Chest X-ray based AI may accurately predict malnutrition risk among bedridden patients
Malnutrition in older people affects both physical functioning and cognition. It also incurs direct and indirect costs to society. According to guidelines, rapid nutritional intervention is recommended, and nutritional therapy should be initiated within 24–48 hours of admission.
Chest radiographic prediction model can be correlated with actual height and weight and play a vital role in rapidly assessing the risk of malnutrition, according to a recent study published in Clinical Nutrition OPen Science. The lead author of this study is Nakao et al.
For proper nutritional assessment, obtaining information related to height and weight is essential. It is challenging to get the size and weight of bedridden elderly patients directly. In the present study, researchers illustrated the potential of a convolutional neural network model for assessing height and weight based on chest radiographs.
The team evaluated radiographs obtained over 15 years of follow-up, including 6,453 and 7,879 radiographs from male and female patients. A convolutional neural network predicted the height and weight of the patients (Juzen NST). A ResNet152 classifier was trained using Fastai (V1.0) running on PyTorch to indicate the height and weight.
The key results of the study are:
- The correlation coefficients between the predicted and measured values using the height prediction model were R=0.855 and R=0.81 for males and females, respectively.
- The correlation coefficients between the values predicted by the weight prediction model and measured values were R=0.793 and R=0.86, respectively.
They said, “Our chest radiographic prediction model highly correlates with actual height and weight. It can be combined with information regarding clinical nutrition factors to assess malnutrition risk rapidly.”
Training the prediction model using chest radiographs from each hospital can be optimized for the most common ethnic groups in the area, they noted. The model illustrates the potential of automated imaging AI for proper nutrition prediction models in older adults.
Reference:
Nakao, Y., Sasaki, R., Mawatari, F., Harakawa, K., Okita, M., Mitsutake, N., & Nakao, K. (2023). Development of Deep-Learning Tool to Predict Appropriate Height and Weight from Chest Radiographs in Bedridden Patients. Clinical Nutrition Open Science. https://doi.org/10.1016/j.nutos.2023.08.005
