This study presents a model for estimating healthy life expectancy in Gyeongsangbuk-do at the city, county, and district level using machine learning. Quality-adjusted life expectancy (QALE) was calculated at each level using Graville correction and life tables. Next, 43 factors related to healthy life expectancy, including demographic and health care policy variables, were obtained from national health data. Machine learning was used to estimate healthy life expectancy. It was confirmed that LightGBM and artificial neural network had superior estimation performance compared to the multiple linear regression model commonly used in healthcare and medical science. Using the artificial neural network model with the best performance, we conducted additional factor analysis using Shapley additive explanations. Our findings confirmed that the depression experience rate and perceived stress rate were the most significant factors affecting healthy life expectancy in all cities, counties, and districts in Gyeongsangbuk-do. However, the sensitivity analysis revealed that the ranking of factors causing an increase or decrease in healthy life expectancy varied across cities, counties, and districts. Thus, it was confirmed that tailored policies, accounting for regional circumstances, are necessary to promote health and enhance equity.