PROJECT TITLE
MediGuard AI – Agentic Medication
Adherence Assistant
Team Name : Code Z
Team Leader : Naitik Raghav
Team Members :
Ashutosh Raghav
Sarthak Raghav
Ketan Singh
Yash Yadav
Vinod Jena
Registered Mail ID : [Link].ug24@[Link]
College Name : IILM University, Gurugram
Mob. No. : 8860839314
IBM SkillsBuild Team ID : IBMBH00218
INTRODUCTION
MediGuard AI
An Agentic AI-powered medication adherence
assistant that helps patients understand
prescriptions, follow medication schedules,
reduce missed doses, and provide safe
healthcare support through intelligent multi-
agent collaboration.
Smart Medication Risk Caregiver
Reminders Tracking Detection Alerts
SDG 3 – Good Health &
Well-being
PROBLEM STATEMENT
Medication non-adherence is a major
healthcare challenge affecting millions of
patients worldwide.
❖ Missed medication doses
Patients often forget to take medicines due to busy
schedules or forgetfulness.
❖ Prescription misunderstanding
Patients may misinterpret dosage, timing, or medication
instructions.
❖ Poor adherence tracking
Repeated missed doses often go unnoticed without
continuous monitoring.
❖ Delayed emergency detection
Medication overdose or incorrect intake may require
immediate intervention.
⚠ Results in poor treatment outcomes,
increased health risks, and avoidable
hospitalizations.
SDG 3 – Good Health & Well-being
Project Contribution
Improves medication adherence
Reduces missed medication
doses
Supports safe medication
management
Enables timely health
interventions
Impact
Healthier Patients •Better
Treatment Outcomes
Safer & Smarter Healthcare
DATA COLLECTION & PREPROCESSING
Data Collection
Medication Dataset (CSV)
Patient Information
Preprocessing
Validate Medication Records
Load Structured Dataset
FEATURE SELECTION & ENGINEERING
Feature Selection
Patient • Prescription • Schedule • Status
Feature Engineering
Prescription Parsing
Shared State Updates
MODEL DEVELOPMENT
Model Used:
Groq Llama-3.3-70B-Versatile:Large Language Model
for prescription understanding, medication
planning, and patient response analysis.
Integrated with LangChain-Groq for intelligent AI
reasoning.
Workflow Development:
Build the multi-agent workflow using LangGraph.
Connect agents through shared state and conditional
routing.
System Optimization:
Improve reliability using memory management,
guardrails, and structured AI workflows.
Model Evaluation
Evaluation Metrics:
Workflow Execution: Verifies successful execution of all LangGraph
agents.
Prescription Parsing: Evaluates AI accuracy in extracting medication
details.
Response Understanding: Checks correct classification of patient
responses.
Risk Assessment: Validates appropriate risk level identification.
Emergency Detection: Ensures guardrails detect medication
emergencies.
Workflow Validation:
Test the system using multiple medication adherence scenarios to
verify reliable agent collaboration.
System Reliability:
Evaluate memory management, conditional routing, and consistent
response generation across different patient cases.
Tools and Resources
Google Colab & LangGraph:
Description:
○ Google Colab provides a cloud-based environment for
developing and executing Python notebooks.
○ LangGraph enables multi-agent workflow orchestration
with shared state and conditional routing.
Usage:
○ Use Google Colab for code execution, dataset handling,
and project development.
○ Use LangGraph with LangChain-Groq to build and manage
the Agentic AI workflow.
Link to our MediGuard Agentic AI
Code:
[Link]
bg-jIr1l9W8U8HRn?usp=sharing
Result or Outcome:
Result or Outcome:
Result or Outcome:
Result or Outcome:
Result or Outcome:
Result or Outcome:
Result or Outcome:
Result or Outcome: