AI algorithms utilizing body measurements predict lung cancer, CVD, and all-cause mortality. Researchers analyzed CT chest scans, showing AI-obtained body composition measurements improved lung cancer death and all-cause mortality prediction. Another study used AI models on abdominal CT images to predict type 2 diabetes risk based on pancreatic health and fat levels, yielding consistent and accurate results. These findings indicate the potential of AI in healthcare for outcome prediction and disease risk assessment.
Recent research has demonstrated the remarkable capability of an AI algorithm to utilize body composition measurements for the prediction of lung cancer, cardiovascular disease (CVD), and all-cause mortality.
The study focused on the application of artificial intelligence in assessing non-contrast low-dose CT chest scans for cancer screenings and predicting various outcomes, including lung cancer, CVD, and overall mortality.
While AI is making strides in the healthcare domain, its specific applications in certain areas remain insufficiently explored. This new research introduces an AI algorithm designed to perform body composition assessments on non-contrast low-dose CT chest scans for lung cancer screening.
The researchers aimed to investigate the potential of using these body composition measurements in predicting disease risks, particularly lung cancer incidence, lung cancer-related deaths, CVD-related deaths, and all-cause mortality, within the context of the National Lung Screening Trial (NLST).
The research involved a secondary analysis of NLST data, from which the researchers obtained body composition measurements such as area and attenuation attributes of skeletal muscle and subcutaneous adipose tissue based on baseline LDCT examinations. Using additional proportional hazards models, the researchers assessed the additional value of these measurements, considering patient characteristics like age and medical history.
The study encompassed a total of 20,768 participants with a mean age of 61, of whom 12,317 were men. During the follow-up period, 865 individuals received a lung cancer diagnosis, and 4,180 individuals passed away. The findings revealed that AI-obtained body composition measurements significantly contributed to risk prediction for lung cancer-related deaths and all-cause mortality. However, the prediction of lung cancer incidence did not benefit from this approach.
The use of AI to predict and evaluate risks related to negative health outcomes has gained significant traction in the healthcare industry.
In another study conducted in April 2022, researchers from the National Institutes of Health Clinical Center developed an AI model that employed pancreas health and fat level factors to assess the risk of type 2 diabetes using non-contrast abdominal CT images.
The algorithm was constructed using hundreds of images from various datasets, analyzing CT attenuation, pancreatic volume, intrapancreatic fat percentage, and three-dimensional fractal dimension measurements. The results demonstrated that low pancreas density and high visceral fat were correlated with diabetes prevalence. The researchers concluded that fully automated analyses of abdominal CT biomarkers are reliable in predicting diabetes due to the consistent and accurate results produced by the model.
According to study authors Ronald M. Summers, MD, Ph.D., and Hima Tallam, an MD/Ph.D. student, this research represents a significant advancement toward the broader application of automated methods in addressing clinical challenges. Moreover, it may provide valuable insights into the underlying reasons for pancreatic changes in patients with diabetes.
These research endeavors highlight the potential of AI in predicting various health outcomes and offer promising opportunities for advancing medical diagnostics and improving patient care.
