PARIS: Personalized Activity Recommendation for Improving Sleep Quality

Meghna Singh,Saksham Goel,Abhiraj Mohan,Jaideep Srivastava

2024-05-29

Abstract:The quality of sleep has a deep impact on people's physical and mental health. People with insufficient sleep are more likely to report physical and mental distress, activity limitation, anxiety, and pain. Moreover, in the past few years, there has been an explosion of applications and devices for activity monitoring and health tracking. Signals collected from these wearable devices can be used to study and improve sleep quality. In this paper, we utilize the relationship between physical activity and sleep quality to find ways of assisting people improve their sleep using machine learning techniques. People usually have several behavior modes that their bio-functions can be divided into. Performing time series clustering on activity data, we find cluster centers that would correlate to the most evident behavior modes for a specific subject. Activity recipes are then generated for good sleep quality for each behavior mode within each cluster. These activity recipes are supplied to an activity recommendation engine for suggesting a mix of relaxed to intense activities to subjects during their daily routines. The recommendations are further personalized based on the subjects' lifestyle constraints, i.e. their age, gender, body mass index (BMI), resting heart rate, etc, with the objective of the recommendation being the improvement of that night's quality of sleep. This would in turn serve a longer-term health objective, like lowering heart rate, improving the overall quality of sleep, etc.

Machine Learning,Artificial Intelligence,Human-Computer Interaction

What problem does this paper attempt to address?

The problem this paper attempts to address is how to use machine learning techniques to improve people's sleep quality through personalized activity recommendations. Specifically, the authors focus on how to generate personalized activity suggestions based on individual biological characteristics and lifestyle constraints (such as age, gender, BMI, resting heart rate, etc.) to help users engage in appropriate activities in their daily lives, thereby improving their sleep quality that night. Additionally, these personalized recommendations can also serve long-term health goals, such as lowering heart rate and improving overall sleep quality.

### Main Issues
1. **Improving Sleep Quality**: How to use the relationship between physical activity and sleep quality to generate personalized activity suggestions through machine learning techniques to help people improve their sleep quality.
2. **Personalized Recommendations**: How to provide effective activity suggestions based on individual differences (such as health status, self-control ability, activity type, duration, and scheduling) to achieve the desired behavior change.
3. **Real-time Recommendations**: How to generate suitable activity suggestions at different times of the day based on the user's current activity status and activity deficiencies to achieve daily activity goals.

### Solutions
1. **Behavior Pattern Recognition**: Identify typical behavior patterns for each user through time series clustering of activity data.
2. **Activity Recipe Generation**: Generate activity recipes that lead to good sleep based on each behavior pattern.
3. **Personalized Recommendation Engine**: Combine the user's health metadata and lifestyle constraints to select the most appropriate activity suggestions and continuously adjust and optimize based on user feedback.

### Methods
1. **Data Collection and Processing**: Use wearable devices (such as activity trackers) to collect user activity data and convert it into minute-level activity labels (light, moderate, vigorous).
2. **Behavior Pattern Clustering**: Use the K-Means clustering algorithm to cluster activity data and identify user behavior patterns.
3. **Activity Recipe Extraction**: Perform sub-clustering on activity data within each behavior pattern to extract activity recipes that lead to good sleep.
4. **Recommendation System Design**: Design a continuous activity recommendation engine that generates personalized activity suggestions in real-time based on the user's current activity status and health metadata.

### Experiments and Results
1. **Dataset**: Use the Hispanic Community Health Study / Study of Latinos (HCHS/SOL) dataset, which includes activity data and heart rate data from 1,782 participants.
2. **Feature Space Construction**: Segment activity data into 24-hour cycles and aggregate data for each time period.
3. **Behavior Pattern Generation**: Generate user behavior patterns through clustering algorithms and verify their purity.
4. **Activity Recipe Extraction**: Extract different activity recipes based on behavior patterns, which have a strong correlation with behavior patterns.
5. **Recommendation System Evaluation**: Verify the effectiveness and accuracy of the recommendation system through retrospective evaluation.

### Discussion
1. **Initial Hypothesis**: The initial hypothesis was to group similar populations through clustering methods and generate activity recipes based on cardiovascular health and activity levels.
2. **Alternative Data Sources**: Preliminary experiments using Fitbit data validated the hypothesis, but the data volume was insufficient to build a robust model.
3. **Heart Rate Time Series Data**: Due to the limitations of heart rate data in the HCHS/SOL dataset, the clustering step based on cardiovascular health was removed.
4. **Deep Learning Application**: Use LSTM autoencoders for feature extraction to reduce the complexity of high-dimensional data and improve clustering effectiveness.

In summary, this paper aims to help users improve sleep quality through appropriate daily activities and achieve long-term health goals by combining machine learning techniques and personalized recommendations.