HackStack

Project Info

Winner

Best Use of InterSystems IntegratedML

Health Insurance 4 Dummies

Devpost

Inspiration

💪🏼 Health insurance, everyone needs it, no one wants to pay for it. As soon-will-be adults, health insurance has been a growing concern. Since a simple ambulance ride easily costs up to thousands of dollars, not having health insurance is a terrible decision in the US. But how much are you supposed to pay for it? Insurance companies publish their rates, but just having formulas doesn't tell me anything about if they are ripping me off, especially for young adults having never paid for health insurance.

What it does

? 🔍 Thus, to prevent being ripped off on health insurance after leaving our parents' household. We have developed Health Insurance 4 Dummies. A website utilizing a machine learning model that determines a fair estimate for the annual costs of health insurance, based on user inputs of their personal information. It also uses a LMM to provide detailed information on the composition of the cost.

How we built it

👷🏼‍♀️ The front-end is built using convex-react, creating an UI that takes inputs from the user. The backend is built using python-flask, which communicates with remote services, InterSystems and Together.AI. The ML model is a feed-forward neural network (MLP) for predicting the cost is built on InterSystems using the H2O ML workflow library, trained on a dataset consist of individual's information and their annual rate for health insurance. The explanation of costs is created using Together.AI's Llama-2 model.

Challenges we ran into

🔨 Full-stack development is tedious, especially when the functions require remote resources. Finding good datasets to train the model. Authentication in connecting and accessing the trained model on InterSystem using their IRIS connection driver. Choosing the right model to use from Together.AI.

Accomplishments we're proud of

⭐ Trained and accessed ML model on a remote database open possibility for massive datasets, integrating LMMs to provide automated information.

What we learned

📖 Full-Stack Development skills, ML model training and utilizing. Accessing remote services using APIs, TLS authentication.

What's next

for Health Insurance 4 Dummys 🔮 Gather larger datasets to make more parameters available and give more accurate predictions.

Analysis

Compare with all teams

View

Metric

Figures cover GitHub contributors during the hackathon window. A co-authored commit counts in full for each author, so per-member totals add up to more than the whole-team figures.

Technology

Found in codeClaimed only
  • CSSIn code
  • FlaskIn code
  • HTMLIn code
  • JavaScriptIn code
  • PythonIn code
  • ReactIn code
  • SQLClaimed

6 of 7 appear in the indexed code. 1 claimed on Devpost could not be matched to code, which may simply mean the tool leaves no trace in the repository.

AI coding agents

No AI coding agent signals were found in this repository.

Detected from committed agent config files and commit authorship. Absence of a signal is not proof an agent was unused.

Codebase size

Source size

13 KB

Source files

12

Counts recognized source files only; vendored directories, binaries and lockfiles are excluded, so this is smaller than the repository on disk.

0 stars