How Dart Replaced Python: Offline Time Series Analysis on Flutter

We're all used to time series analysis being Python and its classic stack — statsmodels, sklearn, numpy, and the rest. This ecosystem has become the de facto standard, and it works perfectly — on a server with plenty of memory and compute power. But what if your app has to work offline, without any server access? What if the user needs data insights at every single moment, regardless of connection?

In this talk, I'll show how I solved this problem on Flutter — with no server calls and no Python backend — using my personal project as an example: an app for managing Type 1 Diabetes, which analyzes blood glucose levels represented as a time series.

The Dart ecosystem isn't empty when it comes to this, but it's far from sufficient either. On pub.dev, you can find packages for basic statistical methods that run directly on the device — a solid foundation. But for more advanced analysis, I used FFI to call C libraries from Dart, where the real heavy lifting tools for time series forecasting are already available.

The outcome of this research is a Dart package I authored — an SDK for offline time series forecasting — along with real-world performance metrics collected in a production business case.

My personal motivation for this work is helping my wife, who has Type 1 Diabetes, to anticipate dangerous blood sugar swings — especially at night, when it's easy to sleep through a critical health condition.

I'm confident this talk will be interesting both from a mobile development perspective (as an unusual, non‑standard use case) and from an ML angle, especially in time series analysis. I'll be happy to share my experience and research results with the community!