Comprehensive Summary
The study analyzes 60 years of global health information from the years 1990-2050, to understand and make predictions about type 2 diabetes mellitus. Researchers specifically looked at understanding how type 2 diabetes presented among working populations (ages 15-64) using an advanced machine learning model known as XGBoost that uses SHAP analysis to forecast future trends in diabetes as well as identity factors driving disease burden. Projections from the machine learning model were alarming: global incidence cases are expected to increase 5x from 6.33 million in 1990 to 32.38 million by 2025, with the areas having the fastest growth being in North Africa and the Middle East. They also saw traditionally known pattens such as wealthier countries having lower diabetes rates actually being reversed with high-income countries such as the United Kingdom, facing accelerating rates. The SHAP analysis also identified age as the predominant contributor to type 2 diabetes mellitus burden, while high blood glucose levels, elevated BMI< and air pollution remained consistently influential factors. Emerging risk factors also include high temperatures (climate change related) and alcohol consumption which showed increasing significance over time.
Outcomes and Implications
This research is important because it highlights the need for finding specific targeted strategies for diabetes prevention for working-age populations. This demographic of people is typically overlooked as well despite playing a large role in being the economic backbone of society. For clinicians, this study emphasizes the importance of early screening and targeted intervention in working-age patients, specifically in monitoring blood glucose levels control, weight management, and environmental exposures. These findings that high-income countries are now experiencing growing rates of diabetes as well shows a warning that this disease has the potential to effect many different communities and that these interventions remain important, even in places with many resources. The utilization of machine learning allows clinicians to also see how powerful AI can be in identifying high-tide populations and regional disparities. While this model was only applied to diabetes, this has the implications to be applied to many different diseases for study purposes. Moreover, this study gives us great insights into high diabetes is progressing as a disease in order to find better prevention methods as well as highlighting how AI can be utilized in this prevention process.