tetano
Editor, Senior Moderator
Sci Adv
. 2021 Mar 19;7(12):eabd4177.
doi: 10.1126/sciadv.abd4177. Print 2021 Mar.
Symptom clusters in COVID-19: A potential clinical prediction tool from the COVID Symptom Study app
Carole H Sudre[SUP] 1 2 3 [/SUP], Karla A Lee[SUP] 4 [/SUP], Mary Ni Lochlainn[SUP] 4 [/SUP], Thomas Varsavsky[SUP] 5 [/SUP], Benjamin Murray[SUP] 5 [/SUP], Mark S Graham[SUP] 5 [/SUP], Cristina Menni[SUP] 4 [/SUP], Marc Modat[SUP] 5 [/SUP], Ruth C E Bowyer[SUP] 4 [/SUP], Long H Nguyen[SUP] 6 [/SUP], David A Drew[SUP] 6 [/SUP], Amit D Joshi[SUP] 6 [/SUP], Wenjie Ma[SUP] 6 [/SUP], Chuan-Guo Guo[SUP] 6 [/SUP], Chun-Han Lo[SUP] 6 [/SUP], Sajaysurya Ganesh[SUP] 7 [/SUP], Abubakar Buwe[SUP] 7 [/SUP], Joan Capdevila Pujol[SUP] 7 [/SUP], Julien Lavigne du Cadet[SUP] 7 [/SUP], Alessia Visconti[SUP] 4 [/SUP], Maxim B Freidin[SUP] 4 [/SUP], Julia S El-Sayed Moustafa[SUP] 4 [/SUP], Mario Falchi[SUP] 4 [/SUP], Richard Davies[SUP] 7 [/SUP], Maria F Gomez[SUP] 8 [/SUP], Tove Fall[SUP] 8 [/SUP], M Jorge Cardoso[SUP] 5 [/SUP], Jonathan Wolf[SUP] 7 [/SUP], Paul W Franks[SUP] 4 8 [/SUP], Andrew T Chan[SUP] 6 [/SUP], Tim D Spector[SUP] 4 [/SUP], Claire J Steves[SUP] 4 [/SUP], S?bastien Ourselin[SUP] 1 [/SUP]
Affiliations
Abstract
As no one symptom can predict disease severity or the need for dedicated medical support in coronavirus disease 2019 (COVID-19), we asked whether documenting symptom time series over the first few days informs outcome. Unsupervised time series clustering over symptom presentation was performed on data collected from a training dataset of completed cases enlisted early from the COVID Symptom Study Smartphone application, yielding six distinct symptom presentations. Clustering was validated on an independent replication dataset between 1 and 28 May 2020. Using the first 5 days of symptom logging, the ROC-AUC (receiver operating characteristic - area under the curve) of need for respiratory support was 78.8%, substantially outperforming personal characteristics alone (ROC-AUC 69.5%). Such an approach could be used to monitor at-risk patients and predict medical resource requirements days before they are required.
. 2021 Mar 19;7(12):eabd4177.
doi: 10.1126/sciadv.abd4177. Print 2021 Mar.
Symptom clusters in COVID-19: A potential clinical prediction tool from the COVID Symptom Study app
Carole H Sudre[SUP] 1 2 3 [/SUP], Karla A Lee[SUP] 4 [/SUP], Mary Ni Lochlainn[SUP] 4 [/SUP], Thomas Varsavsky[SUP] 5 [/SUP], Benjamin Murray[SUP] 5 [/SUP], Mark S Graham[SUP] 5 [/SUP], Cristina Menni[SUP] 4 [/SUP], Marc Modat[SUP] 5 [/SUP], Ruth C E Bowyer[SUP] 4 [/SUP], Long H Nguyen[SUP] 6 [/SUP], David A Drew[SUP] 6 [/SUP], Amit D Joshi[SUP] 6 [/SUP], Wenjie Ma[SUP] 6 [/SUP], Chuan-Guo Guo[SUP] 6 [/SUP], Chun-Han Lo[SUP] 6 [/SUP], Sajaysurya Ganesh[SUP] 7 [/SUP], Abubakar Buwe[SUP] 7 [/SUP], Joan Capdevila Pujol[SUP] 7 [/SUP], Julien Lavigne du Cadet[SUP] 7 [/SUP], Alessia Visconti[SUP] 4 [/SUP], Maxim B Freidin[SUP] 4 [/SUP], Julia S El-Sayed Moustafa[SUP] 4 [/SUP], Mario Falchi[SUP] 4 [/SUP], Richard Davies[SUP] 7 [/SUP], Maria F Gomez[SUP] 8 [/SUP], Tove Fall[SUP] 8 [/SUP], M Jorge Cardoso[SUP] 5 [/SUP], Jonathan Wolf[SUP] 7 [/SUP], Paul W Franks[SUP] 4 8 [/SUP], Andrew T Chan[SUP] 6 [/SUP], Tim D Spector[SUP] 4 [/SUP], Claire J Steves[SUP] 4 [/SUP], S?bastien Ourselin[SUP] 1 [/SUP]
Affiliations
- PMID: 33741586
- DOI: 10.1126/sciadv.abd4177
Abstract
As no one symptom can predict disease severity or the need for dedicated medical support in coronavirus disease 2019 (COVID-19), we asked whether documenting symptom time series over the first few days informs outcome. Unsupervised time series clustering over symptom presentation was performed on data collected from a training dataset of completed cases enlisted early from the COVID Symptom Study Smartphone application, yielding six distinct symptom presentations. Clustering was validated on an independent replication dataset between 1 and 28 May 2020. Using the first 5 days of symptom logging, the ROC-AUC (receiver operating characteristic - area under the curve) of need for respiratory support was 78.8%, substantially outperforming personal characteristics alone (ROC-AUC 69.5%). Such an approach could be used to monitor at-risk patients and predict medical resource requirements days before they are required.