Skip to main content

​
Overview

This endpoint analyzes the user’s diary data from the last 30 days to identify patterns, triggers, safe foods, and time-based correlations. Discoveries are sorted by confidence level and type.

​
Endpoint

GET /insights/discoveries

​
Authentication

This endpoint requires authentication via Bearer token. 
Authorization: Bearer <access_token>

​
Response

​
success
boolean
required
Indicates if the request was successful
​
data
array
required
Array of discovery objects, sorted by confidence (high first) and type

​
Discovery Types

​
Trigger Foods

Foods marked as problematic 2+ times. Includes:
  • Food name
  • Associated symptom types
  • Total number of incidents
  • Whether it’s a recent discovery (last 7 days)
Confidence Levels:
  • High: 5+ incidents
  • Medium: 3-4 incidents
  • Low: 2 incidents

​
Time Patterns

Identifies time periods when symptoms occur most frequently (requires 5+ symptom entries). Time Periods:
  • Morning: 06:00 - 12:00
  • Afternoon: 12:00 - 18:00
  • Evening: 18:00 - 22:00
  • Night: 22:00 - 06:00
Detection Criteria:
  • Pattern shown only if >40% of symptoms occur in one period
  • Minimum 3 occurrences required
Confidence Levels:
  • High: >60% of symptoms in one period
  • Medium: 40-60% of symptoms in one period

​
Safe Foods

Foods consumed 5+ times without being marked as problematic. Shows top 3 most consumed safe foods. Confidence Levels:
  • High: Consumed 10+ times without issues
  • Medium: Consumed 5-9 times without issues

​
Combination Patterns

Problematic food combinations (future feature - currently not implemented in the service).

​
Example Request

curl -X GET https://api.ceboelha.com/insights/discoveries \
  -H "Authorization: Bearer eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9..."

​
Example Response

{
  "success": true,
  "data": [
    {
      "id": "507f1f77bcf86cd799439011",
      "type": "trigger",
      "confidence": "high",
      "title": "você marcou leite",
      "description": "parece que leite não combina muito... você anotou algumas vezes.",
      "data": {
        "foods": ["Leite"],
        "symptoms": ["bloating", "cramps", "diarrhea"],
        "occurrences": 6
      },
      "discoveredAt": "2026-03-01",
      "isNew": true
    },
    {
      "id": "507f1f77bcf86cd799439012",
      "type": "time_pattern",
      "confidence": "high",
      "title": "mais anotações à noite",
      "description": "a maioria das suas anotações são entre 18h e 22h (noite).",
      "data": {
        "timeRange": "18:00 - 22:00",
        "occurrences": 15
      },
      "discoveredAt": "2026-03-03",
      "isNew": false
    },
    {
      "id": "507f1f77bcf86cd799439013",
      "type": "safe_food",
      "confidence": "high",
      "title": "arroz tá de boa!",
      "description": "você comeu arroz várias vezes e nunca marcou nada negativo.",
      "data": {
        "foods": ["Arroz"],
        "occurrences": 12
      },
      "discoveredAt": "2026-03-03",
      "isNew": false
    },
    {
      "id": "507f1f77bcf86cd799439014",
      "type": "trigger",
      "confidence": "medium",
      "title": "você marcou feijão",
      "description": "parece que feijão não combina muito... você anotou algumas vezes.",
      "data": {
        "foods": ["Feijão"],
        "symptoms": ["gas", "bloating"],
        "occurrences": 3
      },
      "discoveredAt": "2026-02-28",
      "isNew": false
    }
  ]
}

​
Understanding Patterns

​
Trigger Identification

Triggers are identified from the ProblematicFoods collection, which tracks foods users have manually marked as causing symptoms. The system aggregates:
  • Total incidents across all markings
  • Unique symptom types experienced
  • Most recent incident date

​
Time Pattern Analysis

The system buckets all symptom entries by hour of day into 4 periods, then calculates what percentage fall into each bucket. Only significant patterns (>40% concentration) are reported.

​
Safe Food Detection

The system tracks food consumption frequency across all meal entries and cross-references with problematic food markings. Foods with high consumption but no negative marks are identified as safe.

​
Error Responses

​
401 Unauthorized

Returned when the request is missing or has an invalid authentication token. 
{
  "success": false,
  "error": {
    "code": "UNAUTHORIZED",
    "message": "Invalid or missing authentication token"
  }
}

​
Implementation Notes

  • Analysis uses data from the last 30 days only
  • Discoveries are sorted first by confidence (high > medium > low), then by type priority (trigger > combination > time_pattern > safe_food)
  • New discoveries are those with lastIncident within the last 7 days (for triggers)
  • Time patterns and safe foods are not marked as new
  • Minimum thresholds ensure patterns are statistically meaningful:
    • Triggers: 2+ incidents
    • Time patterns: 5+ symptoms with >40% in one period
    • Safe foods: 5+ consumptions with no marks
  • Food IDs are used internally but food names are displayed to users
  • The system automatically handles edge cases (no data, insufficient data, etc.)

Build docs developers (and LLMs) love

Get started for freeTalk to us