PharmaSync AI

Intelligent Medication Timing Optimization Tool

A behavioral-pattern AI project for the HealthTech Innovators Hackathon 2025

---

🌟 Overview

PharmaSync AI is a lightweight, behavior-driven tool that analyzes when users actually attempt to take their medication and uses Machine Learning to recommend the best one-hour window where they are most likely to remember.

Instead of focusing on dosage or diagnosis, PharmaSync AI focuses strictly on habit patterns — giving users a personalized, data-driven adherence schedule.

---

🚀 Key Features

🧠 Smart ML Window Prediction

• Dynamic KMeans clustering (1–3 clusters depending on data size)
• Automatically selects the strongest adherence pattern
• Returns:
  • Best time window to take medication
  • One-hour optimized window
  • Confidence score (%)

✍️ Manual Habit Entry (No CSV Required)

Users can enter:

• Time attempted
• Success or missed
  Data updates live inside the app.

📊 Real-Time Visual Analytics

• Success Rate by Time of Day
• Raw Attempts Scatter Plot
• Highlights high-adherence vs low-adherence time ranges.

📈 Confidence Score

Shows how dependable the AI’s recommendation is based on pattern strength.

📤 CSV Export

All entered data can be downloaded for reuse or analysis.

---

🧠 How It Works

1. User records daily attempts (time + success).
2. ML engine converts time → minutes for clustering.
3. Dynamic clustering:
   • If successes = 1 → simple rule model
   • If successes = 2 → k = 1 or 2
   • If successes ≥ 3 → k = 1, 2, 3
4. The model selects the cluster with the highest success density.
5. Produces the optimal 1-hour window + confidence score.

---

🔧 Tech Stack

• Python
• Streamlit
• Pandas
• Seaborn
• Matplotlib
• Scikit-learn

---

---

📁 Project Structure

PharmaSync-AI/
│── app.py                 # Streamlit UI
│── pharmasync_model.py    # ML clustering engine
│── requirements.txt       # Dependencies
│── README.md              # Documentation
│── habit_data.csv (optional)

---

🌱 Future Enhancements

• Multi-user behavior profiles
• Weekday vs weekend adherence modeling
• Personalized reminder engine
• Behavior drift detection
• Mobile UI version

---

👤 Author

Devi Manoharan Enterprise Quality Engineering Specialist
FOE Portfolio Project
PharmaSync AI – HealthTech Innovators Hackathon 2025 Submission