Machine learning-based analysis of factors influencing maximum tongue pressure in patients after cardiac surgery: a cross-sectional study

A low postoperative maximum tongue pressure (MTP) reflects diminished tongue muscle strength, which can compromise oral-phase swallowing function and lead to adverse clinical outcomes. However, the specific perioperative determinants of MTP in cardiac surgery patients remain poorly understood. This study aimed to identify the independent clinical predictors of postoperative MTP using a hybrid machine learning approach. In this cross-sectional study conducted at a teaching hospital in Shanghai from April 2025 to January 2026, a total of 470 adult patients undergoing cardiac surgery were included. MTP was assessed 8–24 h post-extubation, and comprehensive perioperative data were systematically extracted. To overcome multicollinearity and prevent model overfitting, a dual-algorithm feature selection was employed. Variables were ranked by a Random Forest (RF) algorithm, and optimal feature subsets were selected using Least Absolute Shrinkage and Selection Operator (LASSO) regression with 10-fold cross-validation. The overlapping core variables were subsequently incorporated into a multivariable stepwise linear regression model. A total of 470 patients were included. Based on the RF and LASSO models, 12 core features were extracted from 18 candidate variables. Multivariable linear regression confirmed five independent predictors. According to the variable importance ranking and standardized coefficients, prolonged endotracheal intubation duration ( β = -0.254, p < 0.001) emerged as the strongest independent predictor of postoperative MTP, followed by older age ( β = -0.167, p < 0.001) and higher log-transformed NT-proBNP levels ( β = -0.111, p = 0.011). In contrast, greater handgrip strength ( β = 0.221, p < 0.001) emerged as the strongest independent positive predictor of postoperative MTP, followed by higher body mass index ( β = 0.133, p = 0.001). Prolonged endotracheal intubation, advanced age, and diminished physiological reserves (low handgrip strength, elevated NT-proBNP, and lower BMI) are independent predictors of lower postoperative tongue pressure. Although extubation timing is dictated by overall clinical status, close monitoring of tongue strength and early rehabilitative support are essential for patients experiencing prolonged intubation. Furthermore, preoperative physical and nutritional optimization are key strategies to preserve postoperative tongue pressure and facilitate oral-motor recovery in vulnerable cardiac surgery patients.

Authors

Institutions

Publication Details

Journal
BMC Oral Health
Published
2026-10-06
DOI
https://doi.org/10.1186/s12903-026-09860-9
Primary Topic
Dysphagia Assessment and Management
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
article

Machine learning-based analysis of factors influencing maximum tongue pressure in patients after cardiac surgery: a cross-sectional study

Haiou Xia, Aimin Shao, Jia Xin Wan, Tingting Zhang et al.
BMC Oral Health
Dysphagia Assessment and Management
article

Machine learning-based analysis of factors influencing maximum tongue pressure in patients after cardiac surgery: a cross-sectional study

Haiou Xia, Aimin Shao, Jia Xin Wan, Tingting Zhang, Run Huang, Yilei Zhu
article en

Abstract

A low postoperative maximum tongue pressure (MTP) reflects diminished tongue muscle strength, which can compromise oral-phase swallowing function and lead to adverse clinical outcomes. However, the specific perioperative determinants of MTP in cardiac surgery patients remain poorly understood. This study aimed to identify the independent clinical predictors of postoperative MTP using a hybrid machine learning approach. In this cross-sectional study conducted at a teaching hospital in Shanghai from April 2025 to January 2026, a total of 470 adult patients undergoing cardiac surgery were included. MTP was assessed 8–24 h post-extubation, and comprehensive perioperative data were systematically extracted. To overcome multicollinearity and prevent model overfitting, a dual-algorithm feature selection was employed. Variables were ranked by a Random Forest (RF) algorithm, and optimal feature subsets were selected using Least Absolute Shrinkage and Selection Operator (LASSO) regression with 10-fold cross-validation. The overlapping core variables were subsequently incorporated into a multivariable stepwise linear regression model. A total of 470 patients were included. Based on the RF and LASSO models, 12 core features were extracted from 18 candidate variables. Multivariable linear regression confirmed five independent predictors. According to the variable importance ranking and standardized coefficients, prolonged endotracheal intubation duration ( β = -0.254, p < 0.001) emerged as the strongest independent predictor of postoperative MTP, followed by older age ( β = -0.167, p < 0.001) and higher log-transformed NT-proBNP levels ( β = -0.111, p = 0.011). In contrast, greater handgrip strength ( β = 0.221, p < 0.001) emerged as the strongest independent positive predictor of postoperative MTP, followed by higher body mass index ( β = 0.133, p = 0.001). Prolonged endotracheal intubation, advanced age, and diminished physiological reserves (low handgrip strength, elevated NT-proBNP, and lower BMI) are independent predictors of lower postoperative tongue pressure. Although extubation timing is dictated by overall clinical status, close monitoring of tongue strength and early rehabilitative support are essential for patients experiencing prolonged intubation. Furthermore, preoperative physical and nutritional optimization are key strategies to preserve postoperative tongue pressure and facilitate oral-motor recovery in vulnerable cardiac surgery patients.

BMC Oral Health
The University of Queensland (AU), Fudan University (CN), XinHua Hospital (CN)
Openalex Percentile: Top 7%
Dysphagia Assessment and Management
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.