Motivation
This project is motivated by the acute need of thalassemia patients who depend on regular and timely blood transfusions to survive, and face delays, mismatched and scarcity due to inefficiencies in the blood donor system.
It is feasible to leverage AI in tandem with platforms like eRaktKosh to:
- forecast patient needs
- correctly match donor to recipient
- improve the distribution of blood products
Ultimately, the goal is to save lives, while also being *ethical and transparent in a healthcare system that builds trust.
What it does
Integration Features of the AI-Based Blood Transfusion System
Integrate Blood Bank Data with Hospitals Seamlessly connect hospital systems with blood bank inventories to enable real-time access to available blood units across multiple centers.
Match Patients with Best Donors
Use AI algorithms to identify the most compatible donors based on blood type, antibody profiles, location, and medical history.
Filter Donors Based on Availability Automatically filter out ineligible or unavailable donors by checking recent donation dates, health status, and donation frequency.
Notify Available Donors When Needed Instantly alert eligible and nearby donors during emergencies through SMS, app notifications, or email to ensure timely blood supply.
How we built it
Methodology Overview: AI-Based Donor–Patient Compatibility System
Data Collection & Cleaning
- Pre-existing thalassemia datasets are gathered.
- Data is cleaned to remove inconsistencies and missing values.
- Relevant features such as blood group, hemoglobin, and diagnosis are selected and encoded.
- Pre-existing thalassemia datasets are gathered.
Model Training
- A machine learning model is trained to predict donor–patient transfusion compatibility based on selected features.
Real-Time Data Integration
- Donor and patient data is fetched directly from integrated hospital and blood bank systems in real time.
Data Preprocessing & Prediction
- Incoming data is preprocessed using the same pipeline as the training phase.
- Preprocessed data is passed into the trained ML model for compatibility prediction.
- Incoming data is preprocessed using the same pipeline as the training phase.
Model Deployment
- The trained model is deployed on a secure backend server.
- An API endpoint is provided for live compatibility checking.
- The trained model is deployed on a secure backend server.
Web Dashboard Interface
- A user-friendly dashboard enables hospitals and blood banks to:
- View donor and patient records
- Manage compatibility checks
- Instantly match donor–patient pairs
- View donor and patient records
- A user-friendly dashboard enables hospitals and blood banks to:
Challenges Faced During Research
- As all the team members are Computer Science Engineering (CSE) students, exploring the medical field was a new and unfamiliar experience for us. Thalassemia is not a widely known disease, which made the research process a bit more challenging. Understanding the medical terminology, disease mechanisms, and treatment options required extra effort and careful study.
- During model training , the data set has been mostly inconsistent making us do a harder job in preprocessing the data
Accomplishments that we're proud of
- Successfully developed HemoMatch, an AI-powered system capable of efficiently matching Thalassemia patients with suitable blood donors.
- Gained a deeper understanding of the medical data related to blood types, genetic compatibility, and the specific needs of Thalassemia patients.
- Overcame the challenge of working in a domain outside our core field by conducting thorough research.
- Implemented machine learning models capable of analyzing donor-recipient compatibility based on real-world constraints and medical guidelines.
- Built a clean and user-friendly interface to ensure ease of use for healthcare providers.
What we learned
- How to work with healthcare datasets, including preprocessing sensitive data and handling privacy concerns.
- Practical experience in designing and training AI models for matching algorithms based on specific criteria like blood type, age, and availability.
- Deeper insight into Thalassemia as a genetic blood disorder, and how donor-patient compatibility can significantly affect treatment outcomes.
What's next for HemoMatch -An AI that matches patients with the proper donor
- Adding support for multi-language interfaces to make the application more accessible to users in rural and diverse regions.
- Collaborating with medical professionals to validate the AI's recommendations and ensure the system meets clinical standards.
- Expanding the scope of the platform to support patients with other rare blood disorders beyond Thalassemia.
Log in or sign up for Devpost to join the conversation.