Ever since IBM decided to use its Watson supercomputer to help detect cancers, a lot of ink has been spilled regarding the potential of Artificial Intelligence (AI) in the health sector. The model currently being tested by IBM Watson in a number of hospitals demonstrates the opportunities for using image recognition to help radiologists detect anomalies and potential illnesses from the results of an MRI scan. In an article published by the American College of Radiology, researchers at the University of Texas at Austin predict that AI will soon be playing an integral part in radiologists' daily lives, making their work more efficient, accurate and satisfying.

AI ASSISTING RADIOLOGISTS

radiology

Artificial Intelligence is also able to comb through medical records and patients' genomes looking for information likely to help make a diagnosis or decide on a given course of treatment. Medical research could also benefit hugely from input from AI. By going through the latest scientific papers, AI will be able to make the connection between different research projects and spot the most promising discoveries, as well as automating the peer review process. AI could also make it easier to discover new medicines and, by rapidly ensuring that they conform to health regulations, help speed up their time to market. Add to these use cases virtual assistants which keep a remote eye on patients and the use of robots for re-education purposes, and it's easy to see why this technology is arousing a great deal of enthusiasm. 

Most of the recent progress made in AI is in the area of deep learning, a branch of the discipline that enables machines to become better at what they do through experience. Based on deep neural networks, deep learning is inspired by the way the human brain works. Deep learning is behind AlphaGo, the all-conquering 'Go' champion developed by Google DeepMind, the software that runs self-driving cars and also the image recognition technology which enables Watson to analyse an MRI scan. However, while the output of deep learning is astonishing, it needs huge quantities of data in order to fulfill its potential. With the Internet and ubiquitous connected objects there is no shortage of health data but it is often fragmented and difficult to access and use.

new technologies  to safeguard your health

hopital
Shutterstock

Data fragmentation putting the brakes on innovation

The issue is a particularly thorny one in the United States, where the health system is complex and involves a number of different stakeholders. The 1996 Health Insurance Portability and Accountability Act (HIPAA) gives patients the right to access their health data and to control who has access to it, but in practice this access is often rather limited, as the Reuters agency demonstrated in an article published in 2016. Patients therefore largely remain pretty much unaware of the content and value of their medical data. Moreover, patient data is stored by a patient's healthcare provider on a private server, which means they cannot be transferred or used by other players – for example to train an AI program to recognize the symptoms of an illness.

HEALTH DATA

data
We should also point out that health data is not limited to the information collected by doctors on their patients. In our era of new technologies, we also need to think about other data sources – from medical imaging, tests analyzed in laboratories, including blood samples, genetic tests, plus all the varied data generated by wearable electronic devices. This data is stored in different silos– at doctors' offices, hospitals, insurance companies, laboratories, health tech startups, etc. – with very little interaction between any of them. On the other hand, given the private and sensitive nature of the information, it is of course not an option to introduce a policy of total openness and free circulation of such data.

Consequently, startups working with Artificial Intelligence in the healthcare field have to invest a large part of their budget in buying datasets on which to train their algorithms. With a limited budget they can often only acquire a small amount of data, which in turn limits the scope of any solutions they come up with. In other words, the data exists, just as the technology enabling the data to be used exists, but there is no common database which would allow the data to be passed among the various players while still maintaining confidentiality. Until recently there was no technology capable of relaying the information openly, collaboratively and securely. With the Blockchain however, things are starting to change.

using the blockchain to secure health data

données

Shutterstock

Blockchain: the potential to create a huge medical database

A DATA POWERHOUSE

data blockchain
The blockchain, which creates a decentralized digital ledger system, offers anew way of storing information other than in a traditional centralized database. Each participant has a copy on his/her computer, which s/he can consult and modify. The blockchain approach is therefore also highly transparent. But the question is: if anyone and everyone can access the database, how can mischievous people be prevented from altering information to suit their own interests? Taking the example of Bitcoin, how can an ill-intentioned person be stopped from artificially increasing the number of Bitcoins s/he holds?

In order to deal with this issue, each transaction needs to be checked and approved by participants on the network, known as 'miners', who use complex cryptographic techniques. Security is ensured jointly by the members of the system themselves rather than by a central unit. Once the information has been entered, all participants can see it. This mechanism is very simple and devilishly efficient. Marc Andreessen, a long-time investor in Silicon Valley, argues that the Blockchain is the most important invention since the Internet.

sandy
Regard d'expert

Sandy Hathaway

Entrepreneur

Various providers, payers and pharmacies all record this information on the same ledger for a given patient, and they can access the data put there by each other 

This twin approach, whereby the blockchain is both open and secure, makes it the ideal technology for resolving the dilemma which is currently putting the brakes on AI advances in the healthcare sector. Taking part in a forum on the online publishing platform Medium, entrepreneur and investor Sandy Hathaway explained: "Blockchain is a digital ledger that records and shares transactions and interactions in chronological order, providing security and interoperability for healthcare providers and their patients. In healthcare, for example, each patient visit, diagnosis, prescribed treatment, outcome and other key data that goes in the EHR [Electronic Health Record] are considered the transactions. Various providers, payers and pharmacies all record this information on the same ledger for a given patient, and they can access the data put there by each other." Ms Hathaway argued that "blockchains for health data can and should be created as a permissioned ecosystem of trusted collaborators with additional controls that protect personally identifiable information while allowing anonymized sharing of personal health information."

Blockchain: a real opportunity  for the health sector

block

Shutterstock

Enabling patients to monitor their data

doing away with intermediaries and regaining control of the data

data mobile

Such a system would first of all lead to information silos being broken down so as to bring the various datasets into one place, creating a unified register which would in turn allow promising AI-based applications to be developed.Secondly, the transparent and unalterable nature of blockchain content would in addition improve the reliability of data in the health sector by making it easier to correct errors and fostering synergies among the various players.This was the view expressed in an IBM report published in 2017, which looked into the possibility of using the blockchain to store health sector data.

"Blockchains could replace the intermediaries that once existed to secure this data, perform these tasks. Smaller organizations could join ecosystems to take on larger competitors. Private sector participants could gain access to and create new sources of data, whether that’s wellness data streaming from personal devices or information collected by home caregivers", says the report. And as the data is absolutely essential for training Artificial Intelligence programs, it would be desirable to ensure that the data is as reliable as possible.

Using blockchain technology would also help to obtain a better overall view of the value chain. It would be possible, for instance, to trace the timeline of each medicine, from the moment when it was first developed to its arrival in the hands of patients. This would inter alia make it easier to combat the counterfeiting of medicines.

gunjan
Regard d'expert

Gunjan Bhardwaj

Co-founder of Innoplexus

Blockchain and AI could enable a structural shift where all parties share data in a decentralized fashion, wherein the system could still collectively use the data to make smart decisions 

A recent report published in the bio-medical research journal Oncotarget focused on the potential for the blockchain and AI technologies to converge in the field of healthcare. In the report, researchers describe how a health data market could be set up, where patients – who have total control over their data and the access rights – could make their data available to developers, pharmaceutical laboratories and research institutes looking to use it to train their AI software for medical purposes and would in turn be remunerated. This looks like a win-win-win model for all parties concerned.

"Blockchain and AI could enable a structural shift where all parties share data in a decentralized fashion, wherein the system could still collectively use the data to make smart decisions", Gunjan Bhardwaj, co-founder of the connected health startup Innoplexus, told Forbes magazine, explaining: "This could overthrow the legacy hurdles of healthcare – data lying in different places, strong regulations restricting the sharing and analysis of that data, and weak incentives for sharing research and training data."

Drawing up a regulatory framework

This idea is now starting to take shape in the world of e-health. In January2017, IBM and the US Food and Drugs Administration, a federal agency under the United States Department of Health and Human Services, teamed up to test secure exchange of medical data via the blockchain. Last October, IBM set up a new partnership, this time with the Centers for Disease Control andPrevention (CDC), the leading national public health institute in the UnitedStates, so as to continue the drive to explore the potential of the blockchain in the health sector.

the data-friendly blockchain

pills

In the United States, Google's AI subsidiary Deepmind, which developed AlphaGo and recently formed a partnership with the National Health Service (NHS), the public health service in the UK, to work on connected health, has also announced plans to build a blockchain as part of the project. Meanwhile, Californian startup Gem, which specializes in blockchain applications, has linked up with Philips to develop applications in the health sector. Also worth mentioning is the recently-formed partnership between startups Longenesis – which specializes in blockchain technology – and Neuromation, an AI specialist, with the aim of creating a blockchain-hosted global market in health data, following the model described in the Oncotarget report. Russian startup Skychain, which is currently on its first round of fund-raising, has also set out to make health data more accessible to AI developers via the blockchain.

If this kind of system is to be implemented, the health industry will need to rapidly draw up and adopt a regulatory framework covering the new practices so as to ensure that this approach will benefit public health while also maintaining data privacy. Speaking on SAS's Health Analytics channel, health industry analyst Kamaljit Behera underlined that "the healthcare industry needs to establish blockchain consortia to facilitate partnerships and create standards for future implementation on a large scale across healthcare use cases." The involvement of the FDA and CDC with IBM's venture looks highly promising, and meanwhile the Pistoia Alliance, a not-for-profit organization which promotes scientific research, is also keeping a close eye on the potential of blockchain technology in the health sector.

By Guillaume Renouard