Scholarlink

Objectives <p>Acute suppurative cholecystitis (ASC) is a severe form of acute cholecystitis (AC) that disproportionately affects older adults, often leading to delayed diagnosis and increased morbidity. Early identification of ASC in older adults over 80 years is crucial to guide timely and appropriate management. This study aimed to develop a diagnostic model based on readily available clinical indicators to identify whether older adults with acute cholecystitis have progressed to ASC.</p> Methods <p>Older adults (aged 80 to 98 years) diagnosed with AC between January 2017 and December 2024 were retrospectively recruited from two independent centers. Patients were classified into ASC and non-ASC groups based on percutaneous cholecystostomy or laparoscopic cholecystectomy findings. LASSO logistic regression was employed to select variables and construct a diagnostic model using demographic characteristics, comorbidities, laboratory parameters, and non-enhanced CT imaging features. Model performance was evaluated by 10-fold cross-validation and decision curve analysis. Interpretability analyses were conducted using SHapley Additive exPlanations (SHAP) and Local Interpretable Model-agnostic Explanations (LIME).</p> Results <p>102 patients were included (median age 84; 49 males). The predictive model constructed using LASSO logistic regression identified five key variables: sex, necrosis of gallbladder wall, STB (serum total bilirubin), NE (neutrophil percentage), and gallbladder wall thickness. The model achieved an average AUC = 0.82 (95% CI 0.74–0.89) in cross-validation. Sex had the highest SHAP value, indicating its critical role in distinguishing between high- and low-risk patients. In predicting non-ASC, the mean SHAP value of sex was substantially higher than that of other variables, suggesting its dominant influence in identifying non-ASC. Conversely, in predicting ASC, the SHAP values of sex, necrosis of gallbladder wall, and total bilirubin were comparable, indicating that these three features hold similar importance in predicting suppuration.</p> Conclusion <p>The model incorporating five key clinical variables demonstrated potential for identifying older adults with ASC. SHAP and LIME analyses highlighted the relative importance of these features, providing interpretable insights to support early identification and personalized management of high-risk patients.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Construction and validation of a diagnostic model for acute suppurative cholecystitis in older patients: LASSO-Logistic regression analysis using readily available clinical indicators

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年1月31日

查看全文
  • Hong-Yu Long,
  • Guo-Dong Chen,
  • Bai-Qing Chen,
  • Feng Xie

摘要

Objectives

Acute suppurative cholecystitis (ASC) is a severe form of acute cholecystitis (AC) that disproportionately affects older adults, often leading to delayed diagnosis and increased morbidity. Early identification of ASC in older adults over 80 years is crucial to guide timely and appropriate management. This study aimed to develop a diagnostic model based on readily available clinical indicators to identify whether older adults with acute cholecystitis have progressed to ASC.

Methods

Older adults (aged 80 to 98 years) diagnosed with AC between January 2017 and December 2024 were retrospectively recruited from two independent centers. Patients were classified into ASC and non-ASC groups based on percutaneous cholecystostomy or laparoscopic cholecystectomy findings. LASSO logistic regression was employed to select variables and construct a diagnostic model using demographic characteristics, comorbidities, laboratory parameters, and non-enhanced CT imaging features. Model performance was evaluated by 10-fold cross-validation and decision curve analysis. Interpretability analyses were conducted using SHapley Additive exPlanations (SHAP) and Local Interpretable Model-agnostic Explanations (LIME).

Results

102 patients were included (median age 84; 49 males). The predictive model constructed using LASSO logistic regression identified five key variables: sex, necrosis of gallbladder wall, STB (serum total bilirubin), NE (neutrophil percentage), and gallbladder wall thickness. The model achieved an average AUC = 0.82 (95% CI 0.74–0.89) in cross-validation. Sex had the highest SHAP value, indicating its critical role in distinguishing between high- and low-risk patients. In predicting non-ASC, the mean SHAP value of sex was substantially higher than that of other variables, suggesting its dominant influence in identifying non-ASC. Conversely, in predicting ASC, the SHAP values of sex, necrosis of gallbladder wall, and total bilirubin were comparable, indicating that these three features hold similar importance in predicting suppuration.

Conclusion

The model incorporating five key clinical variables demonstrated potential for identifying older adults with ASC. SHAP and LIME analyses highlighted the relative importance of these features, providing interpretable insights to support early identification and personalized management of high-risk patients.