Docs are not available for mobile use. Please use a desktop computer to view the documentation.

Signal Boundaries

Detected boundaries between distinct periods within individual signals

Purpose

The signal_boundaries table stores detected boundaries within individual signals that mark transitions between distinct periods or states. These boundaries help identify when a signal changes significantly, indicating shifts in behavior, context, or conditions.

Schema Definition

Field	Type	Description	Constraints
`id`	uuid	Unique boundary identifier	Primary key, auto-generated
`userId`	uuid	Boundary owner	Foreign key to users.id
`sourceName`	varchar(255)	Source of the signal	Not null
`signalName`	varchar(255)	Name of the signal	Not null
`startTime`	timestamptz	Boundary period start	Not null
`endTime`	timestamptz	Boundary period end	Not null
`confidence`	real	Detection confidence (0-1)	Not null
`detectionMethod`	varchar(100)	Method used for detection	Not null
`boundaryMetadata`	jsonb	Additional boundary information	Optional
`createdAt`	timestamptz	Record creation timestamp	Auto-generated

Relationships

References:

users.id via userId - Boundary owner

Indexes

idx_signal_boundary_user_time - B-tree index on (userId, startTime, endTime) for temporal queries
idx_signal_boundary_source - B-tree index on (sourceName, signalName) for signal filtering

Detection Methods

Common detection methods include:

episodic_rule - Rule-based episodic event detection
ambient_change - Statistical change point detection in ambient data
merged - Combination of multiple detection methods
speed_threshold - Speed or rate-of-change based detection
gap_fill - Filling gaps in data continuity

Usage Examples

Get boundaries for a specific signal

SELECT startTime, endTime, confidence, detectionMethod
FROM signal_boundaries 
WHERE userId = 'user-uuid'
  AND sourceName = 'fitbit'
  AND signalName = 'heart_rate'
ORDER BY startTime;

Find recent high-confidence boundaries

SELECT sourceName, signalName, startTime, endTime, confidence
FROM signal_boundaries 
WHERE userId = 'user-uuid'
  AND confidence >= 0.8
  AND startTime >= NOW() - INTERVAL '24 hours'
ORDER BY confidence DESC, startTime DESC;

Analyze boundary detection patterns

SELECT 
  sourceName,
  signalName,
  detectionMethod,
  COUNT(*) as boundary_count,
  AVG(confidence) as avg_confidence
FROM signal_boundaries 
WHERE userId = 'user-uuid'
  AND startTime >= NOW() - INTERVAL '30 days'
GROUP BY sourceName, signalName, detectionMethod
ORDER BY boundary_count DESC;

Get boundaries by duration

SELECT 
  sourceName,
  signalName,
  startTime,
  endTime,
  EXTRACT(EPOCH FROM (endTime - startTime))/60 as duration_minutes,
  confidence
FROM signal_boundaries 
WHERE userId = 'user-uuid'
  AND (endTime - startTime) > INTERVAL '1 hour'
ORDER BY duration_minutes DESC;

Find overlapping boundaries

SELECT 
  a.sourceName as source_a,
  a.signalName as signal_a,
  b.sourceName as source_b,
  b.signalName as signal_b,
  GREATEST(a.startTime, b.startTime) as overlap_start,
  LEAST(a.endTime, b.endTime) as overlap_end
FROM signal_boundaries a
JOIN signal_boundaries b ON a.userId = b.userId
WHERE a.id != b.id
  AND a.userId = 'user-uuid'
  AND a.startTime < b.endTime
  AND a.endTime > b.startTime
ORDER BY overlap_start;

Get boundary metadata details

SELECT 
  sourceName,
  signalName,
  startTime,
  detectionMethod,
  boundaryMetadata
FROM signal_boundaries 
WHERE userId = 'user-uuid'
  AND boundaryMetadata IS NOT NULL
ORDER BY startTime DESC;