DOCKIT/DOKTER

Currently it is difficult to manage or get meaningful o/p from user's health data. Dokter platform runs ML on encrypted data of user, provides accurate predictions to user & helps share report safely.

Comment 2

Dokter

Cant be evil application submission. Please check the images for presentation and demo (click on radio button above to see next image).

Abstract:

You will own your health data, analytics will run on your encrypted data and results will be only with you. Even our dApp doesn't know about the result. Doctors or Labs can make use of our platform to:

  • Predict medical condition of an user (without owing user data)
  • Run more advanced algorithms for various symptoms.

Innovation:

Our dApp is simple for now, currently http://dokterapp.tk platform has enhanced Breast Cancer prediction algorithm with 96.84% accuracy. We innovated to run Machine learning algorithm on encrypted data. This is our original value addition to society. Later many algorithms can be run based on use case & data. Not just another dApp but a platform and so kindly look beyond UI (look into algorithms) powered by Blockstack.

Problem Statement:

Medical records being one of the most sensitive and an important feature of patient health review, control of this data is very critical. Also understanding context behind the illness and predicting accurate measures before it’s too late for our medical system to analyze.

Design Need:

  • Providing complete control over the data, the user can securely and safely provide anonymous identity to share their medical data with others whom he or she desires.
  • Machine learning algorithms on encrypted data to provide the best in class suggestions.
  • A platform to connect participants and researchers to carry out public policy studies and ML training.
  • Easy to use, interactive platform for user.

Solution:

Upon looking at the existing platforms for blockchain applications we have landed up ourselves in a bigger problem of blockchain itself, it was making algorithms to run slow with on-chain computations. So the issue targeted here is that we have built a platform for off chain computation rather than on chain computation.

For the same we have taken medical data as an important aspect, hence have built an platform powered by Blockstack which targets multiple use cases and helps the developer and user let it expand its capabilities by having a single identity.

The task of managing the data of the users health gets very crucial as the user would have a history of a medical condition and may not share everything with the lab or doctor. So we built an dApp which makes use of this history and we can predict the medical condition of the user with more quality data.

To maintain all this data securely we have developed an application which can have a ledger of users data and he/she can securely store it, share it according to their will and wish. This would largely help insurance, govt organizations, doctors, Data scientists and anyone who would need the data. Advanced ML algorithms can be applied to the data to predict the condition and disease.

We have good business model where all players like User, Doctor, Developer, Insurance, Labs will be benefited and market opportunity for our dApp.

Prototype:

We focused mostly on back-end, algorithm accuracy and running ML on encrypted data was challenging. Try our dApp : www.dokterapp.tk (Note - For Version 1, You need to run locally along with your Blockstack Browser and public Gaia hub is not HIPAA compliant). Only final report will be put on Gaia hub (not health data) and report can be shared with intended people only.

Advantages:

  • Using Blockstack makes our algorithms runs faster with off-chain computation
  • We got more quality data from user directly instead of any surveys or from labs
  • User data is owned by user itself and they can share only the needed data to respective API only
  • Once reports are generated, it can be shared to any doctors or labs.

Future Work:

  • We are keeping the commitment to add more ML programs for various types of cancer, improve UI accordingly to make more intuitive and examples of how user can add what type of data & its format.
  • We are working on HIPAA certification for Cloud storage via private Gaia.
  • Integrating with Lightning Networks for payments.

Built With

Share this project:

Updates

posted an update —

Pitch: You will own your health data, analytics will run on your encrypted data and results will be only with you. Even our dApp doesn't know about the result. Doctors or Labs can make use of our platform to:

Predict medical condition of an user (without owing user data)
Run more advanced algorithms for various symptoms.

Innovation: Our dApp is simple for now, currently http://dokterapp.tk has enhanced Breast Cancer prediction algorithm with 96.84% accuracy. We innovated to run Machine learning algorithm on encrypted data which is on GAIA. This is our original value addition to society. Later many algorithms can be run based on use case & data. Not just another dApp and so kindly look beyond UI.

Log in or sign up for Devpost to join the conversation.