An Analytical Study on the Prevalence of Type-2 Diabetes among American Adults

Abstract

Purpose: This project aims to conduct rigorous analytics using AI/ML tools to examine a) the correlation among risk factors and diabetes, and b) the significance of risk factors in diabetes prediction among American adults. Goals: The overarching goal of this project is to build an ML model that can predict the occurrence of diabetes in an American adult under given conditions (variables, risk factors, etc.) with a very high level of accuracy. The ML model will be deployed to build an interactive dashboard that will display various relevant information, including the prediction of diabetes, confidence intervals, the significance of risk factors, and cohort-related (race, age, socio-economic status etc.) statistics given any user input. Methodology: We will implement a combination of Quantitative and Descriptive research approaches. We will utilize the diabetes dataset from the OCHIN database to a) perform a comprehensive analysis of the dataset, b) build base and ensemble predictive models, compare them, and choose the best-performing model, and c) create an interactive dashboard that will display the relevant outcomes (such as summary statistics, cohort-based summaries, prevention and control measures, diabetes prediction) for any specific user input. This work will provide an opportunity to a) identify and describe the correlation between risk factors and diabetes, b) examine the significance of physiological and lifestyle factors in predicting diabetes, and c) inform the targeted population about diabetes risk, control, and intervention strategies. Deliverables: This project aims to achieve its goals through the following deliverables: 1) Choose an appropriate dataset for the study; 2) Perform a comprehensive EDA using charts and graphs; 3) Perform cohort-based statistical analysis; 4) Build various baseline and ensemble predictive models; 5) Compare their performance and choose the best-performing model; 6) Create an interactive dashboard incorporating the results of this work Results: We recently got access to the OCHIN database and are in the process of data acquisition, which is the most critical component of this project. So, this is a work in progress. We plan to carry out and complete the data analysis by July 2025. Predictive modeling and dashboard development will be completed by December 2025. Conclusions: This project aims to build more robust and equitable models for diabetes prediction. This is a work in progress and we recently got access to the OCHIN database and are in the process of data acquisition. We plan to undertake the data analysis by July 2025. Predictive modeling and dashboard development will be completed by December 2025. The developed model is expected to advance the state of the art in equitable diabetes rate estimation by embedding fairness and demographic sensitivity into predictive modeling, with the potential to transform how healthcare disparities are addressed through more inclusive, transparent, and effective public health strategies. Implications of this research: The implications of this research are far-reaching in both the medical and technological domains. By developing a highly accurate machine learning model tailored to predict the occurrence of diabetes in American adults based on specific risk factors and conditions, this project addresses a critical health disparity with data-driven precision.

Ähnliche Arbeiten