Aslan

Background

In today’s fast-paced healthcare environment, patients often face challenges in accessing, managing, and sharing their health data securely. From wearable devices to lab reports, the sheer volume of health information can be overwhelming and fragmented for the average user.

Inspired by the potential of decentralized identity systems and AI-driven predictive insights, I envisioned a solution that empowers patients to take control of their health journey while fostering collaboration with existing healthcare providers. I guess a skimmed through James Davidson's The Sovereign Individual last weekend, too. Lol.

What it does

My web application, Aslan, provides patients like Alice with a secure platform to sign up using decentralized identifiers (DIDs) and prove medical health data using Zero Knowledge Proofs when chatting with primary care providers like Dr. Singh via an agentic AI-powered chat interface.

Patients can import medical records from wearables and labs, chat with an AI agent for predictive health insights and actionable suggestions, and even schedule appointments or share health data securely.

On the provider side, doctors gain access to AI-powered recommendations, enabling them to prescribe personalized treatment plans and ensure the best possible outcomes for their patients.

How I built it

I developed the app using Next.js for the frontend, paired with a GraphQL API powered by Hypermode Modus for backend operations along with knowledge graphs powered by DGraph, not Neo4J.

To ensure secure authentication, I implemented the Privy SDK and used Identity.com's DID:SOL standard for decentralized logins.

AI functionalities, such as predictive insights and health suggestions, leverage models from Hypermode Modus and HuggingFace, while patient and provider data is managed with DGraph to enable intelligent querying and data relationships.

Challenges I ran into

I couldn't finish optimizing my knowledge graph and building the UI in time but the foundation is there. This is such a rich concept ("The Self-Sovereign Palantir of Healthcare") that I wanted to make sure all edge cases were sastified for a full-stack solution before continuing specifically in one area (data, graphs, API, frontend)

Therefore, you will find attached the deployed DGraph Cloud instance, Hypermode Modus API instance and a vercel demo instance that actually allows you to login in with Privy SDK via your email. You can use https://temp-mail.org/en to test it out if you don't feel comfortable sharing your real email.

Furthermore, integrating decentralized identity (DID) systems with user-friendly interfaces proved to be a delicate balance between security, adhering to regulations and usability.

Creating seamless interactions between the AI agent, datasets, and the scheduling system required significant optimization which I couldn't completely implement in the short time frame that I picked up this project (~3 days).

But, I want to continue to ensure compliance with data privacy and healthcare regulations was an ongoing challenge throughout the build that I look forward to building in the future through differential privacy in addition to Zero Knowledge Proofs and Decentralized Identifiers

Accomplishments that I am proud of

I successfully implemented the beginning foundation for a fully decentralized identity system for secure patient registration, allowing patients to retain ownership of their data.

The integration of chat-based predictive AI insights and health suggestions delivered personalized value to synthetic patients, while the smooth connection between patient data and provider tools ensured a collaborative experience.

Most importantly, I created a solution that demonstrates the possible potential to bridge gaps in modern healthcare.

What I learned

This journey taught me the importance of designing systems that prioritize both security and user experience. I gained valuable insights into the complexities of managing healthcare data and ensuring compliance with privacy regulations like HIPAA and FHIR.

Additionally, I learned how to leverage knowledge graph databases like D-Graph for efficient querying and discovered best practices for integrating AI models into real-world applications.

What's next for Aslan

I aim to expand the platform’s capabilities by integrating additional wearable devices and health data sources, providing patients with a more comprehensive view of their health. Enhancing the AI agent is a top priority, focusing on delivering detailed health insights, predictive analytics, and actionable recommendations. Future updates will include advanced multi-provider collaboration features and streamlined scheduling tools to create a seamless experience for patients and providers.

Achieving true privacy when training the AI agent's model on patient data is a critical milestone, leveraging differential privacy-enabled analysis. I want to maintain a robust cryptographic setup to balance security, smooth user experience, and interoperability across platforms and jurisdictions, all while ensuring FHIR compliance. Addressing long-term data retention, I'll ensure patients retain access to their healthcare data even after periods of inactivity so that they don't lose custody of their data.

I plan to continue leveraging technologies like zero-knowledge proofs (ZKPs), decentralized identifiers (DIDs), and knowledge graph-based ontology with AI to create privacy-first, actionable medical insights. This approach empowers everyday citizens with accessible, secure health analytics in a way reminiscent of platforms like Palantir Gotham or Anduril Lattice.

Looking forward

Looking ahead, the Aslan app will integrate with existing healthcare applications via FHIR compliance, onboarding real practitioners to the beta program for real-world validation, and expand its functionality with health automation and real-time monitoring akin to MyFitnessPal.

Additionally, I aim to support telemedicine prescriptions, pharmacy payments, asset streaming, and encrypted messaging between patients and doctors to ensure secure and confidential communication.

There's a significant amount of work ahead, but the potential of this application is immense.

🦁 Closing Statement: Thank You for Trying Aslan!

I hope this hackathon product demonstration inspired you to continue building out the Hypermode platform for full-stack developers like me. I believe Hypermode is the next Vercel, but for AI Applications! (wen GPU hosting)

Technology stack

• Frontend: Next.js, Zod, ShadCn, TailwindCSS
• Authentication: Privy SDK, DID:PRIVY:XXXX
  
• Backend: Hypermode Modus, GraphQL API
  
• AI Models: HuggingFace via Hypermode Modus
• Database: DGraph
  
• Zero-Knowledge Proofs: Snark.js, Circom
• VC Issuance: DID-JWT-VC library
  

Built With

• modus