Towards Early Public Health Emergency Anticipating with Emerging Non-health Data Per se

Abstract

AbstractPhone calls and text messages provide metadata, and Telecom companies record these communications as Call Detail Records (CDR). Using anonymized CDR datasets enables mappings and early decisions in various domains, including public health. Thus, both transmitted and non-transmitted diseases spatio-temporal models may be improved. These CDR data can offer information on people’s movement patterns and interactions with fine-grained precision. With these available CDR datasets, data mining can be applied to identify population trends and produce movement models to simulate diseases. This enables policies to act with earlier decisions, especially in low- and middle-income countries (LMIC) where it is difficult to obtain valuable data for precise statistical analysis. The present study describes a relevant use case in public health where CDRs enables early warning fine-grained levels both at temporal and spatial levels. The passively gathered data from mobile phones, which act as sensors, allowed us to estimate the population in risk of burden in case of stroke in a country according to population mobility. The designed model integrates and crosslinks CDR data with mobile antenna localization, geolocalised hospitals, and freely available spatial data to perform population mappings over a country. As of result, using data about Senegal country, it has been possible to identify regions where more focus should be put on protecting population health.KeywordsCall data recordMobile dataPublic healthEmergency situations