Multidimensional Digital Phenotyping of High-Risk Chronic Disease Patients With Ventilator Dependence

This project plans to employ a multi-center retrospective and prospective cohort study design. It aims to collect data on ventilator support strategies, duration of invasive mechanical ventilation, incidence of ventilator dependence, high-risk factors for ventilator dependence, and in-hospital mortality rates among different chronic disease populations in the ICU. This will involve combining unstructured data with real-time bedside multi-dimensional high-frequency data (including dynamic changes in data volume, respiratory mechanics, diaphragm ultrasound, EIT, diaphragm electrical activity, and other monitoring parameters) to construct digital phenotypes for chronic disease patients with ventilator dependence and identify high-risk factors for ventilator dependence in this population.

Specifically, the research will:

Utilize an integrated modular intelligent respiratory monitoring system, previously developed by the project team, to achieve dynamic monitoring of multi-dimensional indicators.

Systematically collect dynamic clinical characteristics of mechanical ventilation dependence in chronic disease populations through retrospective and prospective cohort studies, and employ multivariate statistical analysis, machine learning, and other techniques to identify no fewer than 5-6 high-risk factors for ventilator dependence in chronic disease patients.

Establish a data ecosystem suitable for chronic disease patients undergoing mechanical ventilation, build a multi-dimensional high-frequency data platform for chronic ventilator-dependent populations, map the full cycle from intubation to mechanical ventilation support, weaning, and extubation, and construct multidimensional digital phenotypes for high-risk chronic disease patients with ventilator dependence.