SehatChain

Project Title

SehatChain

Project Area of Specialization BlockchainProject Summary

In today’s technologically advanced society, where data is the most valuable asset, no one wants to divulge their data and everyone needs adequate data security. Many strategies have been presented to achieve data security preservation. Data security is achieved using techniques such as k-anonymity, l-diversity, and others. On the other hand, some solutions, such as K-anonymity, are a poor approach to syntactic privacy that is susceptible to attribute disclosure, homogeneity, and background knowledge threats. Thus, in this work, we describe a safe and resilient data-sharing architecture based on blockchain, local differential privacy, and federated learning. This design creates a trustless environment in which data owners no longer have to trust the controllers. The federated learning models allow the entire network to learn in a decentralized manner using their actual data. Data will be sent to data consumers using a secure and safe peer-to-peer communication channel. Because of the robust privacy guarantee, data owners do not need to be concerned about the security of their data. Hopefully this proposed model will provide good privacy and accuracy when compared to current methods in terms of latency, throughput, and accuracy.

Project Objectives

Our first goal will be to develop a permissioned blockchain to secure the data and to confirm the authentication of the identities of Data consumers, Data controllers and Data owners via Registration on permissioned blockchain.

After this our next goal will be to perform data anonymity to make sure the patient’s privacy of the sensitive data will be preserved. For this purpose, Local differential privacy can be applied in which the data is infused with such noise that it is difficult to attribute any of it to a particular attribute.

Federated Learning (FL) is an emerging technology that offers privacy-preserving model training across a large area. In the proposed framework patients’ data is completely secure; they only share their trained models using federated learning.

Project Implementation Method

Electronic medical records can provide many benefits to patients and healthcare professionals if they are adapted by the healthcare organizations but concerns about the privacy of patient’s data can cause hurdles in adopting electronic medical records by masses. We will overcome these privacy issues with the following solution:

In our model the data owners are the patients, the hospital is the data controller, and data consumers are the persons or organizations who desire the patients’ dataset. Permissioned blockchain is used to register all three participants and to interact between data consumers and controllers. Here are some of the steps in the proposed solution that are explained.

• Registration on blockchain: To validate identities, data controllers, consumers, and owners registered on the blockchain.

• Publish date bid on blockchain and specs on IPFS: Firstly, the data consumer publishes a bid on the blockchain and dataset specs of bid on the Interplanetary File System (IPFS). A bid contains the link to the specs of the dataset, expiry date, amount, and tags while the dataset spec file on IPF contains the schema: age, gender, ZIP, symptom, medical procedure, minimum number of records, analytical privacy conditions, description.

• Get and detect bid from permissioned blockchain: Data controller gets the dataset spec and detects all the interest data bids.

• Request for getting data: Hospitals must now communicate with data owners to obtain their datasets, but the data owners will never provide the original data.

• Publish dataset specs: Data controller responds to the bid with an offer after getting the updated global model. At the same time, the data controller will publish the offer’s dataset specs on IPFS, which is an off-chain storage system.

• Interchange identity proofs: It begins with the use of a secure P2P communication channel to exchange identification proofs by both sides.

• Verify identity: Both the data controller and the consumer validate each other’s identity by using a self-sovereign identity ledger.

• Share Global Model: When the verification is established, the global model is shared with data consumers over a secure P2P communication channel.

Benefits of the Project

The main motivation of our project is not only to secure the medical data but provide a secure platform to those identities which do not trust each other. Data anonymity will be another challenge. The use of blockchain has opened the doors of many possible ways to secure important information in a decentralized manner. The immutability and Security of the blockchain with other Machine learning techniques can solve major privacy and data breaches issues. One of the main challenges in this project will be merging Machine learning concepts with blockchain.

Technical Details of Final Deliverable

Our software will have the following modules

• Verification Module: Module to verify the patients/data owners and the consumers through OTP or Email verification.
• Registration Module: Module to register patients and healthcare organizations.
• Patient Module: Module for the patients to review and visit their data that was entered through the registration module.
• Organization Module: Module for healthcare organizations to add new patient data and record into the blockchain
• Bidding Modules: Module to place bids for the data of the patients through the controller.

Final Deliverable of the Project Software SystemCore Industry ITOther Industries Medical , Health , Security Core Technology BlockchainOther Technologies Artificial Intelligence(AI)Sustainable Development Goals Good Health and Well-Being for PeopleRequired Resources

Item Name	Type	No. of Units	Per Unit Cost (in Rs)	Total (in Rs)
			Total in (Rs)	60000
GTX 1080ti (Used)	Equipment	3	20000	60000