This study aims at exploring the indigenous elders’ health literacy of chronic diseases, designing a fall prevention textbook needed for the cultural consistency of the ethnic group, and analyzing such elders’ performance of physical strength and activity stability through the teaching materials. The study consisted of two phases; first, it mainly explored the elders’ health literacy of chronic health. A structured questionnaire survey was conducted to collect 60 samples of the Bunun elders. To analyze the data, a statistical software, version 22.0, was employed. The results indicated: (1) The older the person the higher the level of cultural acculturation; (2) the higher the degree of traditional culture, the lower the health knowledge scores; (3) the higher the health knowledge score, the better the self-care behavior of chronic diseases; (4) the health knowledge and self-care behavior of chronic disease predictors include age, education level, and occupation. The second phase mainly uses the factors of the traditional culture to influence health and understand the common health issues of the indigenous elders on chronic disease management and fall prevention. We interviewed 12 elders to summarize the important indigenous traditions. Then, based on the core health concept, we designed the cultural care for 12 weeks, in line with the course for the fall prevention and treatment for such elders. The tribal intervention is carried out with a quasi-experimental design. Via purposive sampling, some Atayal and Tsou participants, aged 55 years and above, 60 people, 30 in each ethnic group, were selected. The data, collected from pre- and post-intervention, were analyzed with the SPSS 22.0. Results showed that after the 12-week intervention course, the intervention group had better health cognitive scores, and the Tinett gait time was shortened. Meanwhile, we selected statistically significant variables include mood adaptation, ADL, IADL, health cognition, POMA-G and POMA-B, plus an ANN model established to incorporate these variables into the input variables to test whether the course can predict the falls and physical activities of the two ethnic groups. Ultimately, we observed that DNN model could accurately predict that the indigenous elders could have better mood adaption, IADL, health cognitive scores, and balance and gait performance, with accuracy of 80% in Tsou elders and 100% in Atayal elders. It is suggested that, in the future, in caring for the elders with chronic diseases in Taiwan, health education should avoid excessive professional terminology or esoteric language. This can be presented in pictures or multimedia languages to enhance the indigenous elders’ health and life quality.