Technical overview
The goal of Mode’s dosing platform is to allow users to vaporize a precise amount of oil concentrate during a puff without the need for proprietary cartridges. Achieving this goal presents a significant technical challenge, as the amount of vapor delivered over the course of a puff depends on a number of complex variables. To overcome this challenge, we have taken a data-driven approach, building machine learning models that take the above variables (“features”) as input and output a predicted dose.
On-Device Measurements
Measuring relevant signals using sensors that are directly integrated into the device
Ground-Truth Dosing Measurements
Collecting ground truth dosing measurements to generate a dataset for machine learning
Dosing Simulator
Improving feature coverage of our statistical models with a dosing simulator
Model optimization
Optimizing statistical models with a high-throughput machine learning pipeline
Concentrate-agnostic models
Improving puff simulations with data collected from human users
Continuously Improving
We strive for precision accuracy and consistently improving the user experience. We will continue to run tests to optimize our dosing algorithms,with improvements being sent over-the-air through the mode mobile apps.