Dynamic Clinical States and Transitions During the First 72 Hours of Intensive Care After Acute Stroke
Abstract
BackgroundThe condition of a patient with acute stroke often changes within hours of ICU admission. Prognostic work here targets fixed endpoints predicted from admission data, and trajectory phenotyping assigns one label per patient. We used longitudinal ICU data to identify interpretable dynamic clinical states, characterize transitions between them, and relate the current state to later events.
MethodsRetrospective cohort study of 6368 adults with acute stroke in MIMIC-IV v3.1. The first 72 h were divided into twelve 6-hour windows, and a hidden Markov model was fitted to 21 neurological, physiological and organ-support variables. State number was chosen against criteria fixed before fitting: statistical fit, restart stability, state occupancy and clinical interpretability. Generalized estimating equations related the current state to new mechanical ventilation and vasopressor use within 12 h, and to ICU death within 72 h. Eleven sensitivity analyses assessed the robustness of the state solution.
ResultsFour states were selected: neurologically preserved-low support, neurological impairment- low support, impairment-renal dysfunction and impairment-respiratory support (63.3%, 7.8%, 11.8% and 17.1% of windows). Within 72 h, 40.3% of patients changed state at least once, and transitions ran in both directions rather than along a single severity gradient. States were identified without outcome data, yet ICU mortality by last state ranged from 2.9% to 43.9%. Adjusted for age, sex, subtype and Charlson index, the current state remained associated with organ-support escalation and death. State prevalence differed by at most 1.1 percentage points between training and test sets, and 10 of 11 sensitivity analyses gave a stable four-state solution (ARI 0.754-0.955).
ConclusionsThe early ICU course of acute stroke can be represented as movement among a small number of clinically interpretable states. The representation was reproducible in a held-out set and across admission eras, but requires validation in an independent database before any clinical use.
Related
- Dynamic volatolomic profiling reveals treatment-induced metabolic state transitions during radiosensitization in colorectal cancer
- Multimorbidity and dynamic transitions of low back pain states in older adults: a multistate markov analysis of the health and retirement study
- External validation of dynamic clinical states in acute stroke: transportability and independent rediscovery across 176 hospitals
- Baseline and Dynamic Clinical-Laboratory Risk Models for Early Progression or Death During PD-(L)1 Blockade in Advanced NSCLC: Development and Internal Validation.
- Association of the triglyceride-glucose index with dynamic bidirectional transitions of cardiovascular-kidney-metabolic syndrome: a multistate Markov model study based on CHARLS
BibTeX
@misc{p2026dynamic,
title = {{Dynamic Clinical States and Transitions During the First 72 Hours of Intensive Care After Acute Stroke}},
author = {P. LEI and Y. XU and Y. ZHANG},
howpublished = {medRxiv},
year = {2026},
month = aug,
doi = {10.64898/2026.08.30.26361738},
url = {https://www.medrxiv.org/content/10.64898/2026.08.30.26361738v1},
note = {CC BY-NC-ND},
}
RIS
TY - UNPB
TI - Dynamic Clinical States and Transitions During the First 72 Hours of Intensive Care After Acute Stroke
AU - P. LEI
AU - Y. XU
AU - Y. ZHANG
PY - 2026
DA - 2026/08/30
PB - medRxiv
DO - 10.64898/2026.08.30.26361738
UR - https://www.medrxiv.org/content/10.64898/2026.08.30.26361738v1
AB - BackgroundThe condition of a patient with acute stroke often changes within hours of ICU admission. Prognostic work here targets fixed endpoints predicted from admission data, and trajectory phenotyping assigns one label per patient. We used longitudinal ICU data to identify interpretable dynamic clinical states, characterize transitions between them, and relate the current state to later events. MethodsRetrospective cohort study of 6368 adults with acute stroke in MIMIC-IV v3.1. The first 72 h were divided into twelve 6-hour windows, and a hidden Markov model was fitted to 21 neurological, physiological and organ-support variables. State number was chosen against criteria fixed before fitting: statistical fit, restart stability, state occupancy and clinical interpretability. Generalized estimating equations related the current state to new mechanical ventilation and vasopressor use within 12 h, and to ICU death within 72 h. Eleven sensitivity analyses assessed the robustness of the state solution. ResultsFour states were selected: neurologically preserved-low support, neurological impairment- low support, impairment-renal dysfunction and impairment-respiratory support (63.3%, 7.8%, 11.8% and 17.1% of windows). Within 72 h, 40.3% of patients changed state at least once, and transitions ran in both directions rather than along a single severity gradient. States were identified without outcome data, yet ICU mortality by last state ranged from 2.9% to 43.9%. Adjusted for age, sex, subtype and Charlson index, the current state remained associated with organ-support escalation and death. State prevalence differed by at most 1.1 percentage points between training and test sets, and 10 of 11 sensitivity analyses gave a stable four-state solution (ARI 0.754-0.955). ConclusionsThe early ICU course of acute stroke can be represented as movement among a small number of clinically interpretable states. The representation was reproducible in a held-out set and across admission eras, but requires validation in an independent database before any clinical use.
ER -