Abstract:A personally tailored exercise regimen is crucial to ensuring sufficient physical activities, yet challenging to create as people have complex schedules and considerations and the creation of plans often requires iterations with experts. We present PlanFitting, a conversational AI that assists in personalized exercise planning. Leveraging generative capabilities of large language models, PlanFitting enables users to describe various constraints and queries in natural language, thereby facilitating the creation and refinement of their weekly exercise plan to suit their specific circumstances while staying grounded in foundational principles. Through a user study where participants (N=18) generated a personalized exercise plan using PlanFitting and expert planners (N=3) evaluated these plans, we identified the potential of PlanFitting in generating personalized, actionable, and evidence-based exercise plans. We discuss future design opportunities for AI assistants in creating plans that better comply with exercise principles and accommodate personal constraints.

What problem does this paper attempt to address?

The paper attempts to address the problem of how to use large language models (LLMs) to create personalized exercise plans that cater to individuals' complex schedules and personalized needs. Specifically, the paper proposes a conversational AI system called PlanFitting, which aims to help users develop personalized weekly exercise plans based on their goals, available time, and potential obstacles, and make adjustments as necessary. ### Main Issues 1. **Challenges of Personalized Exercise Plans**: - Individuals' schedules are complex and variable, making it difficult to create exercise plans that suit everyone. - Traditional exercise planning methods often require the involvement of professionals, which is costly, less accessible, and lacks customization. - Even the most meticulously designed exercise plans need to be adjusted based on changes in personal circumstances, which is difficult to achieve without ongoing professional support. 2. **Shortcomings of Technical Solutions**: - Previous technical support methods (such as self-reflection, peer support, or crowdsourcing) can generate plans comparable to those of experts but still require significant human and time investment. - These methods rely on detailed information about clients' preferences, goals, and schedules, but this information may be incomplete without guidance. - The need for long-term iteration and adaptation to personal changes makes these methods difficult to sustain and scale. ### Solution - **PlanFitting System**: - Utilizes the generative capabilities of large language models (LLMs) to collect users' personalized constraints (such as exercise goals, available time, and potential obstacles) through natural language conversations. - Recommends suitable types of exercises based on the collected information and generates personalized weekly exercise plans. - Adopts an implementation intention (IF-THEN rule) format to associate exercise plans with specific events in users' lives rather than strictly following a timetable. - Through conversational interaction, helps users express and adjust their personalized needs, enhancing the feasibility and operability of the plans. ### Research Methods - **Preliminary Research**: Conducted interviews with professional exercise planners (5 people) and ordinary users with experience in personalized exercise plans (8 people) to understand current exercise planning practices and challenges. - **System Design and Development**: Based on insights from preliminary research, designed and developed the PlanFitting system, including the design of the conversational system, user interface components, and implementation details. - **User Research**: Conducted an exploratory user study (18 participants) to evaluate how users utilize PlanFitting to develop and adjust exercise plans, and assessed experts' evaluations of the quality of the generated plans. ### Contributions 1. Revealed the process and challenges of personalized exercise planning between clients and professional exercise planners. 2. Designed and developed PlanFitting, a novel exercise planning system driven by large language models, to help generate and iterate personalized exercise plans. 3. Demonstrated through exploratory user research how people use PlanFitting to develop exercise plans and provided professional exercise planners' evaluations of the quality of the generated plans.

PlanFitting: Tailoring Personalized Exercise Plans with Large Language Models

Adaptive and Personalized Exercise Generation for Online Language Learning

Enhancing Digital Health Services: A Machine Learning Approach to Personalized Exercise Goal Setting

Designing an AI Health Coach and Studying Its Utility in Promoting Regular Aerobic Exercise

AI Coach: Deep Human Pose Estimation and Analysis for Personalized Athletic Training Assistance

LLM-Personalize: Aligning LLM Planners with Human Preferences via Reinforced Self-Training for Housekeeping Robots

The Last JITAI? The Unreasonable Effectiveness of Large Language Models in Issuing Just-in-Time Adaptive Interventions: Fostering Physical Activity in a Prospective Cardiac Rehabilitation Setting

A Brief Review of the Efficacy in Artificial Intelligence and Chatbot-Generated Personalized Fitness Regimens

LessonPlanner: Assisting Novice Teachers to Prepare Pedagogy-Driven Lesson Plans with Large Language Models

Qualitative evaluation of artificial intelligence-generated weight management diet plans

EgoPlan-Bench2: A Benchmark for Multimodal Large Language Model Planning in Real-World Scenarios

Supporting Physical Activity Behavior Change with LLM-Based Conversational Agents

Fitness Done Right: a Real-time Intelligent Personal Trainer for Exercise Correction

Using artificial intelligence for exercise prescription in personalised health promotion: A critical evaluation of OpenAI's GPT-4 model

Fine-grained Fitting Experience Prediction

PhysioLLM: Supporting Personalized Health Insights with Wearables and Large Language Models

Educational Personalized Learning Path Planning with Large Language Models

Democratizing Large Language Models via Personalized Parameter-Efficient Fine-tuning

Delighting Palates with AI: Reinforcement Learning's Triumph in Crafting Personalized Meal Plans with High User Acceptance

PEFT-U: Parameter-Efficient Fine-Tuning for User Personalization