Developing a predictive model for anticipating technology convergence: A transformer-based model and supervised learning approach

Abstract

This study proposes a novel approach to anticipating technology convergence in the bio-healthcare sector by integrating text mining based on transformer models and supervised learning methodologies. The overarching goal is to develop a robust method for predicting technology convergence, leveraging the interrelationships between technology topics extracted from patents and research articles. Through the application of advanced techniques and by leveraging the strengths of transformer-based models such as BERTopic with KeyBERT and OpenAI integration to generate technology topics, we identified potential convergence opportunities and explored emerging trends within the dataset. The proposed method seeks to predict technology convergence effectively by employing various machine learning and deep learning techniques to train prediction models by integrating technological similarity, link prediction measures, and causal relationships between technology topics as input features, offering a more accurate and comprehensive understanding of the intricate relationships within the technological landscape. This study contributes to the literature on technology convergence by offering a novel methodology for anticipating future trends and identifying opportunities for interdisciplinary collaboration in the bio-healthcare sector. Overall, the outcomes of this study hold significant implications for businesses seeking to capitalize on emerging convergence opportunities for sustainable growth.