AI for Predictive Donor Availability

Inspiration

->We were inspired by the urgent needs of Thalassemia patients who rely on timely blood transfusions. Despite having active donors, delays happen due to poor coordination. We wanted to build a solution that uses AI to predict donor availability and connect patients with matching donors in real-time, saving lives through smarter timing

What it does

->Predicts when a donor is likely to be available next using AI ->Matches patients with nearby eligible donors instantly ->Sends smart reminders and emergency alerts to donors ->Tracks donor contributions and encourages recurring donations ->Provides a support portal for patients with educational content and chatbot help

How we built it

->Frontend: Flutter app for donors and patients ->Backend: Firebase for real-time data, Node.js for API services ->AI Model: Trained an XGBoost and LSTM model to predict donor availability ->Next_Availability=f(LastDonation,Frequency,Health Data) ->Integrated with Google Maps API and simulated e-RaktKosh-like data

Challenges we ran into

->No access to real blood donation data – we had to simulate it ->Avoiding AI overfitting – solved with model tuning and regularization ->Ensuring user-friendliness for non-tech users ->Managing data privacy and security as per India's DPDP Act 2023

Accomplishments that we're proud of

->Built a working prototype with AI-powered predictions ->Designed a simple, gamified donor interface to encourage repeat donations ->Made a patient-focused support system that’s both helpful and compassionate ->Created a scalable foundation for future NGO and hospital integrations

What we learned

->How to apply AI in healthcare logistics meaningfully ->Importance of data privacy and ethical design ->How to balance technical accuracy with real human impact ->Collaboration with domain experts and users makes the solution stronger

What's next for AI for Predictive Donor Availability

->Real-world pilot testing with Thalassemia centers and NGOs ->Expanding to include platelet and rare blood type tracking ->Integration with national systems like e-RaktKosh and Blood Bridge ->Building multilingual support and offline functionality for wider access ->Exploring blockchain for transparent donor history records

Built With

• ai-model-deployment-?-ai/ml-libraries-xgboost
• and-cloud-functions-google-cloud-platform-(gcp)-?-hosting
• dart
• dialogflow
• firebase
• gcp
• googlemapai
• lstm
• lstm-(tensorflow/keras)-?-predictive-modeling-pandas
• node.js
• numpy
• oauth2
• pandas
• python
• xgboost