Understanding medical terminology is critical for effective patient-doctor communication, yet many patients struggle with complex jargon. This study compares Machine Learning (ML) models and Large Language Models (LLMs) in predicting medical term complexity as a means of improving doctor-patient communication. Using survey data from 252 participants rating 1,000 words along with various lexical features, we measured the accuracy of both model types. The results show that LLMs outperform traditional lexical-feature-based models, suggesting their potential to identify complex medical terms and lay the groundwork for personalised patient-doctor communication.