ThalAssist+

Inspiration

Alright, real talk—the whole ThalAssist+ thing? It’s personal. I’ve seen (way too often) how Thalassemia patients in India basically have to run an obstacle course just to get a blood transfusion. And don’t even get me started on how little people know about the genetic risks. It’s rough on families, emotionally and otherwise. So, with my data science chops and a soft spot for AI-for-good projects, I started thinking: why not use tech to make life even a little bit easier? Maybe help folks get the care they need faster, stop new cases before they happen (hello, education!), and just be there for patients and their people.

What it does

So, hackathon rules say “no code yet”. But I’ve mapped out a plan. Picture this:

BloodBuddy: AI magic to match patients with nearby donors—plus gamified nudges ‘cause hey, people love a good dopamine hit.

ThalGenie: A chill risk assessment tool that nudges people to do pre-marital screening and actually think about genetic stuff.

ThalBot: Multilingual, not-awkward chatbot dishing out health tips, reminders, even diet hacks. Also, just someone to talk to, because why not.

ThalRadar: GIS heatmaps so that NGOs can stop playing guessing games and actually find the right spots for camps/awareness drives.

Oh, and I didn’t just dream this up—I’ve sketched out the features, looked at plugging into stuff like e-RaktKosh and Blood Bridge, and picked a tech stack that won’t make me rip my hair out (think FastAPI, Azure ML, Firebase, all that jazz).

How I plan to build it

ThalAssist+ will be a modular, mobile-first web application built with scalability, accessibility, and low-bandwidth use in mind. Here's the tech plan:

Backend: FastAPI (Python) for managing user data, AI endpoints, and API integrations (e.g., e-RaktKosh, BloodBridge).

Frontend: Flutter or React (depending on time constraints) for a clean, responsive user experience across devices.

Database: Firebase for real-time data sync, authentication, and cloud storage.

AI Models:

• BloodBuddy: A donor-patient matching engine using clustering + real-time location + donor history.
• ThalGenie: A risk assessment engine trained on synthetic/available datasets with user-friendly output.
• ThalBot: Multilingual chatbot using Rasa or Azure Bot Service, tuned for medical intent recognition and empathy.

GIS Module (ThalRadar): Built using QGIS and integrated with Mapbox or Leaflet.js to generate real-time heatmaps for NGOs and public health workers. The design will follow accessibility-first principles: offline fallback, support for older devices, regional language options, and intuitive UI for all ages.

Challenges anticipated

Let’s not sugarcoat it, there’s hurdles:

• Data privacy: Not just a checkbox, it’s a real worry.
• The app has to work even when someone’s stuck with 2G internet on a 2015 phone. No pressure.
• Getting access to APIs like e-RaktKosh? Not exactly a walk in the park.
• Making sure your grandma and your teenage cousin can both use the thing without rage-quitting.

What's next for ThalAssist+

For the hackathon, I’m just aiming to get a barebones MVP running—think chatbot (ThalBot) and basic blood-matching (BloodBuddy). Afterward? Bring in the risk assessment, GIS magic, and maybe even some actual clinical validation if I can rope in the right partners. Here’s hoping, right?

Built With

• ai/ml:-scikit-learn
• azure
• azure-ml
• deployment:
• firebase-or-mongodb-gis/mapping:-leaflet.js
• geopandas-integrations:-e-raktkosh-api
• hugging-face-frontend:-react-native-(app)
• langchain
• openai-api-chatbot:-azure-bot-service
• qgis
• streamlit-or-old-school-html/css-(web)-backend:-python-(fastapi)
• vercel