DJ Fang, chief operating officer and co-founder of AI-powered regulatory consultancy Pure Global, says softer regulations in China and Southeast Asia can mean readier access to patient data, which is in growing demand as medtechs and their partners work to train AI algorithms.

Compared with the US and Europe where regulations around data privacy and informed consent are more stringent, “what you find in China, Southeast Asia, and really the whole of Asia Pacific with the exception of maybe Japan, there are certain hot spots where data, especially patient data, are more easily accessible just because of the nature of consenting in the hospital systems. Because once patients go in the hospital, sometimes they just consent automatically that their data will be used for research to get those treatments,” Fang said in a 21 October interview.

There are still limits that tend to be imposed on the data’s use and transference. “The data cannot leave the country, that’s the biggest thing,” he said.

“Essentially there is the Western chain of thought versus the Asian chain of thought in terms of data availability and supply. --DJ Fang

He noted that under the EU’s General Data Protection Regulation, for example, arrangements can be made for data sharing provided the data are sufficiently anonymized. But stakeholders have pointed to practical challenges of anonymizing data and gray areas in the regulation that companies must navigate carefully.

Similarly, “in the US, data privacy is a big thing.” Some medtechs are finding success with decentralized federated learning frameworks wherein multiple data providers train machine learning models locally within their own systems without sending data to a central server, thereby circumventing potential privacy and security issues.

“In parts of China, there will be some local regulation to support metech startups,” allowing them to “tap into networks and data that are only used by those companies in that region to do research. That’s very unique because they want to foster a lot of research in certain areas, whether to generate economic activities or employment. … But that consent process is sort of implied by the hospitals,” Fang said.

“So essentially there is the Western chain of thought versus the Asian chain of thought in terms of data availability and supply. The ease of getting data will be relatively easier in the Asia Pacific region versus the Western side of things just because of the maturity of the regulation,” he added.

Prior to co-founding Pure Global in 2020 at the height of the COVID-19 pandemic, Fang’s background was in big data and cybersecurity and, before that, finance. As a trader he gained exposure to AI models for predicting changes in financial markets, as well as the data siloes and other access challenges that stand in the way of fulfilling the ultimate promise of Big Data and AI.

AI-Driven Market Access

With more than 20 offices and laboratory locations around the world, Pure Global leverages AI and data to provide medical device and IVD clients with smart, efficient regulatory compliance solutions, “unlocking global markets,” according to its website.

Its global database – searchable using AI and data-powered research tools – comprises more than 100 countries’ regulations, 600,000 distributors, 5 million registered products, and 1 million clinical data records, among other information.

The firm is actively helping clients to navigate market access in 35 countries or regions around the world, not only with respect to core medtech regulatory compliance but also the growing number of data, AI and other regulations that medtechs are subject to as well, Fang said.

“We really manage this through our digitized workflows, essentially the software we built out to handle this. Today, we leverage AI to support a lot of the regulatory documentation needs and monitor regulatory updates that come down,” he said.

For a number of countries, such as Brazil, Pure Global has fully digitized regulatory submission workflows to streamline the process for clients, automate the completion of required forms, and translate to local language, Portuguese in the case of Brazil.

Fang counted Pure Global’s AI-based translation services among the firm’s draws. “Every country will most likely have their language required for submissions. This is very exciting because the quality of the translation accuracy is becoming very high. Normally in the past when you started with Google Translate and got 70% to 80% accuracy, that was pretty good. Now with AI, you can really get above the 90% level,” he said.

Pure Global has developed its own AI translation platform that not only offers higher-quality translation, but also improved capabilities for uploading documents without compromising format, “even with PDFs,” Fang said.

“So this is why we‘re seeing this trend of companies saying, ‘I would like to just use your translation platform’ – especially these larger companies, they definitely have translation work that needs to be done. Of course they’ll have an expert to do the final [quality control], but at least the initial grunt work can be largely handled by AI. So there’s a lot of savings there.”

Fang noted that Pure Global’s regulatory consultants still spend significant time reviewing documents themselves, but AI screening has helped to focus their activities. All told, clients have reported shortened go-to-market timelines. “Where normally it takes, let’s say, a few months, maybe we’ll get it done in, you know, one month,” he said.

AI Diagnostics, Consolidation

Fang expects AI to continue to make pronounced impacts in the diagnostics space.

He pointed to Roche’s announcement in September that it has expanded its digital pathology open environment by adding more than 20 advanced AI algorithms from eight new collaborators, including Deep Bio, DiaDeep, Mindpeak and Owkin. The algorithms are designed to provide enhanced diagnostic capabilities across a range of cancers.

“Roche’s navify Digital Pathology enterprise software, an application for the pathologist’s workflow … creat[es] easy access to third-party innovation. These AI tools are designed to enhance pathology insights, helping benefit cancer patients through precision medicine and enabling targeted treatments,” the firm says.

In Fang’s view, major medtech companies likely will continue to partner with start-ups to bring mature algorithms onto their own platforms rather than developing proprietary AI/ML models in-house. Such arrangements benefit the AI innovator class as well.

In radiology, for example, “a small startup could have the software to analyze images. It’s hard to say, ‘You have a new device that needs to break into all these hospitals and clinics.’ That’s just not practical. But [medtech’s] big guys are playing in the market already, so what is more practical is for these bigger companies to integrate the start-ups’ technology into their platforms,” he said.

‍