Critical Illness
Conditions
Keywords
Intensive Care Unit, Physiological State Space, Trajectory, Electronic Health Records, Physiological Resilience
Brief summary
Critically ill patients can deteriorate rapidly across multiple organ systems. Most intensive care unit (ICU) risk tools rely on measurements collected at a single time point and may not fully capture how physiology evolves or how quickly a patient recovers from disturbance. This single-center retrospective observational study will use routinely collected data from the ICCA reporting database at Zhongshan Hospital, Fudan University to develop and internally validate a research prototype called the ICU Physiological State Space Monitor. Adult ICU admissions will be represented as daily state vectors across 10 physiological domains. The study will characterize each patient's position and movement in a multidimensional state space, identify high-risk regions and possible critical transitions, and quantify physiological resilience using trajectory features such as variability, autocorrelation, curvature, recovery slope, and cross-domain coupling. The primary validation outcome is a composite clinical deterioration event within 72 hours after an eligible index patient-day. The monitor is an analytic and visualization framework for retrospective research and will not be deployed for real-time clinical decision-making in this study.
Detailed description
This is a single-center, retrospective, non-interventional methodological cohort study using routinely collected structured data from the ICCA reporting-layer database at Zhongshan Hospital, Fudan University. The source population comprises adult ICU patients admitted between September 2021 and May 31, 2026 who have an extractable and uniquely identifiable ICU encounter. No intervention will be assigned, no clinical decision will be altered, and no additional examination, treatment, follow-up visit, or biospecimen collection will be performed. The unit of participant-level registration is the patient, while the principal analytic structure is the ICU encounter and the patient-day. ICU admission time will serve as the common temporal anchor. Monitoring, laboratory, fluid, medication, organ-support, diagnostic, and demographic data will be extracted from the ICCA reporting layer and aggregated primarily into consecutive 24-hour windows. Alternative 12-hour or shorter windows may be evaluated in sensitivity analyses. Daily physiological state vectors will be constructed across 10 domains: oxygenation; ventilation and respiratory drive; hemodynamic perfusion; renal-fluid balance; hepatic-metabolic clearance; inflammation-immune activation; coagulation-blood integrity; brain-autonomic regulation; bioenergetic and acid-base status; and treatment-support burden. Prespecified data-governance rules will be used for variable-source mapping, unit harmonization, time alignment, limited carry-forward, missing-data handling, and physiological-range checks. Principal component analysis will provide the primary low-dimensional state-space representation. UMAP may be used for supplementary visualization and local-structure exploration but will not be the sole primary analytical framework. Enhanced trajectory features will include state position, displacement, trajectory length, speed, acceleration, turning angle, curvature, local variability, lag-1 autocorrelation, recovery slope, and cross-domain coupling. These features will be used to characterize high-risk regions, possible basin crossings and critical transitions, and physiological resilience. Eligible records will be divided chronologically into an earlier derivation cohort and a later temporal validation cohort, with an intended split of approximately 70% and 30%, respectively. The exact cutoff will be locked before analysis. The primary analysis will evaluate the association between state-space and trajectory features and a composite clinical deterioration outcome occurring within 72 hours after each eligible index patient-day. Multivariable logistic regression or discrete-time risk models will be used for the primary outcome. ICU and in-hospital mortality may be evaluated using Cox regression or competing-risk methods, and other outcomes will be analyzed with appropriate regression or time-to-event models. Internal validation will use the temporal validation cohort and bootstrap resampling to assess discrimination, calibration, and robustness. The source data will remain in the hospital-controlled information environment. Investigators will use a deidentified research dataset without direct identifiers and without access to the reidentification key. Only aggregated results, model parameters, and approved figures will be released. The study seeks a waiver of written informed consent because it is retrospective, non-interventional, uses deidentified existing data, and does not affect participants' current or future care.
Interventions
OTHERMultidimensional ICU Physiological State and Trajectory Features
The exposure of interest is the participant's multidimensional physiological state and trajectory derived from routinely collected ICU monitoring, laboratory, fluid, medication, organ-support, diagnostic, and demographic data. Daily state vectors will cover 10 physiological domains. Enhanced trajectory features will include state position, displacement, trajectory length, speed, acceleration, turning angle, curvature, local variability, lag-1 autocorrelation, recovery slope, and cross-domain coupling. The investigators will not assign any exposure, treatment, or clinical intervention.
Sponsors
Shanghai Zhongshan Hospital
Study design
Observational model
COHORT
Time perspective
RETROSPECTIVE
Eligibility
Sex/Gender
ALL
Age
18 Years to No maximum
Healthy volunteers
No
Inclusion criteria
* Age 18 years or older. * Admission to an ICU at Zhongshan Hospital, Fudan University between September 2021 and May 31, 2026, with a uniquely identifiable ICU encounter in the ICCA reporting-layer database. * ICU length of stay of at least 24 hours. * Availability of core variables from at least four physiological domains and at least one patient-day suitable for modeling.
Exclusion criteria
* Duplicate import of the same ICU encounter that cannot be reliably deduplicated. * Severe missingness, inconsistency, or irreconcilable error in key identifiers, ICU admission/discharge times, or core timestamps. * Test data, demonstration data, or records determined through data governance to be non-genuine patient records. * Inability to lock the primary source mapping for key variables, or overall data quality insufficient to support construction of the physiological state vector.
Design outcomes
Primary
| Measure | Time frame | Description |
|---|---|---|
| Composite Clinical Deterioration Within 72 Hours | Within 72 hours following each eligible index patient-day | A binary composite outcome defined by the occurrence of at least one of the following events during the 72-hour window after an eligible index patient-day: ICU death; new initiation of invasive mechanical ventilation; new initiation of continuous renal replacement therapy (CRRT); new initiation of extracorporeal membrane oxygenation (ECMO); or a prespecified significant escalation in vasoactive medication support. A participant/index window meeting more than one component will be counted once for the composite outcome. |
Secondary
| Measure | Time frame | Description |
|---|---|---|
| ICU mortality | From ICU admission through ICU discharge, an average of approximately 5 days | Death from any cause before discharge from the ICU. |
| In-Hospital Mortality | From hospital admission through hospital discharge, an average of approximately 30 days | Death from any cause before discharge from the index hospitalization. |
| Duration of Invasive Mechanical Ventilation | From ICU admission through ICU discharge, an average of approximately 5 days | Total number of calendar days during which invasive mechanical ventilation is used during the ICU stay. |
| Duration of Continuous Renal Replacement Therapy | From ICU admission through ICU discharge, an average of approximately 5 days | Total number of calendar days during which CRRT is used during the ICU stay. |
| Duration of Vasoactive Medication Exposure | From ICU admission through ICU discharge, an average of approximately 5 days | Total number of calendar days with recorded vasoactive medication support during the ICU stay. |
| ICU Length of Stay | From ICU admission through ICU discharge, an average of approximately 5 days | Time from ICU admission to ICU discharge, reported in days. |
| Hospital Length of Stay | From hospital admission through hospital discharge, an average of approximately 30 days | Time from hospital admission to hospital discharge, reported in days. |
Countries
China
Outcome results
None listed