  
Set Up

The land deformation in line of sight (LOS) direction can be measured using time series InSAR. InSAR can successfully measure land subsidence based on LOS in many big cities, including the eastern and western regions of Bangkok which is separated by Chao Phraya River. There are differences in prosperity between both sides due to human activities, land use, and land cover. This study focuses on the land subsidence difference between the western and eastern regions of Bangkok and the most possible cause affecting the land subsidence rates. The Radarsat-2 single look complex (SLC) was used to set up the time series data for long term monitoring. To generate interferograms, StaMPS for Time Series InSAR processing was applied by using the PSI algorithm in DORIS software. It was found that the subsidence was more to the eastern regions of Bangkok where the vertical displacements were +0.461 millimetres and -0.919 millimetres on the western and the eastern side respectively. The districts of Nong Chok, Lat Krabang, and Khlong Samwa have the most extensive farming area in eastern Bangkok. Besides, there were also three major industrial estates located in eastern Bangkok like Lat Krabang, Anya Thani and Bang Chan Industrial Estate. By the assumption of water demand, there were forty-eight wells and three wells found in the eastern and western part respectively. The number of groundwater wells shows that eastern Bangkok has the demand for water over the west, and the pumping of groundwater is a significant factor that causes land subsidence in the area.Keywords: Subsidence, InSAR, Radarsat-2, Bangkok

Sign in / Sign up

Export Citation Format

Share Document