Symptom clusters in COVID-19: A potential clinical prediction tool from the COVID Symptom Study app

• DOI: 10.1126/sciadv.abd4177
• Publication Date: 2021-03-19T19:22:25Z
• Authors: Carole H. Sudre; Karla A. Lee; Mary Ni Lochlainn; Thomas Varsavsky; Benjamin Murray; Mark S. Graham; Cristina Menni; Marc Modat; Ruth C. E. Bowyer; Long H. Nguyen; David A. Drew; Amit D. Joshi; Wenjie Ma; Chuan-Guo Guo; Chun-Han Lo; Sajaysurya Ganesh; Abubakar Buwe; Joan Capdevila Pujol; Julien Lavigne du...; Alessia Visconti; Maxim B. Freidin; Julia S. El-Sayed...; Mario Falchi; Richard Davies; Maria F. Gomez; Tove Fall; M. Jorge Cardoso; Jonathan Wolf; Paul W. Franks; Andrew T. Chan; Tim D. Spector; Claire J. Steves; Sébastien Ourselin
• Keywords: Coronavirus

As no one symptom can predict disease severity or the need for dedicated medical support in coronavirus 2019 (COVID-19), we asked whether documenting time series over first few days informs outcome. Unsupervised clustering presentation was performed on data collected from a training dataset of completed cases enlisted early COVID Symptom Study Smartphone application, yielding six distinct presentations. Clustering validated an independent replication between 1 and 28 May 2020. Using 5 logging, ROC-AUC (receiver operating characteristic – area under curve) respiratory 78.8%, substantially outperforming personal characteristics alone (ROC-AUC 69.5%). Such approach could be used to monitor at-risk patients resource requirements before they are required.