Hyperledger Foundation Hyperledger Global Forum 2021, 11 Jun 2021

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Blockchain-based Decision-support and Longitudinal Data Storage, with... - Alevtina Dubovitskaya

Description

Blockchain-based Decision-support and Longitudinal Data Storage, with Illustration for Diabetes Type 2 Management - Alevtina Dubovitskaya, HSLU (Lucerne University of Applied Sciences and Arts)

Clinical decision pathways evolve and demand multiple assessment steps, access to distributed data sources, and orchestration of various processes executed by diverse consortia of stakeholders. Alevtina will present a demo of a blockchain-based system for decision support and longitudinal data storage, with a use case based on the treatment of diabetes type 2. The system allows encoding dynamic processes and rules that access relevant data and guide clinicians and patients towards optimal decisions in a shorter time. The Business Process Management (BPM) technology is layered on top of a blockchain-based architecture that leverages private data collections and multiple channels for privacy and separation of concerns. The use of blockchain also allows collecting data from the everyday care of broad populations, opening an opportunity for a fundamentally new style of medical research.

Sessions Details & Slides (if available)- https://sched.co/j3do
Access All Sessions: https://events.linuxfoundation.org/hyperledger-global-forum/program/schedule/

----
Hyperledger – Open Source Blockchain Technologies

Hyperledger is an open source community focused on developing a suite of stable frameworks, tools and libraries for enterprise-grade blockchain deployments.
It serves as a neutral home for various distributed ledger frameworks including Hyperledger Besu, Fabric, Sawtooth, Indy, as well as tools like Hyperledger Avalon, Cactus and libraries like Hyperledger Aries, Ursa. Learn more about Hyperledger projects: https://www.hyperledger.org/use

Case Studies: https://www.hyperledger.org/learn/case-studies
Training & Certification: https://www.hyperledger.org/learn/training
Tutorials: https://www.hyperledger.org/use/tutorials
Webinars: https://www.hyperledger.org/learn/webinars
Events: https://www.hyperledger.org/events
Vendor Directory: https://www.hyperledger.org/use/vendor-directory

Subscribe to the Hyperledger Newsletter: https://www.hyperledger.org/newsletter

Follow-us on Twitter @Hyperledger

Learn about Hyperledger Membership: https://www.hyperledger.org/about/join

#Hyperledger #HyperledgerGlobalForum2021 #Blockchain

A

Hi welcome everyone. My name is alif tina and I would like to in this presentation uh I would like to share with you um our work in progress um in the research in the framework of the research project that uh aims to build blockchain, based decision support and longitudinal data storage with illustration for diabetes type 2 management. So this work is done in collaboration with rick who luca matsoler christian, reynolds and peter novotny.

A

So evidence-based medicine requires massive data acquisition and it's accessible for diagnosis, treatment and research purposes. In order to actually put in place uh such um approach, evidence-based medicine, uh we need to address following um following things: we need to enhance enhanced decision, support system. We need to be able to collect historical patient data and to share it in a secure and privacy aware manner and share it, both for the diagnosis for the treatment and also for research purposes.

A

So in order to um uh enhance those those three different points, we propose to use business process and rule management and blockchain to to assist healthcare decision and allow implementation and automatization on the following the latest clinical protocols.

A

um Our use case scenario is a chronic disease management and we have chosen diabetes, type 2 and the patient that have diabetes. They have to go to doctor about every three six months and basically the following tasks happen. Every time that comes to the doctor, so first, the doctor um using is using some guidelines, the existing guidelines and his experience to identify the best treatment plan for the patient.

A

Then, when the plan is selected, then it has to be checked against other medications that patients is maybe taking and also the the safety of the drug and possible adverse effects. Then uh there has to be a check for especially for the um expensive and new medication.

A

There is a need to um check whether this medication will be reimbursed by the insurance company. Then uh the patient has to accept either possible adverse effects, or maybe they need to pay for the indication in case if it's not reimbursed by his insurance company and then also um the data has to be collected. The data about patient reactions on the treatment and the progress and in the in the best case scenarios some data could also be collected for the medical research. So here we try to uh modeling those flow.

A

We also try to specify different databases that are public or private, and also uh the the data or whether those databases contain uh only the data or also the rules such as, for example, um guidelines for the from the medical literature.

A

So the the difficulty of implementing such system that would basically allow a smooth execution of those tasks and optimization or and their optimization uh lies in the following.

A

So what we have here is the multiple dynamic uh data ware processes such as execution of guidelines, simulation of reimbursement, decision making and collecting of the of the outcome data, and we have overlapping set of participants that may have only partial access to different uh types of data, including sensitive patient data, and also all those participants can in different manner, influence the full decision making regarding the treatment- and here we're talking about patients, clinicians researchers, insurance and pharmaceutical companies and regulators.

A

So we propose the following architecture that captures this um flow with the tasks that I just presented and also allows to achieve the following functional requirements. So our system architecture consists of three layers that could be represented. Each of each of them can be represented as the separate blockchain or separate channel and uh first layer uh concerns intelligent logic. So this is the this um blockchain, um that is, that can be uh seen by different participants.

A

uh But only designated um actors could write. Data could could basically uh add lines and then those guidelines can be verified by other participants like different medical doctors, the same as the reimbursement and their prescribing information can also be verified by the regulators.

A

Then the second layer is the data processing layer that basically, um is a workflow orchestration orchestration of those tasks that is initiated by the clinician, and that is more concerned to concerned, with the specific case, specific specific patient case and involves patient data and specific medical guidelines that are required for this patient.

A

Given uh it's given his condition, then uh we have also data privacy layer that responsible for consent, management and here is where our patient can specify what data can be uh used by the doctor, for example, to execute those tasks and to for which doctor which doctor can access uh which type of data generated by also different practitioners or maybe stored in different hospitals. We also assume that there is a off-chain data storage uh that can have either sensitive uh patient information or also data for medical research, so the system works as follows.

A

uh When a patient comes to the doctor, the doctor can uh provide some input information about the condition of the patient. So in our case it could be diabetes, type 2. and then, um from this first lecture from the intelligent logic ledger, the guidelines are sent to the data processing ledger and then um then we know basically, which patient data is required.

A

In order to uh run to execute those guidelines, then, uh when we know which type which date is required, then the information sent to the data privacy ledger and to to check whether there is a consent from this patient, so that this doctor can access this type of data. And then um the rules are executed.

A

And here we see that the either the data can be sent with the consent of the patient to the data processing ledger or uh we can use the private data collection in order to uh keep the data uh and in the hospital and not share it. On the ledger and like this, it goes for different tasks next, one uh for the, for example, for the prescribing information where this flow would actually be repeated.

A

uh So now, uh let's focus on how we, um how we implement this uh medical, how we implement this uh first layer with intelligent logic, what it consists of, so we started for this specific example. We started with the official recommendations for the treatment with the guidelines for treatment of type 2 diabetes, where we had the text and from this text we constructed a case management models which basically um shows different outcomes based on the different uh rules and different input information. So here we first related whether uh using business process, management and business process.

A

Modeling annotation could be also an option and we figured out that using case management is actually a better because it fits better the the structure of those text guidelines. Also, we don't. We cannot have it as structured as business process, as is in bpmn in business process management, because sometimes we don't know which data arrives first and um it's really.

A

The the structure of the guidelines is really fits for the cmm, then, uh from my case management models, we created jsons that basically describe the the the rules for um that basically implements the logic of the guidelines and what what we have represented in cmmn and then we used open source json rule engine. uh That also leaves on this uh intelligent ledger.

A

So here is a specific example where we have a part of our case management model, uh for example. uh This is a representation of the logic of how touch as the treatment for patients with chronic kidney disease where we have to have- and here is the representation in json manner that describes uh for each of the outcomes, depending on the input input, data such as bmi and eegfr.

A

uh What are the rules and based on that that will be applied to the data to the input data that will specific participation and that will basically automatically suggest to the to the medical doctor that this is the um suggested medication. According to the.

A

So now um in this in this small demo, I would like to demonstrate um how we implemented this specific piece of logic uh and um on our intelligent lecture. um That is, that we call guideline channel uh because we used uh hyperledger fabric and we um plan to have three different channels. So so far we have uh two channels. One is the guideline channel that stores all this information about the guidelines and the second channel is the treatment channel that is concerned with the specific patient cases and implementation of those guidelines.

A

So basically, here you will see a short video where a doctor is able to uh initiate to to basically get the the required um required model from the guideline channel and then it is sent to the treatment channel and then it is executed using um patient data.

A

So here is just the demonstration that we have those two guideline channels created. We have hospitals, insurance, company and regulators, where a doctor is has rights. Rights to write on this treatment channel and the insurance company and regulators are able to view the information.

A

So here's just the screenshots from our patient data, smart contract that is deployed on the treatment channel. As you can see here and here's the demo video so here we have our case models, so this is the content of the of the first layer with intelligent logic with our guideline channel. So here the doctor can see that okay, he needs to check the information for the patient with the kidney disease, the logic that will be used.

A

So he can see that and then he could choose uh already from the from the patients from the available from the informations from the patients that in his database he could uh initiate the specification case uh where the logic from that is coming from the intelligent ledger will be executed.

A

So now he chooses to execute this logic for the patient one and see the execution, and that is that, basically, one of those um four different outcomes has been uh chosen based on the logic that was um that was coming from the intelligent, ledge, intelligent logic, ledger and then executed on the treatment channel.

A

So now I think I don't have time to demonstrate the demo just mentioned that we also have uh already implemented a case where the doctor may not have all the information. uh As you can see here, all the all the tasks and, for example, not all the information may be available and doctor has possibility to also manually, enter a manual entry and add uh different data. That then, would help to identify whether to identify one of those three outcomes.

A

So the next steps we have basically uh first is to extend our framework uh to connect it to the consent. Management ledger, implement the guidelines and also integrate other rules like prescription management and also a drug drug interaction, and we have also different research challenges like how to achieve systematic sharing use of data for research, including uh achieving privacy uh and security of the data, and also ensure that we can uh that we can translate from those textual guidelines that we can correctly translate them into the into the intelligent logic.

A

And this is really a very, uh very big, interesting part on how to also ensure the verification done by different parties and also um make sure that this system uh can be uh adopted in the in the clinical workflow. So we have to also work with medical doctors, and this is also one of our next steps to to help to build better interfaces and make sure that this can be easily integrated in the clinical data flow.

A

So thank you very much for your attention. Are there any questions.