Enhanced Sarcopenia Detection in Nursing Home Residents Using Ultrasound Radiomics and Machine Learning

Strategic Insights -

Researchers at Sichuan University have conducted a study to investigate the effectiveness of ultrasound radiomics combined with machine learning in enhancing sarcopenia diagnostic accuracy among older adults in long-term care. The study involved 628 residents from 15 nursing homes in China, where sarcopenia diagnosis was followed by the AWGS 2019 criteria. The researchers used ultrasound of thigh muscles to extract conventional parameters (muscle thickness and echo intensity) and radiomic features, which were then analyzed using five machine learning algorithms, including logistic regression.

Key Takeaways:

  • The study found that ultrasound radiomics, especially when integrated with conventional parameters and clinical data using logistic regression, significantly improves sarcopenia diagnostic accuracy in nursing home residents.
  • The diagnostic accuracy of the ultrasound radiomic models was superior to that of the models based on conventional ultrasound parameters, regardless of muscle group.
  • The integrated models achieved area under the receiver operating characteristic curve (AUC) values of 0.85 (95% CI: 0.79-0.91) for rectus femoris, 0.81 (95% CI: 0.75-0.87) for vastus intermedius, and 0.83 (95% CI: 0.77-0.90) for quadriceps femoris.
  • The conventional ultrasound models had AUC values of 0.70 (95% CI: 0.63-0.78) for rectus femoris, 0.73 (95% CI: 0.65-0.80) for vastus intermedius, and 0.75 (95% CI: 0.68-0.82) for quadriceps femoris.
  • The study concluded that the ultrasound radiomics approach holds promise for enhancing sarcopenia screening and early detection in long-term care settings.

Statistics:

  • 61.9% sarcopenia prevalence among the 628 residents
  • 628 residents from 15 nursing homes in China
  • 5 machine learning algorithms used in the study, including logistic regression
  • 70% training set, 30% validation set
  • AUC values for the integrated models: 0.85 (95% CI: 0.79-0.91) for rectus femoris, 0.81 (95% CI: 0.75-0.87) for vastus intermedius, and 0.83 (95% CI: 0.77-0.90) for quadriceps femoris
  • AUC values for the conventional ultrasound models: 0.70 (95% CI: 0.63-0.78) for rectus femoris, 0.73 (95% CI: 0.65-0.80) for vastus intermedius, and 0.75 (95% CI: 0.68-0.82) for quadriceps femoris

Sources:

  • Enhanced Sarcopenia Detection in Nursing Home Residents Using Ultrasound Radiomics and Machine Learning. Journal of the American Medical Directors Association, 2025:105830
  • Sichuan University, Center of Gerontology and Geriatrics, West China Hospital, Chengdu, People's Republic of China
  • Elsevier Science Inc, Ste 800, 230 Park Ave, New York, NY 10169, USA (publisher of Journal of the American Medical Directors Association)
  • Shuyue Luo, Hongbo Fu, Yan Zhuo, Rongna Lian, Xiaoyan Chen, Wenhua Jiang, Lei Wang, and Ming Yang (authors of the study)