Rakt Hak

LifeLink uses AI to instantly connect Thalassemia patients with nearby blood donors. Our platform tracks donation and predicts future donor availability using time series forecasting.

Comment

Rakta Hak

🩸 Core Problem Statement

How can we develop real-time systems to connect Thalassemia patients with blood donors and predict donor availability?

🎯 Primary Solution Components

1. Real-Time Donor-Patient Matching System

  • GPS-based matching of nearby eligible donors (using google-maps-SDK & PlacesAPI)
    Dataset: E-RaktKosh APIs and https://www.friends2support.org/
  • Immediate notifications via app/SMS/WhatsApp
  • Blood group compatibility and last donation validation
  • Patient urgency flag to prioritize cases
  • Website development for centralized access
  • Urgent alerts based on geolocation and Places API
  • Centralized donor profiles with comprehensive information
  • Advance notifications - donors can notifywhether they are coming or not via app, which:
    • Give organizations estimate of number of people attending to better arrange resources
    • Enable time slot assignments for better coordination

2. AI Chatbot for Awareness & Support

Problem: Lack of awareness leads to unplanned pregnancies with Thalassemia major

Solution:

  • AI-powered awareness chatbot (multilingual)
  • Educates about prevention, testing, carrier screening
  • Guides families on treatment, subsidies, NGOs, government schemes

3. Regional Cluster Network

  • Enable small regional clusters (e.g., by district) where patients and donors are locally connected for faster action
  • Rather than relying only on centralized systems

📊 Donor Recognition & Analytics

🔍 Contribution Tracking System

How can we ensure seamless tracking of donor contributions to encourage recurring participation?

Implementation Strategy:

  • Every donor gets a unique ID upon registration (via phone number or Aadhaar)
  • Each donation event is logged with:
    • Date & time
    • Location (blood bank/hospital)
    • Recipient ID (if applicable)
    • Verified by authorized personnel (staff/admin QR scan)

Technical Implementation:

  • QR code generation per donor
  • Blood donation centers can scan the donor's code to log real donations
  • Alternatively, link with e-RaktKosh API to sync donation records automatically if donor data is public or accessible via consent

Centralized Contribution Record

Allow donors to:

  • View and download all past donations
  • Use their record for incentives (e.g., college CSR hours, government benefits if any)
  • Opt in to make their contribution public (e.g., leaderboard)

Donor Impact Score

Build a "Donor Impact Score" — a cumulative value based on:

  • Frequency of donation
  • Urgency of cases responded to
  • Willingness to donate rare groups
  • Engagement level with the platform

Donor Dashboard with Analytics

Key sections:

  • Total donations
  • Total lives impacted (estimated)
  • Locations donated at
  • Badges/milestones earned
  • Next eligible date
  • Streak (e.g., 3 consecutive donations)

This makes their impact tangible and builds an emotional loop

🔮 Forecasting: Predictive AI Models for Donor Availability

We use Prophet time-series forecasting to predict blood collection patterns and enable proactive donor outreach.

📊 State-Level Blood Collection Forecasting

🎯 Goal:

Use historical state-wise blood collection data to predict future blood supply trends — enabling proactive donor activation before shortages occur.

🗂️ Data Source:

  • Government Dataset: State/UT-wise blood units collected (2018-2022 H1)
  • Key Variables: Blood units collected, number of reporting blood banks, temporal patterns
  • Coverage: 35 States/UTs with varying data completeness

⚙️ Why Prophet?

  • Handles missing data (common in government datasets)
  • Automatic seasonality detection (festivals, emergencies affect donations)
  • Changepoint detection (identifies COVID-19 disruptions automatically)
  • Uncertainty quantification (provides confidence intervals for risk assessment)

🛠️ Implementation Approach:

1. Data Preprocessing
# Transform to Prophet format: ds (date), y (blood units), regressors
df_prophet = transform_to_timeseries(state_data)
# Handle missing values and outliers
# Add external regressors (blood bank count, population)
2. Model Training
# Train separate Prophet model per state
model = Prophet(yearly_seasonality=True, changepoint_prior_scale=0.05)
model.add_regressor('blood_banks_count')
forecast = model.predict(future_dataframe)
3. Validation & Accuracy
  • Backtesting: Train on 2018-2021 → Test on 2022 data
  • Target MAPE: <20% for reliable predictions
  • Cross-validation: Time-series split validation
4. Forecast Output
{
  "state": "Delhi",
  "next_6_months": {
    "predicted_collection": 180500,
    "confidence_range": [165000, 196000],
    "trend": "decreasing",
    "risk_level": "high"
  },
  "donor_activation_needed": 72200
}

🎯 Smart Interventions:

Early Warning System
  • 3-month advance alerts for predicted shortages
  • State-wise risk scoring (Low/Medium/High)
  • Seasonal pattern recognition (Ramadan, Diwali impact on donations)
Proactive Donor Outreach
  • Targeted campaigns 4-6 weeks before predicted shortages
  • Personalized messaging based on historical donation patterns
  • Resource allocation for blood drives in high-need areas

📈 Expected Impact:

  • 25-30% improvement in donor response rates through timely outreach
  • Reduced emergency shortages with predictive intervention
  • Optimized blood bank operations with demand forecasting
  • Data-driven policy decisions for health departments

🚀 Technical Stack:

  • Prophet for time-series forecasting
  • Python/Pandas for data preprocessing
  • FastAPI for prediction endpoints
  • Real-time dashboard for stakeholder monitoring

🔔 Automated Engagement System

Automated Reminders & Nudges

Use simple cron jobs to:

  • Send reminders at 90-day cooldown expiry
  • Celebrate donation anniversaries
  • Alert if patient nearby needs the same blood group

Delivery Channels:

  • WhatsApp (via Firebase cloud messaging)

- Email (via email.js)

📅 Scheduling and Supply Management

Donor Scheduling

  • Organize rotation schedules so that donors are invited to give blood as soon as they become eligible (every 8-12 weeks for whole blood)
  • Use IT systems or donor management software to remind and re-invite donors as they reach eligibility

Blood Drives

  • Arrange frequent blood donation camps in schools, colleges, companies, and communities
  • Coordinate event dates so that blood supplies are steady throughout the year, not just during annual drives

Emergency Contact Lists

  • Maintain a list of registered donors who can be contacted during urgent situations or shortages
  • Sometimes called a "walking blood bank"

🎯 Personalization Features

Personalization and Patient Groups

  • Patient-Donor Matching: In some cases, allocate regular donors specifically for patients with rare blood types or who have developed antibodies (to minimize delays and complications)

🔗 System Integration

Integration with existing systems like e-RaktKosh and Blood Warriors' Blood Bridge initiative

Real-Time Blood Group Availability

Example Workflow:

  1. Patient requests B+ in Surat
  2. Your system:
    • Queries e-RaktKosh
    • Gets a list of nearby blood banks where B+ is available
    • Also checks your local donor DB
    • Sends both options to the patient

Blood Bank Locator + Stock Info

  • On your site: "Find Nearest Blood Bank with A+"
  • Call e-RaktKosh and show stock + contact info
  • Use e-RaktKosh as a fallback or cross-check layer, while your core system handles:
    • Real-time donor tracking
    • Donor reminders
    • Patient urgency alerting
    • Last-mile coordination

Fallback Message: "B+ blood is available at XYZ Blood Bank (5.3 km away). Contact: 1234567890"

🚀 Expected Impact

This system aims to create a seamless, efficient, and scalable solution for connecting Thalassemia patients with blood donors while building a sustainable donor engagement ecosystem. Using time series forecasting to help health organizations better plan their blood camps base on predictive donor availability.

Built With

Share this project:

Updates