Tough Questions with Mathieu
Mathieu Galtier, Head of Algorithms and Research.
Why is it definitely not always easy working at a startup.
There’s the long hours spent at the office, making things just perfect for the next day; the creativity and commitment it takes to actually produce a product; and the uncertainty of not knowing what tomorrow will bring. Our team takes it all in stride, though— they’re all committed to optimizing sleep and creating a movement around better rest.
To get to the bottom of what really goes on at the Rythm offices, we asked Head of Algorithms and Research, Mathieu Galtier, why he joined the company and how we do what we do every day.
What is the most exciting thing about working at Rythm?
The most interesting thing about working at Rythm is being able to participate to the redefinition of science. We design algorithms to really help people sleep better, which will not just stay in the lab or in an academic paper — we are actually taking science from labs to people. This also opens up the possibility for much wider analyses, which have never been done before.
It’s the first time humans have managed to record sleep over such a large sample of the population and for such a long time. We are creating an entirely new approach to understand sleep!
How do you accurately track sleep using an EEG device?
When one falls asleep, one’s brain does not shut down. On the contrary, sleep is an incredibly active state, where neurons are firing constantly. The two main neuron firing patterns correspond to deep sleep and REM sleep.
For instance, deep sleep is the time for dramatic synchronization in cortical neurons. Neurons alternate with ~0.5 seconds excitation phases and ~0.5 seconds of complete silence. These slow oscillations are synchronized over a large part of the cortex and they generate an electric field that can be measured even outside the skull.
The electrodes on the Dreem headband measure this information. We can track the average of the excitation of hundreds of millions of neurons at the scale of a millisecond. This makes it possible to identify which sleep stage the dreamer is experiencing.
How does Dreem improve deep sleep quality?
Dreem focuses on auditory stimulation to engage deep sleep. Stress, aging, and neurodegenerative diseases tend to desynchronize the cortex during deep sleep. The large and global slow oscillations are not so large and global anymore.
The idea with Dreem is to send short sounds (~100ms) at the precise moment needed to strengthen the slow oscillations. With our advanced machine learning algorithm, we identify the ideal timing of stimulation right at the beginning of certain slow oscillations. This raises the excitability of both the cortical area and the auditory area.
Thus, the waves of neural excitation propagate more easily and on a larger scale. Deep sleep is better quality. This method is based on a large body of academic research and the recent findings of several recognized professors like Jan Born (studies), Ken Paller and Giulio Tononi.