# Project export: PredictAPulseAI This document was generated by HackStack to give an AI agent context about a hackathon project. Sections are labeled with their provenance; content marked as truncated was cut to keep this document small. ## Project metadata - Hackathon: Cal Hacks 10.0 - Tagline: AI-backed health predictor to help you choose the most personalized insurance plan for you. - Devpost: https://devpost.com/software/predict-a-pulse-ai - GitHub: https://github.com/themeghamind/PredictAPulseAI - Demo: https://predict-a-pulse-ai-rofr.vercel.app/ - Team: 3 GitHub contributor(s) — drkchn (50 commits), Megha Jain (16 commits), Adrian Bao (5 commits) ## Devpost submission (written by the team) ### Inspiration The American Healthcare system is expensive and complicated. Everyone wants the best, most cost-effective insurance plan, but choosing one can feel like a daunting task. We were motivated to build a product that could deal with large benefits summaries containing opaque language and support our users' unique medical needs, all the while maintaining a high level of user personalization. ### What it does PredictAPulseAI provides users with a health questionnaire, where the answers serve as input for an ML model for heart attack risk classification. Our ML model was trained using a dataset that has features for causes of heart attacks and predicts future heart attacks. Afterward, users upload summary benefits of insurance policies to PredictAPulseAI, and it combines all of this data to find the most cost-effective insurance policy given your risk for heart attacks. ### How we built it Frontend: JS, React, Next.js, TypeScript, HTML/CSS, Material UI Backend: Python, Flask, MindsDB.api, Tesseract OCR, Open AI GPT (3.5) Database: SQL, CockroachDB Classification Model: Heart Attack Kaggle Dataset, MindsDB ### Challenges we ran into Some of the challenges we ran into were dealing with the limitations of Intel Cloud, specifically its inability to connect to Cockroach DB and port-forward for our custom backend API. ### Accomplishments we're proud of Successful implementations of functions. Replaced cookie usage. Bridged front-end to back-end, API integration, and database implementation. ### What we learned Insurance comparisons and how it helps users based on their current health conditions. We expanded our knowledge of SQL with CockroachDB, ML prediction models with MindsDB, and Flask web server with our custom-written API endpoint. ### What's next This project is not limited to predicting heart attacks and reducing the cost of treatment for our users. Other leading causes of death, such as cancer, can be effectively predicted to do the same pipeline. Most importantly, our project would save lives. We plan to adapt our project to an intuitive mobile application to increase accessibility. We also plan to effectively market our idea and utilize sponsorships from insurance companies to gain funding. ## README (from the GitHub repository) # PredictAPulseAI Personalized AI health outcome predictor and insurance selection guide Deployed Website: https://predict-a-pulse-ai-rofr.vercel.app/ https://github.com/themeghamind/PredictAPulseAI/assets/57309755/eb6f003d-fac2-4944-8f2d-c92eda53162d ## Inspiration The American Healthcare system is expensive and complicated. Everyone wants the best, most cost-effective insurance plan, but choosing one can feel like a daunting task. We were motivated to build a product that could deal with large benefits summaries containing opaque language and support our users' unique medical needs, all the while maintaining a high level of user personalization. ## What it does PredictAPulseAI provides users with a health questionnaire, where the answers serve as input for an ML model for heart attack risk classification. Our ML model was trained using a dataset that has features for causes of heart attacks and predicts future heart attacks. Afterward, users upload summary benefits of insurance policies to PredictAPulseAI, and it combines all of this data to find the most cost-effective insurance policy given your risk for heart attacks. ## How we built it Frontend: JS, React, Next.js, TypeScript, HTML/CSS, Material UI, Vercel Backend: Python, FastAPI, MindsDB.api, Tesseract OCR, OpenAI GPT 3.5 Database: CockroachDB Classification Model: Heart Attack Kaggle Dataset, MindsDB Regression-Based Predictor ## Challenges we ran into Some of the challenges we ran into were dealing with the limitations of Intel Cloud, specifically its inability to connect to Cockroach DB and port-forward for our custom backend API. ## Accomplishments that we're proud of Successful implementations of functions. Replaced cookie usage. Bridged front-end to back-end, API integration, and database implementation. ## What we learned Insurance comparisons and how it helps users based on their current health conditions. We expanded our knowledge of SQL with CockroachDB, ML prediction models with MindsDB, and Flask web server with our custom-written API endpoint. ## What's next for PredictAPulseAI This project is not limited to predicting heart attacks and reducing the cost of treatment for our users. Other leading causes of death, such as cancer, can be effectively predicted to do the same pipeline. Most importantly, our project would save lives. We plan to adapt our project to an intuitive mobile application to increase accessibility. We also plan to effectively market our idea and utilize sponsorships from insurance companies to gain funding. ## Detected evidence (automated analysis) Indexed codebase: 11 recognized source files, 31 KB. - CSS (language) — detected in the code - FastAPI (technology) — detected in the code - JavaScript (language) — detected in the code - Next.js (technology) — detected in the code - OpenAI (technology) — detected in the code - Python (language) — detected in the code - React (technology) — detected in the code - Tailwind CSS (technology) — detected in the code - TypeScript (language) — detected in the code - Flask (technology) — claimed on Devpost, not found in the code - SQL (language) — claimed on Devpost, not found in the code - Vercel (technology) — claimed on Devpost, not found in the code ## Codebase structure (from repository index) ### Files (17 of 17) ``` .eslintrc.json .gitignore .vscode/settings.json app/fileupload/page.tsx app/globals.css app/layout.tsx app/page.tsx app/questions/page.tsx app/results/page.tsx fastapi_backend.py next.config.js package.json postcss.config.js README.md requirements.txt tailwind.config.ts tsconfig.json ``` ### Dependencies - package.json: @emotion/react@^11.11.1, @emotion/styled@^11.11.0, @fontsource/roboto@^5.0.8, @mui/icons-material@^5.14.15, @mui/material@^5.14.15, @types/node@20.8.3, @types/react@18.2.25, @types/react-dom@^18, autoprefixer@^10, axios@^1.7.2, cors@^2.8.5, encoding@^0.1.13, eslint@^8, eslint-config-next@13.5.4, next@^13.5.4, postcss@^8, react@^18, react-dom@^18, react-drag-drop-files@^2.3.10, tailwindcss@^3.3.5, tesseract.js@^5.1.0, typescript@5.2.2 - requirements.txt: charset-normalizer, fastapi, openai@==0.28, pandas, pillow, pydantic, python-dotenv, pytz, requests, six, sqlparse, tzdata, urllib3, uvicorn ### Recent commits (newest first) - Merge pull request #25 from themeghamind/fix-readme-video - Update README.md - Merge pull request #24 from themeghamind/readme-update-1 - Update README.md - Merge pull request #22 from themeghamind/clientside-storage - implemented client-side localstorage of data and api calls - implemented client-side localstorage of data and api calls - Merge pull request #21 from themeghamind/BabyMochi-patch-1 - Update README.md - Fixed apostrophe - Made frontend cleaner - Merge pull request #20 from themeghamind/drkchn-patch-2 - Update README.md - Merge pull request #19 from themeghamind/drkchn-patch-1 - Update README.md - Fixed up Results page [FINAL] - Fixed up Results page - Completed results page, bypassing CORS issue - Merge pull request #17 from themeghamind/derek-megha-frontend - Small edits ## Key source files (fetched from GitHub, selected and truncated for size) ### requirements.txt ``` uvicorn fastapi openai==0.28 requests urllib3 charset-normalizer pytz tzdata sqlparse six python-dotenv pandas pillow pydantic ``` ### package.json ``` { "name": "nextjs", "version": "0.1.0", "private": true, "scripts": { "dev": "next dev", "build": "next build", "start": "next start", "lint": "next lint" }, "dependencies": { "@emotion/react": "^11.11.1", "@emotion/styled": "^11.11.0", "@fontsource/roboto": "^5.0.8", "@mui/icons-material": "^5.14.15", "@mui/material": "^5.14.15", "axios": "^1.7.2", "cors": "^2.8.5", "encoding": "^0.1.13", "next": "^13.5.4", "react": "^18", "react-dom": "^18", "react-drag-drop-files": "^2.3.10", "tesseract.js": "^5.1.0" }, "devDependencies": { "@types/node": "20.8.3", "@types/react": "18.2.25", "@types/react-dom": "^18", "autoprefixer": "^10", "eslint": "^8", "eslint-config-next": "13.5.4", "postcss": "^8", "tailwindcss": "^3.3.5", "typescript": "5.2.2" } } ``` ### app/layout.tsx ```typescript import './globals.css' import type { Metadata } from 'next' import { Inter } from 'next/font/google' const inter = Inter({ subsets: ['latin'] }) export const metadata: Metadata = { title: 'InsuraComp', description: 'Heart attack predictor and insurance policy guide powered by AI', } export default function RootLayout({ children, }: { children: React.ReactNode }) { return ( {children} ) } ``` ### app/page.tsx ```typescript "use client" import Image from 'next/image' import Link from 'next/link' import { useRouter } from "next/navigation" export default function Home() { const router = useRouter() return (
Pulse Logo

InsuraComp

Choose a policy
that’s right for you

We help you cut the crap from health insurance policies
and choose something that suits your needs - not your employer’s
or your insurance company’s, but your wallet’s!

{/*
) } ``` ### app/fileupload/page.tsx ```typescript "use client"; import { useRouter } from 'next/navigation'; import React, { useState } from 'react'; import Tesseract from 'tesseract.js'; import Image from "next/image"; export default function Home() { const router = useRouter(); const [isTranscribingA, setIsTranscribingA] = useState(false); const [isTranscribingB, setIsTranscribingB] = useState(false); const [planAFile, setPlanAFile] = useState(null); const [planBFile, setPlanBFile] = useState(null); const [planAText, setPlanAText] = useState(''); const [planBText, setPlanBText] = useState(''); const handlePlanADrop = (event: React.DragEvent) => { event.preventDefault(); const file = event.dataTransfer.files[0]; setPlanAFile(file); extractTextFromFile(file, setPlanAText, setIsTranscribingA); }; const handlePlanBDrop = (event: React.DragEvent) => { event.preventDefault(); const file = event.dataTransfer.files[0]; setPlanBFile(file); extractTextFromFile(file, setPlanBText, setIsTranscribingB); }; const extractTextFromFile = ( file: File, setText: React.Dispatch>, setIsTranscribing: React.Dispatch> ) => { setIsTranscribing(true); const reader = new FileReader(); reader.onload = (e) => { const target = e.target; if (target && target.result) { Tesseract.recognize( target.result as string, 'eng', { logger: (m) => console.log(m), } ).then(({ data: { text } }) => { setText(text); setIsTranscribing(false); }).catch((error) => { setIsTranscribing(false); }); } else { setIsTranscribing(false); } }; reader.readAsDataURL(file); }; const handleUpload = async () => { const plan_details = { plan_a_details: planAText, plan_b_details: planBText, }; localStorage.setItem("plan_details", JSON.stringify(plan_details)); router.push('/results'); }; return (
Pulse Logo

InsuraComp
e.preventDefault()} >

Drop Plan A File Here

{planAFile &&

{planAFile.name}

}
e.preventDefault()} >

Drop Plan B File Here

{planBFile &&

{planBFile.name}

}
); } ``` ### app/results/page.tsx ```typescript "use client"; import { useEffect, useState } from 'react'; import Image from 'next/image'; import axios from 'axios'; export default function Home() { const [status, setStatus] = useState(''); const [riskStatement, setRiskStatement] = useState(''); const [planDetails, setPlanDetails] = useState({ plan_a_details: '', plan_b_details: '' }); const [checkboxValues, setCheckboxValues] = useState({ gender: false, age: false, chol: false, bp: false, }); const [isLoaded, setIsLoaded] = useState(false); useEffect(() => { const storedCheckboxValues = localStorage.getItem('checkbox_values'); if (storedCheckboxValues) { try { setCheckboxValues(JSON.parse(storedCheckboxValues)); } catch (error) { console.error('Error parsing checkboxValues', error); } } const storedPlanDetails = localStorage.getItem('plan_details'); if (storedPlanDetails) { try { setPlanDetails(JSON.parse(storedPlanDetails)); } catch (error) { console.error('Error parsing planDetails', error); } } setIsLoaded(true); }, []); useEffect(() => { if (!isLoaded) return; const fetchRiskData = async () => { const data = { age: checkboxValues.age ? 1 : 0, sex: checkboxValues.gender ? 1 : 0, cholesterol: checkboxValues.chol ? 1 : 0, blood_pressure: checkboxValues.bp ? 1 : 0, }; try { const response = await axios.post('http://127.0.0.1:8000/checkrisk', null, { params: data }); if (response.status === 200) { console.log(response); if (response.data.risk_score === '1') { setRiskStatement("You have a high probability of suffering a heart attack in the near future."); } else if (response.data.risk_score === '0') { setRiskStatement("You have a low probability of suffering a heart attack in the near future."); } } else { console.error('Failed to send answers'); } } catch (error) { console.error('Failed to send answers: ', error); } }; fetchRiskData(); }, [checkboxValues, isLoaded]); useEffect(() => { const updateRiskAndFetchRecommendations = async () => { if (riskStatement) { const recommendation_data = { risk: riskStatement, plan_a_details: planDetails.plan_a_details, plan_b_details: planDetails.plan_b_details }; console.log("Retrieving recommendations with", recommendation_data); try { const response = await axios.post('http://127.0.0.1:8000/retrieverecs', null, { params: recommendation_data }); if (response.status === 200) { setStatus(response.data.status); } else { console.error('Failed to retrieve recommendations'); } } catch (error) { console.error('Error fetching data:', error); } } }; updateRiskAndFetchRecommendations(); }, [riskStatement, planDetails]); return (
Pulse Logo

InsuraComp

Your Results

{status ? ( <>

{riskStatement}

, newValue: string[]) => { let updatedValues = { gender: false, age: false, chol: false, bp: false, }; newValue.forEach(name => { updatedValues[name as keyof typeof updatedValues] = true; }); setCheckboxValues(updatedValues); localStorage.setItem("checkbox_values", JSON.stringify(updatedValues)); }; const handleSubmit = async () => { router.push('/fileupload') }; return (
Pulse Logo

InsuraComp

Health Questionnaire

Select each question if it applies to you

{ return checkboxValues[key as keyof typeof checkboxValues]; })} onChange={handleCheckboxChange} aria-label="questionnaire" sx={{ '& .MuiToggleButton-root': { borderRadius: '0.5rem !important', '&:first-of-type': { borderTop: 'none', }, }, '& .MuiToggleButton-root:first-of-type': { borderTopLeftRadius: '0.5rem !important', borderTopRightRadius: '0.5rem !important', }, '& .MuiToggleButton-root:last-child': { borderBottomLeftRadius: '0.5rem !important', borderBottomRightRadius: '0.5rem !important', }, }} > Are you male?