@ishitaverma

So basically this is an AI that generates RPG loot. Like when you’re playing a game and you kill a boss and it drops a sword with stats, ForgeAI does that but with AI.

Give it something like type=sword rarity=legendary material=obsidian effect=lifesteal and it spits out a full item, name, lore, and stats like [+18 ATK | +15% Lifesteal | -20 Max HP]. 🔥

The Forge is where you pick the item type, rarity, material, and effect, hit a button, and ForgeAI generates a loot card with rarity colored glowing borders. Grey for common, all the way to a black shimmer for Void-Touched.

There’s also a Gacha system where you pull 3 random items at once with a chest opening animation. Built-in pity system so you’re guaranteed a Legendary every 10 pulls and a Void-Touched every 50. 🎲Everything goes into an Inventory with equip slots like Head, Chest, Weapon, Offhand, and Ring on a character silhouette. Mix and match your best drops.

The Codex tracks everything, word cloud of all the lore generated, rarity distribution charts, and a power score leaderboard of your best items. 📖

The whole UI is pixel-art themed, black background, orange and gold accents, glowing borders, chest animations, shimmer effects by rarity. Feels like an actual game. 🕹️

Project: Multi-disease prediction app (ML + DL) with a Streamlit dashboard

• 🎯 Goal: Build a Streamlit app that predicts risk for 5 diseases - Heart Disease, Diabetes, Breast Cancer, Chronic Kidney Disease, and Parkinson’s

• 📊 Datasets sourced (public mirrors of the classic Kaggle/UCI sets):

  • Cleveland Heart Disease
  • Pima Indians Diabetes
  • Breast Cancer Wisconsin (Diagnostic)
  • UCI Chronic Kidney Disease
  • UCI Parkinson’s voice dataset

• 🤖 Models trained per disease:

  • Heart → Logistic Regression, Random Forest, XGBoost
  • Diabetes → SVM, KNN, Keras MLP (GPU)
  • Breast Cancer → SVC, Random Forest, Keras ANN (GPU)
  • CKD → Decision Tree, LightGBM
  • Parkinson’s → Random Forest, Keras ANN (GPU)

• 💾 Saved each model with its scaler + feature schema (JSON) so the app can auto-generate input forms per disease

• 🖥️ Next: wire up the 3-page Streamlit app

  • Dashboard - EDA & visual insights per disease
  • Prediction - pick a disease → fill clinical inputs → get risk score
  • About - disease descriptions + illustrations

• ⚙️ Training pipeline uses GPU (TensorFlow/CUDA) for the deep learning models