Perme Score and Machine Learning for Detecting ICU-acquired Weakness: A Prospective Observational Cohort Study

Area of Science:

• Critical Care Medicine
• Neurology
• Physical Therapy

Background:

• Intensive care unit-acquired weakness (ICUAW) is prevalent in critically ill patients on prolonged mechanical ventilation (MV).
• Standard diagnostic tools like handgrip strength tests (HGST) are often impractical in critical care settings.
• Early identification of ICUAW is crucial for patient recovery and reducing intensive care unit (ICU) length of stay.

Purpose of the Study:

• To evaluate the Perme ICU Mobility Score as a screening tool for ICUAW.
• To determine the association between the Perme Score and ICUAW, using HGST as the diagnostic reference.
• To assess the utility of machine learning models in identifying ICUAW predictors.

Main Methods:

• A prospective observational study was conducted in a tertiary ICU involving adult patients on MV for at least 7 days.
• ICUAW was diagnosed using sex-specific HGST cutoffs.
• Logistic regression, LASSO, and Random Forest models were employed to analyze the association between the Perme Score and ICUAW.

Main Results:

• ICUAW was present in 78.4% of the 97 patients studied.
• Lower Perme Scores were significantly associated with ICUAW (p < 0.001).
• A Perme Score cutoff of ≤9 demonstrated 76.3% sensitivity and 71.4% specificity for detecting ICUAW.

Conclusions:

• The Perme Score is a feasible tool for screening ICUAW in mechanically ventilated patients.
• A Perme Score cutoff of ≤9 can aid in early functional risk stratification.
• Clinical context is essential when interpreting the Perme Score due to its limitations in specificity and negative predictive value.