A new method for spatio-temporal transmission prediction of COVID-19
© 2022 Elsevier Ltd. All rights reserved..
COVID-19 is the most serious public health event of the 21st century and has had a huge impact across the world. The spatio-temporal pattern analysis and simulation of epidemic spread have become the focus of current research. LSTM model has made a lot of achievements in the prediction of infectious diseases by virtue of its advantages in time prediction, but lacks the spatial expression. CA model plays an important role in epidemic spatial propagation modeling due to its unique evolution characteristics from local to global. However, no existing studies of CA have considered long-term dependence due to the impact of time changes on the evolution of the epidemic, and few have modeled using location data from actual diagnosed patients. Therefore, we proposed a LSTM-CA model to solve above mentioned problems. Base on the advantages of LSTM in temporal level and CA in spatial level, LSTM and CA are integrated from the spatio-temporal perspective of geography based on the fine-grained characteristics of epidemic data. The method divides the study area into regular grids, simulates the spatial interactions between neighborhood cells with the help of CA model, and extracts the parameters affecting the transition probability in CA with the help of LSTM model to assist evolution. Simulations are conducted in Python 3.4 to model the propagation of COVID-19 between Feb, 6 to Mar 20, 2020 in China. Experimental results show that, LSTM-CA performs a higher statistical accuracy than LSTM and spatial accuracy than CA, which could demonstrate the effectiveness of the proposed model. This method could be universal for the temporal and spatial transmission of major public health events. Especially in the early stage of the epidemic, we can quickly understand its development trend and cycle, so as to provide an important reference for epidemic prevention and control and public sentiment counseling.
Medienart: |
E-Artikel |
---|
Erscheinungsjahr: |
2023 |
---|---|
Erschienen: |
2023 |
Enthalten in: |
Zur Gesamtaufnahme - volume:167 |
---|---|
Enthalten in: |
Chaos, solitons, and fractals - 167(2023) vom: 25. Feb., Seite 112996 |
Sprache: |
Englisch |
---|
Beteiligte Personen: |
Wang, Peipei [VerfasserIn] |
---|
Links: |
---|
Themen: |
---|
Anmerkungen: |
Date Revised 30.01.2023 published: Print-Electronic Citation Status PubMed-not-MEDLINE |
---|
doi: |
10.1016/j.chaos.2022.112996 |
---|
funding: |
|
---|---|
Förderinstitution / Projekttitel: |
|
PPN (Katalog-ID): |
NLM351008616 |
---|
LEADER | 01000naa a22002652 4500 | ||
---|---|---|---|
001 | NLM351008616 | ||
003 | DE-627 | ||
005 | 20231226050308.0 | ||
007 | cr uuu---uuuuu | ||
008 | 231226s2023 xx |||||o 00| ||eng c | ||
024 | 7 | |a 10.1016/j.chaos.2022.112996 |2 doi | |
028 | 5 | 2 | |a pubmed24n1169.xml |
035 | |a (DE-627)NLM351008616 | ||
035 | |a (NLM)36589549 | ||
040 | |a DE-627 |b ger |c DE-627 |e rakwb | ||
041 | |a eng | ||
100 | 1 | |a Wang, Peipei |e verfasserin |4 aut | |
245 | 1 | 2 | |a A new method for spatio-temporal transmission prediction of COVID-19 |
264 | 1 | |c 2023 | |
336 | |a Text |b txt |2 rdacontent | ||
337 | |a ƒaComputermedien |b c |2 rdamedia | ||
338 | |a ƒa Online-Ressource |b cr |2 rdacarrier | ||
500 | |a Date Revised 30.01.2023 | ||
500 | |a published: Print-Electronic | ||
500 | |a Citation Status PubMed-not-MEDLINE | ||
520 | |a © 2022 Elsevier Ltd. All rights reserved. | ||
520 | |a COVID-19 is the most serious public health event of the 21st century and has had a huge impact across the world. The spatio-temporal pattern analysis and simulation of epidemic spread have become the focus of current research. LSTM model has made a lot of achievements in the prediction of infectious diseases by virtue of its advantages in time prediction, but lacks the spatial expression. CA model plays an important role in epidemic spatial propagation modeling due to its unique evolution characteristics from local to global. However, no existing studies of CA have considered long-term dependence due to the impact of time changes on the evolution of the epidemic, and few have modeled using location data from actual diagnosed patients. Therefore, we proposed a LSTM-CA model to solve above mentioned problems. Base on the advantages of LSTM in temporal level and CA in spatial level, LSTM and CA are integrated from the spatio-temporal perspective of geography based on the fine-grained characteristics of epidemic data. The method divides the study area into regular grids, simulates the spatial interactions between neighborhood cells with the help of CA model, and extracts the parameters affecting the transition probability in CA with the help of LSTM model to assist evolution. Simulations are conducted in Python 3.4 to model the propagation of COVID-19 between Feb, 6 to Mar 20, 2020 in China. Experimental results show that, LSTM-CA performs a higher statistical accuracy than LSTM and spatial accuracy than CA, which could demonstrate the effectiveness of the proposed model. This method could be universal for the temporal and spatial transmission of major public health events. Especially in the early stage of the epidemic, we can quickly understand its development trend and cycle, so as to provide an important reference for epidemic prevention and control and public sentiment counseling | ||
650 | 4 | |a Journal Article | |
650 | 4 | |a CA | |
650 | 4 | |a COVID-19 | |
650 | 4 | |a LSTM | |
650 | 4 | |a Modeling | |
650 | 4 | |a Spatio-temporal | |
700 | 1 | |a Liu, Haiyan |e verfasserin |4 aut | |
700 | 1 | |a Zheng, Xinqi |e verfasserin |4 aut | |
700 | 1 | |a Ma, Ruifang |e verfasserin |4 aut | |
773 | 0 | 8 | |i Enthalten in |t Chaos, solitons, and fractals |d 1995 |g 167(2023) vom: 25. Feb., Seite 112996 |w (DE-627)NLM114377014 |x 0960-0779 |7 nnns |
773 | 1 | 8 | |g volume:167 |g year:2023 |g day:25 |g month:02 |g pages:112996 |
856 | 4 | 0 | |u http://dx.doi.org/10.1016/j.chaos.2022.112996 |3 Volltext |
912 | |a GBV_USEFLAG_A | ||
912 | |a GBV_NLM | ||
951 | |a AR | ||
952 | |d 167 |j 2023 |b 25 |c 02 |h 112996 |