This work was the first challenge of my PhD, supervised by @thomasandrillon.bsky.social, funded by the @erc.europa.eu, was done at the @institutducerveau.bsky.social. Thanks to the APHP Isabelle Arnulf, the clinicians in her team, the para-medical staff, and @inserm.fr.

12/ IH remains a diagnosis of exclusion—but maybe it doesn’t have to.

This work shows how quantitative EEG markers can help clarify what makes IH distinct.

🧵Thanks for reading!

📄 Read the preprint here: www.biorxiv.org/content/10.1...

🔁 Questions, and feedback welcome!

11/ Bottom line:

IH with long sleep time isn’t marked by “bad sleep” but by paradoxically deep and structured sleep, paired with excessive daytime sleepiness.

⏰ It’s time to look beyond the night

→ into daytime arousal systems and brain function during wakefulness.

10/ Can we use this info for diagnosis?

Using just one night of EEG, our model classified IH vs. controls with an AUC of 0.74 🎯

Still far from perfect, but it’s a step toward more objective, EEG-based diagnostics.

9/ Why does slow wave density matter?

Slow waves are linked to sleep inertia, and may reflect heightened sleep pressure.

Two (non-exclusive) hypotheses:

🧠 Increased synaptic plasticity during wake

🔥 Altered brain metabolism

Either way → sleep may be normal, but wake is off.

8/ We also explored slow wave activity.

💡 IH patients showed increased slow wave density during N3 despite no change in overall delta power.

This implies that global spectral analyses may miss subtle but meaningful changes in sleep microarchitecture.

7/ These changes were not mirrored by infraslow modulations of sigma power

📉 suggesting that while local sleep dynamics are altered, global regulatory rhythms (like those influenced by noradrenaline) may stay intact.

6/ Zooming into the microstructure 🔬

📈 IH patients had more sigma power & higher spindle density

🧠 Spindles were more clustered, pointing to over-stabilized thalamocortical networks

→ This might underlie severe sleep inertia and reduced responsiveness to stimuli.

5/ We also analyzed moment-to-moment variability (s/o @rherzoga.bsky.social).

➡️ IH patients showed less similarity between consecutive NREM epochs.

Not due to awakenings or hybrid stages like insomnia (see thread : shorturl.at/QDlol) or NT1,

but perhaps a different kind of sleep instability 🤔

4/ But sleep isn’t just sub-stages—sleep is a graded and dynamic process!

So, we turned to the analysis of hypnodensity 📊:

IH showed more mixed wake/N1 epochs, especially during intra-sleep awakenings.

⚠️ Are these a marker of sleep intrusions into wakefulness and difficulty fully waking up?

3/ First, macrostructure:

🛌 IH participants had more N3 and higher sleep efficiency

😴 Less REM and reduced wake after sleep onset

📉 These results might partly reflect the way sleep tests are run (e.g., waking at 6:30am for MSLT → less REM at the end of the night)

2/ IH with long sleep time is paradoxical: people sleep a lot but still wake up feeling unrested.

So we looked beyond classic metrics like REM %, sleep efficiency, or MSLT latency…

…and dove into sleep microarchitecture. 🧠💤