Vagif Aliyev
Bridging sensor data and human comprehension: Designing a Large Language Model-Based coaching interface for runners
At: WearM.ai
An AI Coach Runners Can Trust
Wearable sensors give runners a large amount of data, but turning that into actionable improvements is the hard part. This project filled that gap by designing and testing an AI-powered coaching interface. Using a Large Language Model (LLM), the system translates complex bio-mechanical sensor data into clear, personal running advice delivered through a conversational coach.
Results were very positive, and the prototype scored an excellent usability score (SUS 86.8). Users found it trustworthy and actionable due to three key factors: transparency that showed the thought process, reasoning, and rationale behind decisions, the ability of the coach to provide responses consisting of interactive visuals with conversational explanations, and allowing users to control their coaching style and preferences. This work provides a validated framework for building effective, human-centred AI coaches.
Exam date: 14-07-2025