LIORA
Privacy-first menstrual wellness and cycle forecasting application operating offline with zero cloud tracking.
The Idea
Menstrual tracking applications handle some of the most sensitive personal data imaginable. Yet, the vast majority of existing solutions operate as data-harvesting tools, monetizing user health information or leaving it vulnerable on centralized servers. LIORA was conceived from a simple conviction: wellness technology should serve the user, not the surveillance economy. The data must stay on the device.
The Problem
Existing tracking apps suffer from two major flaws: they are highly invasive regarding data privacy, and their prediction algorithms often rely on rigid, generalized calendars that fail to account for the unique, shifting biological variance of individual users. This creates a dual failure of trust and accuracy.
The Approach & The Hathaway Algorithm
To solve the prediction accuracy problem without compromising privacy, I developed the Hathaway Algorithm. Instead of transmitting data to a cloud-based machine learning model, LIORA processes predictions entirely on the device.
The algorithm uses adaptive probabilistic modelling to dynamically adjust predictions based on historical tracking data. It learns the user's specific variance patterns, shifting away from generic calendar-based predictions toward personalized, localized intelligence.
The System Architecture
LIORA is built with Flutter to provide a fluid, native-feeling experience across Android devices. The data layer relies on an encrypted local SQLite database. While the system operates 100% offline by default, users have the option to enable encrypted, client-controlled synchronization via Firebase if they require cross-device functionality.
Current State & Field Testing
LIORA has moved from a conceptual prototype to a field-tested application. It was rigorously tested with over 100 users across Kerala's educational institution network, allowing me to refine the Hathaway Algorithm against real-world biological variance.