This study examines historical community health and environmental conditions data from medieval Andalusia during the period 318–879 AH (930–1474 CE). Through a mixed-method approach of quantitative and qualitative analyses, the research utilized Natural Language Processing (NLP), machine learning algorithms, and predictive modeling on extensive historical records, archaeological finds and geographical data from the Rome region. The study disclosed strong spatial and temporal variability of public health outcomes in the regions of Andalusia. Coastal regions were 30% less likely to suffer from waterborne diseases than inland areas, and rural hinterlands had the worst health indicators, especially where access to water was limited. In the time-series and phase transition analysis, we described three separate phases of public health progress: Phase I—an initial period of steady improvement (318–500 AH), Phase II—a period of equilibrium with sporadic outbreaks of diseases (501–700 AH) along with gradual deceleration in progress, and Phase III—progressing toward decline (701–879AH). Statistical analyses indicated high correlations between environmental determinants and health outcomes; for instance, the correlation of water quality was r = − 0.82 (p = 0.001), with urban green spaces yielded r = − 0.65 (p = 0.002) and waste management exhibited a negative correlation at r = − 0.78 (p = 0.001) with prevalence of disease A multiple regression showed that environmental factors explained 73% (R2 = 0.73, p < 0.001) of the variance in health outcomes. The results presented here show the potential of these AI tools to analyze historical data and provide a timely reminder about how environmental conditions affect public health in medieval Islamic societies, which has implications for twenty-first-century urban health planning.