Inspiration
Our journey to create AssistDoc is fueled by a deeply personal tragedy. The persistent and heartbreaking issue of preventable medical errors and deaths occurring annually in Côte d'Ivoire hit home when I lost my own brother due to a lack of crucial information and medical errors. This profound experience illuminated the devastating impact that fragmented patient histories and the immense workload on our dedicated doctors have on clinical decision-making. It became clear to us that while human expertise is irreplaceable, technology, specifically Artificial Intelligence, could serve as a powerful ally to mitigate these challenges and significantly elevate patient safety.
In building AssistDoc during this intensive hackathon, we embarked on a steep learning curve. We gained invaluable insights into the intricacies of healthcare data management, the paramount importance of data security and patient privacy, and the nuanced application of AI in a highly sensitive domain. We learned to prioritize core functionalities for rapid prototyping, focusing on establishing a secure, centralized data backbone and developing an initial AI recommendation engine.
How we built it
Our project was built using Django for a robust backend, enabling efficient data management and API development. For the frontend, we leveraged Django's templating capabilities combined with Tailwind CSS to achieve a clean, responsive, and quickly designed user interface. PostgreSQL, hosted on Supabase, was chosen as our database management system, providing the scalability and reliability essential for a national health platform, a significant upgrade from local SQLite, the Gemini API for medical artificial intelligence, collaborating via GitHub. While our initial architecture is monolithic for speed, we've planned a future transition to a microservice architecture, isolating hospitals and using a message broker like Kafka for seamless event synchronization.
Challenges we ran into
We encountered migration issues with the Gemini API during integration with Django, requiring several technical adjustments to ensure system compatibility and stability.
What we learned
This project allowed us to explore the capabilities of the Gemini API in depth. We discovered its exceptional potential for complex medical analysis and learned how to optimise its performance for critical healthcare applications.
What's next for AssistDoc
Our immediate vision for AssistDoc is to launch a phased rollout in Ivorian hospitals, progressively integrating the system into their daily operations. Beyond Côte d'Ivoire, we plan to expand our solution across West Africa, adapting it to the unique healthcare landscapes of neighboring countries.
To support this ambitious growth, our architecture will undergo a significant shift. We're moving from our initial monolithic design to a robust microservices architecture. This will allow us to treat each hospital, or even specific medical departments, as independent services. This approach dramatically enhances scalability, data security through isolation, and system resilience.
Alongside this architectural evolution, we will develop advanced features such as telemedicine capabilities and predictive analytics. We're also committed to enriching our AI model with local medical data, ensuring its recommendations are culturally and contextually relevant. Key to widespread adoption will be establishing strong partnerships with the Ministry of Health and other critical stakeholders, alongside comprehensive training programs for doctors to ensure proficiency and foster consistency in medical prescribing practices across the region.
Log in or sign up for Devpost to join the conversation.