/GraphPrescribe

"Cost-effective Digital Prescription using Pharmaceutical Knowledge Graph" is a AranongoDB-based knowledge graph containing comprehensive medication data. When provided with a prescription, the system efficiently analyzes the medicines and suggests reliable, cost-effective alternatives.

Primary LanguageJupyter NotebookMIT LicenseMIT

❄️ GraphPrescribe

Also known as "Cost-effective Digital Prescription using Pharmaceutical Knowledge Graph" is a AranongoDB-based knowledge graph containing comprehensive medication data. When provided with a prescription, the system efficiently analyzes the medicines and suggests reliable, cost-effective alternatives.

  • For instance, if the initial prescription costs 900 rupees for three medicines priced at 300 rupees each, the software identifies trustworthy substitutes priced at 100 rupees each, totaling 300 rupees. This results in substantial savings of 600 rupees for the user, maintaining reliability while significantly cutting costs.

  • This is a group project made during internship at AICOE JIO by Aryan Rathore and Reva Bharara under the careful guidance of our mentor Mr. Krishna Kumar Tiwari (head of knowledge graph platform at AICOE JIO platforms limited)

  • This project is proof of concept and is at a prototyping phase

❓Need of the application

Having witnessed the financial strain caused by my father's battle with bone TB and the ensuing ICU expenses, I was motivated to create this project. We aim to help families in similar situations by providing cost-effective healthcare solutions. This software uses a medicine database to offer affordable alternatives, aiming to ease the financial burden while ensuring quality care for those facing substantial medical expenses.

🔬 Research article

🏁 Result

  • Finding a alternative medicine for 1 prescription image

  • The original prescription was Rs.2577 but GraphPrescribe found an alternative prescription of Rs.114 saving 2463 Rs/- image

image

  • Finding a alternative medicine for 1 medicine (Corex LS Sugar Free Syrup) image

image

The original medicines was Rs.103 but GraphPrescribe found an alternative medicines named Ascoril LS Syrup which Rs.97 saving 6 Rs/-

image

  • for Augmentin 625 Duo Tablet (original 452/- alternative 12/-) image

  • The medical_knowledge_graph formed image

image

image

💽Prerequisites to understand the approach

💡 What are knowledge graphs and how to build them?

  • The blog series attached is a 3-part series also made in our JIO internship. this will familiarize you with the concepts needed to understand the approach of this project.
  • https://medium.com/everythingisconnected

📚 About the dataset

📕 a_to_z_medicine_data_web_scraping.csv

  • This csv contains all the info of the medicines.
  • This dataset has 38074 rows where each row represents one medicine.
# Column Non-Null Count Dtype
0 name 38074 non-null object
1 manufacturer 38074 non-null object
2 chemicals 38074 non-null object
3 uses 38074 non-null object
4 side_effects 38074 non-null object
5 Habit Forming 38074 non-null object
6 Therapeutic Class 38074 non-null object
7 Chemical Class 38074 non-null object
8 Action Class 38074 non-null object
9 MRP 38074 non-null int64

image

🖋️ insights

image

image

image

image

image

🛣️ Roadmap of the project

🌐 1_medicine_data_web_scraping.ipynb

  • Objective: The primary goal of this notebook is to execute web-scraping procedures on publicly accessible medicine websites.

  • Data Collection: Utilize web-scraping techniques to extract relevant data from diverse public medicine websites.

  • Dataset Creation: Compile the extracted information into a comprehensive medicine dataset.

  • Saved Dataset: The Compile data was stored in the a_to_z_medicine_data_web_scraping.csv

image

🛠️ 2_medicine_data_statistics_insights.ipynb

  • Objective: This notebook aims to conduct Preliminary, Exploratory data analysis and extract meaningful insights from the gathered medicine dataset before integrating it into the knowledge graph.

  • The insights have been attached above.

🕸️ 3_main_graph_creation.ipynb

  • Objective: The primary objective of this notebook is to construct a knowledge graph of medicines. This graph will serve the purpose of identifying medicine replacements and uncovering hidden insights within the dataset

  • Data Acquisition: Read data from the "a_to_z_medicine_data_web_scraping.csv" file containing medicine information.

  • Database Setup: Create the "drug_repurposing_3" database on ArangoDB and upload columns as collections.

  • Graph Construction: Construct edge collections and upload the remaining dataset, organizing them as edges in the knowledge graph.

image

image

💊 4_graph_queries_medical_kg.ipynb

  • Objective: using qraph aql and a medicine name find an alternative maedicine with a lower cost image

🗔 5_final_prescription_test.py

  • Objective: given a prescription of medicines find alternative medicines that are lower in cost and show the comparision graph
  • TO DO: add proper GUI to this code

🔜 future ugrades

  • The project as of now is just a proof of concept so i does not has proper website or GUI.
  • Only 38k~ medicines have been added for this to be functional we will need to add more medicines
  • Special filters for paitents with medical conditions like sugar, BP or pregnancy need to be added
  • Make a project of the idea rather than a proof of concept.

👥credits and contact info:-