AMNC23: This is what needs to happen for AI to improve healthcare

"In order for efficacious therapeutics to reach patients globally, we have to find ways to make them more efficiently."

So said Ena Cratsenburg, Chief Business Officer of Ginkgo Bioworks, at the World Economic Forum's Annual Meeting of the New Champions, in a session on Healthcare Systems of Tomorrow.

The session also featured Christine Zhou, Senior Vice-President; President, Region China, Novo Nordisk, Lu Yimin, President, China General Technology (Group) Holding Co, Ltd and Ren Minghui, Professor, School of Public Health, Peking University.

It came as the Forum published a report on Scaling Smart Solutions with AI in Health: Unlocking Impact on High-Potential Use Cases.

The use cases identified for private-public acceleration include diagnosis, infectious disease intelligence and clinical trial optimization. Other use cases for further exploration include identifying new drugs and triaging patients, with actions recommended to scale AI in healthcare.

Countries across the world are working on AI regulation to ensure we avoid the pitfalls of the technology, such as bias, but also working out how to implement it in useful ways.

The UK, for example, has announced a $26 million AI Diagnostic Fund that will include the use of AI tools to analyze chest X-rays to diagnose lung cancer sooner, the leading cause of cancer death in the UK.

AI is an "important enabling technology" to help us scale up healthcare in future, said Ena Cratsenburg in the Forum's Healthcare Systems of Tomorrow session.

Here are some of the key quotes.

AI is only as good as the data you have to train the models, added Ena Cratsenberg and right now we are "scratching the surface" of what biology can do.

"One of the things that is going to help engineer biology in a way that can bring forward useful products is finding a way to do a lot of experiments in a very efficient way.

"Automation, robotics can help us do millions of experiments. By leveraging the latest advancements in those technologies, we can do a lot of experiments and generate a lot of data. For any good AI model, you need to have good-quality data to train the model.

"Imitating the data in a way that allows you to gain the insights for you to design something rationally, and understand the result of changes you make, is incredibly important. By having a standard way to curate the data and analyze the data in a useful way, that is when AI is connected with empirical experiments ... will help develop insights that are useful."

Ren Minghui said governments have to take a balanced approach to make sure risks can be mitigated and managed well before such innovation is recommended for wider use.

"They can encourage sharing information and technological expertise across-the-board with different entities because it is very important, and can reduce unnecessary overlapping, speeding up new ideas and innovation in health and medicine development."

He said governments can also help support the financing of innovation through taxation policy, to incentivize companies to invest in research and development.

Novo Nordisk's purpose is driving change to defeat chronic diseases like diabetes and obesity, said Christine Zhou.

"We increased investment in research and development and we are applying new technologies like artificial intelligence to speed up the innovation processes and fostering a partnership to make sure we tap into external innovation capabilities.

"But innovation from pharma alone is not enough to address ever-increasing healthcare challenges. We see that collaboration and partnership among stakeholders such as societies, academia, and industry peers, as instrumental for us to work together to create an ecosystem to drive change, and to increase access to the patients in need."

"There are some really exciting AI models out there," said Ena Cratsenberg, "and while I think it is very possible for these models to design a protein that may improve service function and have the characteristics that you want to be a good drug, the fact of understanding why that protein works still doesn't exist.

"So these AI models are one of many different tools we need in our toolbox in order for us to really understand how and why a drug works and we can use that insight and understanding to develop other drugs.

"I don't think AI will replace fundamental research, it is a part of the solution that we have. It is a tool and the more we understand the scientific insights, the better we can train these models and it is an ongoing leap from there."

Christine Zhou said the AI technology is bringing a lot of impact to healthcare development and its utilization has been increasing across different sectors in the value chain, including drug discovery, clinical trials, manufacturing, quality control, and even manufacturing.

But it's the "convergence between technology, drug development and biology that will lead to better drugs being developed faster and more accurately".

Christine Zhou also warned that if unregulated, AI can have negative impacts.

"In my view, the difference between AI and the human mind is they all know how to do things but only the human mind knows why. Therefore, it is difficult for AI to replace the human mind in certain fields. If AI is not regulated, it could potentially cross red lines and have a negative impact. If you look at the hypothetical examples in healthcare fields, unverified medical information used by generative AI could mislead the public.

"We need to make sure we create a guardrail for AI development and education."

The Chinese regulators have required filing of AI algorithms and some countries also have clear requirements for a disclaimer of the contents generated by AI, she added. Pharma companies must make sure self-regulation behaviour is embedded with ethical guardrails in AI utilization.

"AI allows us to have more ability to see what is happening in a more democratized way," said Ena Cratsenberg.

"Having the ability to access data and see how it impacts different aspects of the delivery of the medicines is going to help address and flag problems that need to be solved, but there is a lot that we need to do."

Ren Minghui said AI was used for diagnosis in the COVID-19 response: "It was cost-effective and the technology can be used widely for lung cancer and lung diseases."

But besides innovation, it can also help with service delivery for ageing populations.

"We must move our health systems from hospital to home-based services. AI can help with these changes and trends. It is good, but we need regulation to ensure we can benefit and mitigate the risks to human beings."

Lu Yimin added: "AI is making the health system more equal, fairer. We are using it to help local hospitals improve diagnostic capacity. With a digital twin, we are able to offer innovation companies, and pharma companies more insights to better serve our people."

Watch the whole session here: