Development of forecast models for COVID-19 hospital admissions using anonymized and aggregated mobile network data.
Jalil Taghia, Valentin Kulyk, Selim Ickin, Mats Folkesson, Cecilia Nyström, Kristofer Ȧgren, Thomas Brezicka, Tore Vingare, Julia Karlsson, Ingrid Fritzell, Ralph Harlid, Bo Palaszewski, Magnus Kjellberg, Jörgen Gustafsson
Author Information
Jalil Taghia: Ericsson Research, Ericsson, 164 40, Kista, Sweden. jalil.taghia@ericsson.com.
Valentin Kulyk: Ericsson Research, Ericsson, 164 40, Kista, Sweden.
Selim Ickin: Ericsson Research, Ericsson, 164 40, Kista, Sweden.
Mats Folkesson: Ericsson Research, Ericsson, 164 40, Kista, Sweden.
Cecilia Nyström: Ericsson Business Area Cloud Software and Services, Ericsson, 164 40, Kista, Sweden.
Kristofer Ȧgren: Telia Company AB, 169 94, Solna, Sweden.
Thomas Brezicka: Department of Quality and Patient Safety, Sahlgrenska University Hospital, 413 45, Gothenburg, Sweden.
Tore Vingare: Department of Analysis and Project Management, Sahlgrenska University Hospital, 413 45, Gothenburg, Sweden.
Julia Karlsson: Department of Analysis and Project Management, Sahlgrenska University Hospital, 413 45, Gothenburg, Sweden.
Ingrid Fritzell: Department of Analysis and Project Management, Sahlgrenska University Hospital, 413 45, Gothenburg, Sweden.
Ralph Harlid: Södra Älvsborgs Sjukhus, Hospital Management, 501 82, Borås, Sweden.
Bo Palaszewski: Department of Data Management and Analysis, Västra Götalandsregionen, 405 44, Gothenburg, Sweden.
Magnus Kjellberg: AI Competence Center, Sahlgrenska University Hospital, 413 45, Gothenburg, Sweden.
Jörgen Gustafsson: Ericsson Research, Ericsson, 164 40, Kista, Sweden. jorgen.gustafsson@ericsson.com.
Reliable forecast of COVID-19 hospital admissions in near-term horizons can help enable effective resource management which is vital in reducing pressure from healthcare services. The use of mobile network data has come to attention in response to COVID-19 pandemic leveraged on their ability in capturing people social behavior. Crucially, we show that there are latent features in irreversibly anonymized and aggregated mobile network data that carry useful information in relation to the spread of SARS-CoV-2 virus. We describe development of the forecast models using such features for prediction of COVID-19 hospital admissions in near-term horizons (21 days). In a case study, we verified the approach for two hospitals in Sweden, Sahlgrenska University Hospital and Södra Älvsborgs Hospital, working closely with the experts engaged in the hospital resource planning. Importantly, the results of the forecast models were used in year 2021 by logisticians at the hospitals as one of the main inputs for their decisions regarding resource management.
