HPLC Training for All: Integration of Multi-Language Artificial Intelligence (AI) Avatars in Virtual Reality (VR) Digital Twin Laboratories

Stephen Hilton,Ayah Al-Dargazelli,Amy Locks,Mae Taylor,Mireia Montaner

DOI: https://doi.org/10.26434/chemrxiv-2024-q511z

2024-04-08

Abstract:The integration of High-Performance Liquid Chromatography (HPLC) training into academic curricula remains a challenge, due to the complexities of its theoretical concepts and the cost-prohibitive nature of HPLC instrumentation. This challenge in training is further compounded for students who are taught in the English language and for whom English is a Foreign Language (EFL), and who may struggle to understand traditional learning materials. This paper introduces a pioneering Virtual Reality (VR) platform for immersive HPLC training that is coupled with a conversationally intelligent Artificial Intelligence (AI) avatar capable of multilingual support. This approach provides students with an on-hand tutor for adaptive, immersive learning practices in whatever language they feel comfortable speaking in. The effectiveness of the AI models were evaluated via a survey of postgraduate Chemistry students at UCL, and the responses were analysed computationally using a series of evaluation metrics. Evaluation from TF-IDF, BLEU and BERT statistical analysis demonstrated that the AI avatar had superior linguistic and instructional performance across English, Spanish, and German, particularly in complex concept explanations, validating the potential for multilingual AI support in education. Survey feedback revealed that the AI Avatars with a knowledge bank significantly improved HPLC learning outcomes, with participants rating their engagement and the quality of learning above traditional methods.

Chemistry

What problem does this paper attempt to address?

This paper introduces a method that utilizes virtual reality (VR) and artificial intelligence (AI) to solve the challenges of high-performance liquid chromatography (HPLC) training. HPLC is an important technique in fields such as pharmaceuticals, but due to its complex theory and expensive equipment, training faces challenges in academic courses, especially for non-native English-speaking students. The paper proposes an innovative VR platform that combines a multilingual conversational AI avatar to provide students with adaptive and immersive learning experiences, regardless of the language they use.

The research evaluates the effectiveness of the AI model through a survey conducted among chemistry graduate students at UCL and demonstrates the superior language and educational performance of the AI avatar in English, Spanish, and German using statistical analysis methods such as TF-IDF, BLEU, and BERT, particularly in explaining complex concepts. Feedback shows that the AI avatar with a knowledge base significantly improves HPLC learning outcomes, and students rate its engagement and learning quality higher than traditional methods.

The paper also discusses how VR eliminates language barriers and provides non-native English-speaking students with fair and high-quality scientific education opportunities by simulating practical experience. The research team has developed software called LAB427 Digital Twin HPLC, which includes multilingual AI avatars to seamlessly integrate theory and practice. The results indicate that AI with a knowledge base performs better in various languages compared to AI without a knowledge base, and it has received positive feedback from students, highlighting the potential of AI in personalized education and improving learning outcomes. Future research will expand to broader environments and long-term impacts.