Abstract WP437: Where Are We Located Now With Artificial Neural Network? Application of Deep Learning for the Development of Culturally Sensitive Social Support Interventions for Dementia Caregivers

Introduction: Little is known about culturally sensitive interventions for Hispanic and African American family caregivers for persons living with vascular dementia. Applications of deep-learning neural networks have expanded rapidly in recent years, from the first uses in character recognition (1989) to the high-profile AlphaGo (2016) from Google DeepMind. The purpose of this study is to visualize topics from health literature that use deep learning, exploring the techniques’ ability to provide a foundation for developing culturally sensitive Twitter-based social support interventions for Hispanic and African American dementia caregivers. Methods: The corpus of deep learning literature was extracted from 506 studies mentioning deep learning from PubMed library. Text mining was conducted using AutoMap to identify topics and semantic relations from the corpus. Results were visualized as network clusters of frequently applied topics and isolated topics. Results: Literature applying deep learning are limited to 1) imaging and radiology, 2) genomics, 3) cancer (N=506 studies). Topics on dementia and drug discovery were identified as an emerging area for the adaptation of deep learning techniques (Figure 1). Only seven journals in the PubMed library were identified as having published more than 10 studies applying deep learning methods. Conclusion: Despite its popularity in art and science research, deep learning is at an early stage of adoption in health science, largely focused on limited topics, e.g., imaging and disease diagnosis. Applying deep learning methods may provide insights for developing culturally sensitive Twitter-based interventions for Hispanic and African American dementia caregivers. Implementing education on applying deep learning to imaging and vascular-dementia prediction may be a good start to adapt, deepen and broaden the scope of health analytics to improve vascular dementia patients’ health outcomes.