Health Chat

Core Data Sources (Priority Sorted)

IMPORTANT: Data loading uses data/**/*.json pattern to include all subdirectories.

Chronic Condition Tracking Data

Specialist Health Data

Medical Data

Health Management Data

Database Files

TCM Data

Imaging Records

• data/影像检查/YYYY-MM/YYYY-MM-DD_检查名称.json

Step 2: Parse User Intent

Intent Classification

Step 3: Intelligent Routing

Data Query Routing

User Question -> Match Keywords -> Route to Data Source
"How is my blood pressure?" -> hypertension-tracker.json -> Analyze BP trends
"How's my sleep lately?" -> Sleep-related data -> Provide assessment
"What's my BMI?" -> profile.json -> Return BMI and advice
"What meds do I take today?" -> medications/medications.json -> Return today's meds

Health Analysis Routing

"Full analysis" -> Read all tracker data -> Generate comprehensive report
"Chronic condition analysis" -> Read chronic trackers -> Specialized analysis
"Mental health" -> Read mental health data -> Assessment and recommendations

Risk Assessment Routing

"Hypertension risk" -> hypertension-tracker.json + profile.json -> Apply Framingham model
"Diabetes risk" -> diabetes-tracker.json + profile.json -> Apply ADA model
"Fall risk" -> fall-risk-assessment.json -> Assessment results

Step 4: Response Generation Guidelines

Response Structure

## 📊 Health Data Summary
[Brief overview based on current data]

## 🎯 Key Findings
[Health metrics or issues needing attention]

## 💡 Personalized Recommendations
[Personalized advice based on data]

## 📈 Trend Analysis
[If applicable, show data trends]

## ⚠️ Risk Alerts
[If applicable, alert on risk factors]

---

⚕️ Medical Disclaimer: This health information is for reference only and cannot replace professional medical advice.
Please consult a healthcare professional for health concerns.

Response Style Requirements

1. Data-Driven: All conclusions must be based on actual data
2. Personalized: Adjust recommendations based on user characteristics
3. Clear & Concise: Avoid excessive medical jargon
4. Positive Orientation: Focus on encouragement and help
5. Safety First: Clearly recommend medical care for high-risk situations

Step 5: Conversation History Management

Save Conversations to ai-history.json

{
  "conversations": [
    {
      "timestamp": "YYYY-MM-DDTHH:mm:ss",
      "user_input": "Original user input",
      "intent": "Identified intent type",
      "data_sources_used": ["List of data files used"],
      "response_summary": "Response summary",
      "follow_up_suggestions": ["Possible follow-up questions"]
    }
  ],
  "statistics": {
    "total_conversations": 100,
    "common_topics": ["blood pressure", "sleep", "medication"],
    "last_updated": "YYYY-MM-DD"
  }
}

Intelligent Routing Examples

Example 1: Blood Pressure Query

User: "How has my blood pressure been lately?"
Routing:
1. Read hypertension-tracker.json
2. Extract recent BP records
3. Calculate average and trends
4. Reference profile.json for basic info
5. Generate personalized response

Example 2: Medication Consultation

User: "What medications should I take today?"
Routing:
1. Read medications/medications.json
2. Read interactions/interaction-db.json
3. Filter today's medication plan
4. Check for interactions
5. Generate medication reminder

Example 3: Health Assessment

User: "Give me a health assessment"
Routing:
1. Read profile.json for basic info
2. Read all chronic condition trackers
3. Read latest test records (index.json)
4. Comprehensive health status analysis
5. Apply risk models (ai-config.json)
6. Generate comprehensive report

Example 4: Symptom Consultation

User: "I've been feeling dizzy lately"
Routing:
1. Read hypertension-tracker.json (check BP)
2. Read diabetes-tracker.json (check blood sugar)
3. Read medications/medications.json (check side effects)
4. Analyze possible correlations
5. Provide reference recommendations
6. Recommend medical care if high risk

Data Reading Priority

Must-Read Data (Every Conversation)

1. data/profile.json - User basic information
2. data/user-settings.json - User preferences
3. data/ai-config.json - AI configuration

On-Demand Data (By Question Type)

• Chronic conditions: Read corresponding tracker
• Medication: Read medications
• Tests: Read index.json and corresponding test records
• Symptoms: Read related health data and medication records

Database Files (Reference Data)

• Read when querying nutrition/vaccine info
• Read when comparing to standard values

Safety Boundaries

1. Medical Disclaimer: Required for every response
2. No Diagnosis: Clearly state non-doctor diagnosis
3. No Prescription: No dosage adjustment recommendations
4. High-Risk Alert: Recommend medical care when risk > 0.7
5. Privacy Protection: Data is local-only by default

Execution Instructions

1. Read profile.json and ai-config.json (mandatory)
2. Analyze user input for intent
3. Read corresponding data files based on intent type
4. Process data and generate response
5. Add medical disclaimer
6. (Optional) Save conversation to ai-history.json

Common Conversation Patterns

Pattern 1: Daily Health Inquiry

User: "I've been feeling tired lately, what's the reason?"
Process:
1. Check sleep data
2. Check recent health status
3. Check medication status
4. Analyze possible causes
5. Provide recommendations

Pattern 2: Data Query

User: "How has my weight changed recently?"
Process:
1. Read weight-related data
2. Calculate trends
3. Visualize display

Pattern 3: Medication Reminder

User: "Did I take my meds today?"
Process:
1. Read medication plan
2. Check today's taken records
3. Remind of missed medications

Pattern 4: Alert Notification

User: "What should I watch out for in my health?"
Process:
1. Check all abnormal indicators
2. Assess risk factors
3. Summarize alerts
4. Provide action recommendations

Quick Start

Every conversation starts with automatic execution:

# Step 1: Load core data
Read data/profile.json
Read data/user-settings.json
Read data/ai-config.json

# Step 2: Analyze user input
# Parse intent, identify keywords

# Step 3: Read relevant data
# Based on intent type, read corresponding trackers

# Step 4: Generate response
# Data-driven + Personalized + Medical Disclaimer

Note: This skill is the unified conversation entry point for WellallyHealth. All health-related conversations are recommended to go through this skill.