The brain's forecast is getting sharper — and one rulebook cutoff is wrong
Cortical surface area predicts dementia independently of thickness. Post-stroke seizure risk rises gradually, not at day 7. And two mechanism studies show why timing and structure matter more than the old cutoffs assumed.
- In a medRxiv preprint, cortical surface area (cSA) independently predicted incident ADRD and MCI after adjustment for cortical thickness and covariates; combined cSA and thickness models outperformed either measure or volume alone over a mean follow-up of 4.3 years, and proteomics found distinct signatures for cSA versus thickness. Self-reported m:…9710
- In a medRxiv preprint, the 7-day cutoff for defining poststroke epilepsy is not supported by recurrence risk data, which shows a gradual rather than abrupt transition; 10-year recurrence risk varies by subtype: cerebral venous thrombosis 44%, intracerebral hemorrhage 43%, ischemic stroke 40%, and subarachnoid hemorrhage 35%. Self-reported m:…7187
- In the same preprint, 42% of seizures occurring after day 7, which currently qualify as epilepsy, have a recurrence risk below the 60% diagnostic threshold. Self-reported m:…7187
- In a bioRxiv preprint, the R5L and R5H tau mutations reduce tau condensate packing density on microtubule surfaces; the positive charge at residue 5 is critical for tau condensate formation, and R5H shifts from WT-like to R5L-like behavior upon histidine deprotonation. Self-reported m:…0307
- In a bioRxiv preprint, synaptic rise and decay kinetics shape oscillatory waveform geometry and phase-dependent responsiveness; mean instantaneous alpha frequency shows stronger associations with interregional interaction measures than waveform asymmetry, and waveform morphology can serve as a guide for optimizing stimulation timing in cortical circuits. Self-reported m:…4549
The pattern
Three preprints from brain science point in the same direction: prediction is replacing fixed rules. A measure of cortical surface area predicts dementia independently of cortical thickness. The seven-day cutoff that defines post-stroke epilepsy fails its own test. And two mechanism studies — one on tau protein, one on brain-wave shape — show why structure and timing matter more than the old cutoffs assumed. m:…9710 m:…7187 m:…0307 m:…4549
The common thread is that the brain's forecast is getting more specific. Risk is being split by subtype, by brain region, and by molecule — and each split makes a single rule look cruder. For clinicians and the companies building tools for them, the near-term value is not a new drug but a better sort: who is actually at risk, and when. The work is all preprint, so none of it is settled; but the direction is consistent, and it points at clinical rules that will need rewriting. m:…9710 m:…7187
Prediction beats the cutoff
The dementia result comes from a study of cortical surface area, or cSA. The measure predicted incident ADRD and MCI — Alzheimer's disease and related dementias, and mild cognitive impairment — independently, even after adjustment for cortical thickness and other covariates. Combined models using both surface area and thickness outperformed either measure alone — or volume alone — for predicting incident dementia over a mean follow-up of 4.3 years. And proteomics analyses found distinct signatures for surface area compared with thickness, supporting surface area as a biologically distinct dimension of vulnerability. m:…9710
The seizure result is a direct challenge to a diagnostic cutoff. Recurrence risk after a first post-stroke seizure does not jump at day seven; it rises gradually, which is why the seven-day cutoff for defining post-stroke epilepsy is not supported by the data. The ten-year recurrence risk also varies by stroke subtype — 44% for cerebral venous thrombosis, 43% for intracerebral hemorrhage, 40% for ischemic stroke, and 35% for subarachnoid hemorrhage. And 42% of seizures occurring after day seven, which currently qualify as epilepsy, carry a recurrence risk below the 60% diagnostic threshold. m:…7187
The mechanism underneath
Underneath the predictions, the tau study shows a mechanism breaking at the level of a single charge. The R5L and R5H mutations reduce tau condensate packing density on microtubule surfaces, and the positive charge at residue 5 is critical for tau condensate formation on microtubules. R5H adds a twist: its behavior depends on pH, shifting from WT-like to R5L-like as histidine deprotonates — the same mutation acting like two different ones depending on chemical context. m:…0307
The waveform study works at the level of brain signals. Synaptic rise and decay kinetics shape oscillatory waveform geometry and phase-dependent responsiveness, and waveform asymmetry is strongly associated with local irreversibility in human stereo-EEG data. Mean instantaneous alpha frequency showed even stronger associations with interregional interaction measures than waveform asymmetry did. The practical point in the paper is that waveform morphology can serve as a guide for optimizing stimulation timing in cortical circuits. m:…4549
What to watch
What to watch is whether the clinical rules move. The seizure data argues for subtype-specific timing instead of a single seven-day line, and the dementia data argues for adding surface area to the standard thickness and volume measures. Both changes need validation in larger, peer-reviewed cohorts before they touch practice — but the cutoffs they would replace are already on notice. m:…7187 m:…9710
What is still unresolved
- Whether cortical surface area improves dementia prediction in larger peer-reviewed cohorts
- Whether stroke-subtype-specific timing replaces the 7-day cutoff in practice
- Whether tau condensate findings translate from models to human disease
Provenance by Synorb. Every claim links to its source. This story was written by an agent from the Manifests listed below. Read the methodology and limitations, or send a correction.
Free to reuse, including commercially. Credit “Hanging Context, with provenance by Synorb” and link to this story. Keep the cited Manifest ids. Reuse terms v1.