From 01af3ab6e2e5cd6d1adb25140acfcfc0979a6e09 Mon Sep 17 00:00:00 2001
From: Moksha2502 <161910293+Moksha2502@users.noreply.github.com>
Date: Sat, 3 Jan 2026 11:57:04 +0530
Subject: [PATCH] Create README.md

---
 README.md | 152 ++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 152 insertions(+)
 create mode 100644 README.md

diff --git a/README.md b/README.md
new file mode 100644
index 00000000..18e03f65
--- /dev/null
+++ b/README.md
@@ -0,0 +1,152 @@
+
+---
+
+# Drug–Drug Interaction Checker (Graph + RAG)
+
+## Problem Statement
+
+When a patient is prescribed multiple medications, some drugs may react with each other. These reactions are called Drug–Drug Interactions (DDIs). DDIs can:
+
+* Reduce the effectiveness of treatment
+* Cause unexpected side-effects
+* Lead to serious or life-threatening complications
+
+Doctors and pharmacists usually check interactions manually using references or tools. This can be time-consuming and may still lead to oversight.
+
+### Objective
+
+To build a system that automatically:
+
+1. Takes a list of medicines as input
+2. Detects risky interaction pairs
+3. Explains the interaction mechanism in simple language
+4. Uses both graph-based relationships and RAG (Retrieval-Augmented Generation) for clarity and accuracy
+
+---
+
+## Dataset
+
+We use the DDInter dataset (Drug–Drug Interactions).
+
+Dataset link:
+[https://www.kaggle.com/datasets/montassarba/drug-drug-interactions-database-ddinter](https://www.kaggle.com/datasets/montassarba/drug-drug-interactions-database-ddinter)
+
+The dataset contains:
+
+* Drug names
+* Drug IDs
+* Drug–drug interaction pairs
+* Description of interactions
+* Mechanism or effect details
+
+This allows us to build a knowledge graph of how drugs interact.
+
+---
+
+## System Design
+
+### Step 1: Input
+
+The user enters a list of medicines, for example:
+
+Warfarin, Aspirin, Ibuprofen
+
+---
+
+### Step 2: Build a Drug Interaction Graph
+
+We use a graph structure where:
+
+* Each node represents a drug
+* Each edge represents an interaction between two drugs
+* Edge properties store risk details and mechanism text
+
+This helps quickly check whether two drugs interact.
+
+---
+
+### Step 3: Interaction Detection
+
+For every pair of drugs entered by the user, we check if an interaction exists in the dataset.
+
+If an interaction exists:
+We flag it and retrieve the mechanism text.
+
+If no interaction exists:
+We report that no interaction was found.
+
+---
+
+### Step 4: RAG-Based Explanation
+
+We use LangChain to retrieve relevant text from the dataset and convert it into simple explanation.
+
+So instead of only saying “Interaction detected”, the system explains why the interaction is risky.
+
+Example explanation:
+Warfarin and Aspirin together increase bleeding risk because both reduce clotting ability in the body.
+
+---
+
+## Technology Stack
+
+Programming Language: Python
+Graph Library: NetworkX or Neo4j (optional)
+AI Pipeline: LangChain (RAG-based explanation)
+Data Handling: Pandas
+
+---
+
+## Example Output
+
+### Input
+
+Warfarin, Aspirin, Metformin
+
+### Output
+
+Pair: Warfarin — Aspirin
+Risk Level: High
+Interaction Reason: Increased bleeding risk
+Explanation: Both drugs interfere with clotting, which increases bleeding tendency.
+
+Pair: Warfarin — Metformin
+Risk Level: None Detected
+
+Pair: Aspirin — Metformin
+Risk Level: None Detected
+
+---
+
+## Expected Outcomes
+
+* Detect risky drug interaction pairs
+* Provide reason and mechanism behind the interaction
+* Ensure responses are explainable
+* Improve medication safety
+* Support healthcare decision-making
+
+This can be useful for doctors, pharmacists, hospitals, and healthcare applications.
+
+---
+
+## Assumptions
+
+* Users provide correct drug names
+* Dataset covers the majority of common interactions
+* The system currently checks interactions only between two drugs at a time
+* Severity level is based on dataset information
+* Patient-specific factors such as age, dosage, or disease conditions are not considered in this version
+
+---
+
+## Future Enhancements
+
+* Add risk severity scoring
+* Add patient-specific risk assessment
+* Build an easy-to-use web interface
+* Support multiple languages
+* Integrate real-time medical databases
+
+---
+