BForecast

BForecast is a unified AI-driven platform designed to transform India’s fragmented blood ecosystem into a proactive, insight-led infrastructure for thalassemia patients, donors, blood banks and NGOs.

Comment

BForecast: Making Blood Flow Smarter, Faster, Fairer

What Inspired This

In India, thalassemia patients face a recurring and often exhausting challenge: securing timely, compatible blood transfusions. Despite national efforts like e-RaktKosh, the procurement journey remains largely reactive, manual, and high-stress—especially for children who depend on transfusions every few weeks.

This platform was born from observing the invisible burden placed on patients’ families and NGOs who, without structured tools, rely on WhatsApp forwards and cold calls to fill urgent blood requests. The real need isn’t just more data—it’s real-time intelligence, seamless coordination, and empathetic, tech-enabled support.

Why Each Module Was Built

BForecast is a unified, AI-powered platform designed to transform India’s fragmented blood ecosystem into a proactive, insight-led, and seamlessly coordinated system. Each module addresses a specific operational gap and integrates directly with real-world platforms like e-RaktKosh and community-driven APIs.

1. Conversational Chatbot

Purpose: Provide 24/7 personalized support for thalassemia patients

Problem Addressed: Patients often lack reliable access to health information, records, and coordination tools.

Key Functions:

  • Natural Language Understanding for intent detection
  • Retrieval-Augmented Generation using e-RaktKosh and Blood Warriors data
  • Appointment booking, smart reminders, and escalation to human agents
  • Answers to common and complex thalassemia-related queries

2. Location-Based Donor Center Finder

Purpose: Help patients locate nearby donor centers in real time

Problem Addressed: Lack of visibility into local blood availability leads to wasted travel and delayed care.

Key Functions:

  • Geocoding and proximity ranking of centers
  • Real-time inventory from e-RaktKosh and Blood Bridge API
  • Route planning with wait-time and status updates

3. Patient Medical Record Manager

Purpose: Centralized vault for treatment history and reports

Problem Addressed: Fragmented records across PDFs, paper, and apps make clinical decisions difficult.

Key Functions:

  • GDPR/HIPAA-compliant secure storage
  • Interoperability with FHIR-based EMRs and e-RaktKosh imports
  • OCR ingestion of external PDFs
  • Visual timelines and critical-lab-value alerts

4. Notification & Engagement Hub

Purpose: Sustain engagement with patients and donors

Problem Addressed: Donor retention is low, and patients miss key health tasks due to limited follow-up.

Key Functions:

  • Smart outreach across email, SMS, WhatsApp, push
  • A/B tested campaigns triggered by ML predictions
  • Dynamic engagement based on donor-churn risk and location

5. Analytics & Reporting Dashboard

Purpose: Real-time visibility for blood-bank administrators and NGOs

Problem Addressed: Lack of decision-making support tools and delayed reporting

Key Functions:

  • Live KPIs showing stock forecasts vs actuals
  • Segment explorer for analyzing donor behavior
  • Custom alerts for pipeline gaps, low inventory, and churn risk

6. External Systems Integration

Purpose: Complement existing infrastructure, not duplicate it

Problem Addressed: Data silos between platforms and limited cross-system actionability

Key Integrations:

  • e-RaktKosh API for transfusion logs, inventory, and procurement updates
  • Blood Warriors' Blood Bridge for real-time donor availability
  • Bi-directional data flow for operational efficiency and compliance

How the Architecture Works

[ \text{Data Flow} \Rightarrow \text{Intelligence Layer} \Rightarrow \text{Personalized Actions} ]

  1. Ingestion Layer
    • Pulls data from EMRs, e-RaktKosh, Blood Bridge, and user input
  2. Feature Store
    • Enriches data with behavioral, demographic, and temporal features
  3. Core ML Models
    • Forecast donor availability, alert shortages, personalize outreach
  4. Orchestration Layer
    • Triggers chatbot flows, alerts, and smart campaigns
  5. Frontend Modules
    • Modules consume shared APIs and data to deliver a unified experience

Challenges and Constraints

Area Challenge Strategy
Data Quality Fragmented or inconsistent logs Model-based anomaly detection and fallbacks
Tech Literacy Low smartphone usage in some areas Multi-channel delivery: SMS, IVR, WhatsApp
Privacy & Trust Sensitive health data Privacy-by-design, consent-based sharing
Model Drift Region-specific behavior variation Continuous feedback loops & tuning
API Stability External system reliability Graceful degradation + manual sync options

Assumptions

  • NGOs and hospitals are open to pilot deployments
  • Patients are willing to share data in return for better outcomes
  • e-RaktKosh APIs and community APIs will remain accessible
  • Conversational AI will resolve ≥ 85% of common queries autonomously
  • Regulatory frameworks (DPDP, HIPAA) can be adhered to at scale

Hackathon Timeline Snapshot

Day Milestone
Day 1 Core ingestion + Chatbot MVP + donor finder UI
Day 2 Medical record manager + ML models for donor engagement
Day 3 Admin dashboard + notifications + full end-to-end orchestration

Closing Thoughts

BForecast is more than just a tool—it’s a lifeline.
By using AI to connect intent, need, and availability across the blood ecosystem, it offers patients predictability, dignity, and timely care. It gives donors a sense of meaningful participation, and it provides NGOs and hospitals with the tools to act, not just react.

When intelligence is applied with empathy, infrastructure becomes care.

Built With

Share this project:

Updates