By design, blockchain technology provides AI systems with a more quantitative and qualitative engine to power the use of actionable information: smart, trustworthy automation from end-to-end. Combined, both technologies can bring lots of value to the AI-driven healthcare world. Following the previous Parkinson’s case study, we should also consider how AI and blockchain can come together to improve clinical trials, as they enhance the quality of interactions between numerous stakeholders including patients, medical professionals, pharmaceutical companies and regulators.
Such new models can enhance the recruitment of patients. According to Deep 6 AI, a company trying to improve the clinical trial enrolment process by transforming the way researchers identify suitable patients, “86% of clinical trials fail to recruit sufficient patients”. Deep 6 AI has designed algorithms to quickly identify patients and match them to complex trial criteria, thereby reducing the risk of failure of a trial due to unsuitable patients. This said, lots of barriers remain to ensure patient data will be used in the right way and we can envision how valuable decentralised models could be in securing, anonymising and processing patient data usage. It’s time to consider new models.
The AI & blockchain convergence requires new models and human-centric services
1/ Propose a new “consent as a service” model
The organiser of a clinical trial should always obtain the participants’ informed consent to the use of their personal data, particularly given its sensitive nature. This should involve the organiser disclosing to the participants all pertinent information regarding the trial, including the protocol and the potential risks so that the participant is able to make a fully informed decision as to whether to allow their personal health data to be used. Collecting and recording patients’ consent on the blockchain can guarantee reliability and traceability of, and facilitate access to, data. In this regard, researchers at the University of California recently succeeded in creating a proof-of-concept trial method for ensuring the integrity of clinical trial data using blockchain, whereby all data is recorded using an algorithm (SHA256 or Secure Hash Algorithm, like a signature for a text or a data file as an almost-unique, fixed-size 256-bit hash).
2/ Use AI to define a new health value economic model
To foster health data sharing, it is essential that people become aware of the value generated, as it will benefit many sectors of the healthcare industry as well as many various other patients. For instance, it will allow the R&D and innovation sectors to delve into new research. On the other hand, as various stakeholders will benefit from this, citizens need to be empowered and motivated: we must reinvent healthcare economics and reward systems. To unlock the data economy’s full potential, it is vital that we define a new health value economic model, a financial system with sustainable and inclusive foundations, enabling the auto-regulation of this new health data economy.
3/ Understand the shift in data ownership to define new legal models
We must define how data should be regulated, as health data sharing is a complex process. It involves many stakeholders, making it essential to create new data-sharing protocols. First and foremost, we must ensure patient safety, as the use of health data could give rise to potential legal issues. An essential legal matter arises when patients want to value their health data. To give patients control over their data and reward them for sharing it, we should take into consideration new collaborative models for data sharing. Thanks to their core processes, AI and blockchain can allow the data economy to increase its impact throughout the world, but we need to ensure new legal models would be capable of fostering this collaboration.
4/ Ensure data quality in regulating data provenance
Could clinical trials involving AI and blockchain still face the same difficulties, as traditional data recording means protecting data privacy? Getting quality data for training AI models is complex, but working with large quantities of data in the healthcare space is the real challenge, due to the privacy issues associated with personal health information. Solutions currently being considered to protect data privacy are the pseudonymisation of patient data, restricting access to the blockchain, and recording sensitive data off-chain. De-identification of electronic health records via AI algorithms provides the opportunity to use such data for research without it affecting patient privacy. That said, data provenance has more importance than ever, making new tracking models essential.
Claire Douangmala – MALTEM
Pierre Robinet – OGILVY CONSULTING
Lucas Nicolet-Serra – SIMMONS & SIMMONS LLP
Daryl Arnold – OCEAN PROTOCOLE