Model Pipeline
Wearable PPG. 10 hour full-night recording
Home-Based Sleep Staging with the Spica Platform
AI-Powered Sleep Analysis. Anywhere.
Spica enables remote, home-based sleep staging using wearable devices and artificial intelligence. Participants wear a compatible device overnight while the Spica app automatically collects physiological signals. Our deep learning engine analyzes the data and provides accurate sleep stage predictions in an intuitive research dashboard.
Clinically Validated
Secure & GDPR Compliant
Scalable for Large Studies
Understanding Sleep Architecture
Sleep cycles through distinct stages that play vital roles in physical recovery, memory, emotional regulation and overall well-being.
1
Wake
Awake Alert
2
Light Sleep (N1/N2)
Light sleep Restorative sleep
3
Deep Sleep (N3)
Body repair Recovery
4
REM Sleep (Rapid Eye Movement)
Memory consolidation Learning
Complete Sleep Staging Workflow with Spica
Wearable Devices
- ● Polar 360 wearable
- ● Smart rings with PPG
- ● Compatible PPG wearables
- ● Multi-device integration
Spica App
- ● Overnight data recording
- ● Automatic synchronization
- ● Secure local storage
- ● Study task reminders
AI Sleep Staging Engine
- ● Deep learning analysis
- ● 30-second sleep staging
- ● High accuracy
- ● Cross-subject validation
Spica Dashboard
- ● Remote participant monitoring
- ● Sleep data visualization
- ● Sleep results review
- ● Structured data export
Wearable Devices
- ● Polar 360 wearable
- ● Smart rings with PPG
- ● Compatible PPG wearables
- ● Multi-device integration
Spica App
- ● Overnight data recording
- ● Automatic synchronization
- ● Secure local storage
- ● Study task reminders
AI Sleep Staging Engine
- ● Deep learning analysis
- ● 30-second sleep staging
- ● High accuracy
- ● Cross-subject validation
Spica Dashboard
- ● Remote participant monitoring
- ● Sleep data visualization
- ● Sleep results review
- ● Structured data export
Supported Physiological Signals
Spica collects multiple physiological and behavioral signals including:
Sleep
Sleep Metrics
Cardiovascular
Heart Rate (HR) HRV PPG Metrics
Behavioral
Motion Activity Step Count
Thermoregulatory
Skin Temperature
Results
The model performs well under cross-subject evaluation on two distinct datasets, indicating effective learning of subject-independent sleep patterns. The lower performance in the 4-class setting is mainly attributed to Deep Sleep class imbalance, as balancing methods were avoided to preserve temporal continuity. Cross-dataset results further demonstrate strong generalization across datasets, highlighting the robustness and transferability of the proposed approach.
Cross Subject Evaluation
MESA Dataset (3 Class)
MESA Dataset (4 Class)
Cross Dataset Evaluation
ABC Dataset (Method 3)
Why It Matters
Accurate sleep stage detection enables early detection of sleep disorders, personalized recommendations, and better long-term health outcomes.
Early Detection
Identify hidden sleep disturbances early
Personalized Insights
Tailored recommendations for better sleep
Better Outcomes
Improve recovery, health and quality of life
- Memory & Learning: Strengthens memory consolidation
- Physical Recovery: Supports tissue repair and immune function
- Emotional Well-Being: Regulates mood and reduces stress
- Performance: Enhances focus, energy and productivity
From Sleep Signals to Meaningful Insights
Clinical Applications
- ● Sleep disorder screening support
- ● Insomnia pattern detection
- ● Circadian rhythm analysis
- ● Longitudinal patient monitoring
Research Applications
- ● Digital phenotyping of sleep
- ● Large-scale cohort studies
- ● Biomarker discovery in sleep & psychiatry
- ● Remote patient monitoring studies
Personalized Insights
- ● Sleep quality beyond duration
- ● Recovery and fatigue estimation
- ● Behavioral and Lifestyle feedback
- ● Better daily performance and well-being
Let’s Transform Sleep Into Better Outcomes
Partner with us to bring advanced sleep intelligence to more people, everywhere.
