9–13 Sept 2024
Turin, Italy
Europe/Rome timezone

Machine learning time series causal effect analysis of the interplanetary magnetic field’s Bz component and geomagnetic Ap on cardiovascular health event frequency

Not scheduled
1h
Turin, Italy

Turin, Italy

Centro Congressi Unione Industriali Torino Via Vela, 17 - 10128 Torino
Poster Space weather and the solar-heliospheric connections Coffee break and poster session 2

Description

Associations between space weather and events on Earth, e.g., geomagnetic perturbations affecting power grids, are well-known. Effects on human health and physiology are less well investigated, and current evidence, suggesting, e.g., less frequent cardiovascular events during phases of high geomagnetic disturbance, builds on small patient cohorts and a wide array of statistical tests. Recently, the availability of large space weather time series along with huge epidemiologic observational databases facilitates evidence synthesis by employing advanced machine learning approaches to uncover associations between both.
The objective is to investigate potential causal effects of the solar wind’s southward IMF component and strong geomagnetic disturbance on cardiovascular health, in particular on cardiac arrest events as a hard endpoint.
Time series of the IMF’s Bz component (Wind satellite, L1) and Ap from the year 2015 were merged with emergency service use time data from the U.S. for cardiac arrest events. A causal impact analysis using counterfactual reasoning was performed to test for causal effects of the March 15 CME and the subsequent St. Patrick’s day G4 level geomagnetic superstorm, for a 14-day period after the CME.
Statistically significant, negative causal impact on cardiac arrest event frequency in the U.S. are identified for the southward (Bz) IMF component (-7% [-12%, -5%], p=0.001) and Ap (-9% [-10%, -4%], p=0.001).
Results show a significant impact of IMF and geomagnetic disturbance on cardiac arrest frequency in the U.S., potentially due to alterations in the activation of the autonomous nervous system. More research is necessary to uncover mechanistic models.

Primary author

Michael Marschollek (Hannover Medical School, TU Braunschweig)

Presentation materials

There are no materials yet.