Synapse
⌘+K
Synapse
PulseExploreJournal ClubResearchersJournals
Instagram
HomeJournal ClubExplore
June 9, 2025Open Access

Beyond proportional recovery in wake-up stroke: unsupervised recovery clusters based on the NIHSS

View Full Paper
Ask AI
Bookmark
Share

Why the study?

Concerns exist regarding linear models like the Proportional Recovery Rule, including methodological dependence of fitter classifications, mathematical coupling, and ceiling effects on clinical scales.

Does Repeated Spectral Clustering of NIHSS scores identify distinct recovery patterns beyond the traditional Proportional Recovery Rule in patients with wake-up stroke?

Population

201 patients from the WAKE-UP trial moderately impaired at onset and still impaired at 22/36 hours

Comparison

Repeated Spectral Clustering recovery patterns vs Proportional Recovery Rule linear fit

Design

Clustering analysis of clinical trial cohort

Follow-up

90 days

Key result

Repeated Spectral Clustering of NIHSS recovery ratios identified six distinct recovery clusters in wake-up stroke patients, revealing that most patients experience non-proportional recovery trajectories.

Authors

AZAndrea ZanolaABAntonio Luigi BisognoVVVeronika Vadinova

Discussion

Loading...

Member takes

Overview

Should not yet alter recovery expectations in wake-up stroke; leaves open whether spectral clustering refines prognostication beyond proportional models.

Study Design

Type

Observational (n=201)

Blinding

Double-blind

Randomization

Randomized

Multicenter

Yes

Structured PICO

Does Repeated Spectral Clustering of NIHSS scores identify distinct recovery patterns beyond the traditional Proportional Recovery Rule in patients with wake-up stroke?

P
Population
201 patients from the WAKE-UP trial with wake-up stroke, moderately impaired at onset (total NIHSS > 4) and still impaired at 22/36 hours (total NIHSS > 0).
I
Intervention
Repeated Spectral Clustering (RSC) applied to NIHSS recovery ratios (calculated from 22/36 hours to 90 days post-stroke).
C
Comparator
Proportional Recovery Rule (PRR) linear models.
O
Outcome
Recovery patterns identified by Repeated Spectral Clustering based on the recovery ratio of total NIHSS scores from 22/36 hours to 90 days post-stroke.surrogate

Unsupervised clustering of NIHSS recovery ratios identifies six distinct recovery trajectories in wake-up stroke patients, challenging the traditional dichotomous 'fitters vs. non-fitters' proportional recovery rule.

Limitations

  • For NIHSS at 22/36 hours less than 6, the models do not discriminate accurately between patients due to a ceiling effect.
  • The clustering approach was limited to only one variable (the recovery ratio).

Cite This Study

Zanola et al. (2025) conducted an observational in Wake-up stroke (n=201). Intravenous alteplase (rtPA) vs. Placebo was evaluated on Recovery ratio (RR) based on total NIHSS from 22/36 hours to 90 days. Repeated Spectral Clustering of NIHSS recovery ratios identified six distinct recovery clusters in wake-up stroke patients, revealing that most patients experience non-proportional recovery trajectories.

synapsesocial.com/papers/6a163fcbed257bd69ec519dchttps://doi.org/10.1101/2025.06.04.25328943
View Full Paper
Ask AI
Bookmark
Share