Modeling ICU occupancy during COVID-19

Over the past few weeks, the question we’ve all heard again and again is: How many ICU beds will we need tomorrow? As COVID-19 continues to put pressure on healthcare systems, reliable forecasts can make the difference between preparation and overwhelm.

Here at UCA Datalab, we’ve been working to answer that question with data. Alongside David Gómez-Ullate and Leopoldo Gutiérrez, I’ve helped develop a real-time model to forecast ICU occupancy by COVID-19 patients across Andalucía. The model uses publicly available data from Andalusian hospitals and is updated daily to provide decision-makers with fresh projections for critical care demand.

We made the results open and accessible online from day one. Because in a crisis like this, transparency matters as much as accuracy.

The work has been intense: just logistic curves, uncertainty intervals, and a flood of CSVs day after day. But it’s been deeply motivating. This might be one of the clearest cases I’ve experienced of data science being called into action, not in theory but in real time. And yes, we barely slept that week, but if a few late nights can help hospitals prepare, it’s a trade we’d make again without hesitation.

We hope this small contribution can help inform public health planning and ease the burden on our medical teams. We’ll keep improving the model and updating the data—one forecast at a time.

Stay safe,

Daniel Precioso