A groundbreaking study has unveiled a new way of predicting longevity—through the use of abdominal CT scans analyzed by explainable artificial intelligence (AI). This innovative biological age model, developed by researchers at the University of Wisconsin Hospitals and Clinics, diverges from the conventional reliance on chronological age
Researchers derived the CT-based biological age model utilizing abdominal imaging data from over 123,000 adults, correlatively measuring factors such as skeletal muscle, visceral fat, and bone density. Unlike previous methodologies, which focused primarily on cellular indicators of aging, this study leverages macroscopic data, promising more personalized assessments of health and longevity.
The results demonstrated the CT model significantly outperforms traditional demographic data. Specifically, it achieved an index of prediction accuracy (IPA) of 29.2, compared to just 21.7 for models relying on age, sex, and race.
This CT-based biological age model does not solely reflect patient age; it indicates the cumulative effects of lifestyle habits, genetic predisposition, and various health conditions. The authors note, “Biological age is a potentially useful construct... to reflect the cumulative physiologic effect of lifestyle habits, genetic predisposition, and superimposed disease processes.”
Throughout its analysis, the study revealed key insights about the biomarkers most predictive of longevity. Muscle density emerged as the most significant metric, highlighting the connection between skeletal health and survival. Aortic calcium score and visceral fat density were also identified as important contributors. These findings align with existing literature emphasizing the importance of maintaining healthy muscle mass and cardiovascular health as one ages.
During the study's follow-up, which averaged over five years, more than 22% of the participants had died. By examining the data, researchers noted drastic differences between the biomarkers of those who survived and those who did not. For example, subjects in the highest risk quartile had survival hazard ratios indicating significantly elevated risk, validating the model's predictive capabilities.
Interestingly, when researchers excluded individuals diagnosed with serious conditions like cancer within five years of their CT scans, the predictive robustness of the model increased considerably, with hazard ratios rising to 24.79, indicating powerful ramifications for early diagnostics and preventive healthcare measures.
“Our CT-based approach could also be used to augment existing clinical risk prediction models,” the authors remark, underscoring the opportunity for enhanced healthcare strategies. The team envisions leveraging existing abdominal CT scans—which are commonly performed for various medical assessments—for broader screening applications, particularly with the rising incidence of metabolic syndrome and associated conditions.
The integration of AI algorithms amortizes the intricacies of analyzing such imaging studies, paving the way for rapid and reproducible assessments without the tedious manual efforts previously required.
Overall, the study positions the CT-based biological age model as not only capable of estimating longevity but as providing actionable insights for health interventions. The advancement of CT technology and the employment of AI heralds a paradigm shift toward more personalized medicine, prompting clinicians to use comprehensive body composition data for patient assessments.
Future studies, the researchers claim, will seek to expand the demographic diversity of the cohorts involved, ensuring the model's applicability across varied groups. They also plan to explore additional socioeconomic factors to enrich assessment capabilities.
With the compelling evidence provided by this study, there is optimism about the potential for CT-derived metrics to redefine how health professionals evaluate aging and longevity, making strides toward enhancing public health approaches and individual health outcomes.