With the rapid pace at which AI is increasingly being used in healthcare systems we need to evaluate how far we have come, where to improve, and prepare for the future.
by Saskia Kendall
In early July, researchers at Flinders University in Australia designed a vaccine using artificial intelligence (AI). Although AI had been used in drug discovery for many years, the sheer size of the impact was impressive and a world first. Not only was the discovery process significantly quicker, cheaper and more effective using AI, but the researchers believe the vaccine stimulates the immune system much more than traditional flu vaccines, potentially making it more effective. Should the vaccine make it to market, the potential impact on the research and development of medicines could be vast and drug discovery may well be unrecognisable within a decade.
The impact of AI is far reaching in healthcare – from early detection, diagnosis, treatment, patient experience, research, disease monitoring, treatment adherence through to customer experience. The Flinders University example is just one example of how Asia Pacific is not only at the forefront of the AI healthcare revolution but is potentially one of its biggest beneficiaries. While trying to unlock the potential, it is also important to understand the challenges, the risks and what regulation needs to be put in place to safeguard patients and health systems more widely.
AI is essentially a computer using algorithms that allows it to learn independently without following any explicit programming. It uses trial-and-error, pattern matching, rules, deep learning and cognitive computing to estimate conclusions from data without direct human input. Access to big data sets combined with the speed of AI has been a game changer and has meant that almost everything we do in our day-to-day lives is now touched by AI.
Projections and Meeting Healthcare Demands
There is a heated global race to unlock the potential of AI in the healthcare sector and Asia Pacific is at the leading edge, with China in the top three markets globally for this innovation. The region’s governments are actively trying to develop ecosystems to create the right mix of finance, highly skilled workforce and projects. The reasons are two-fold – firstly the potential market for these innovations is significant – Accenture has estimated the AI health market will be worth $6.6 billion by 2021. But secondly and most importantly, AI has the potential to make healthcare more efficient while maintaining (or improving) the quality of services. It is estimated that these developments could save health economies globally $150 billion by 2026.
This is the holy grail for health systems, particularly in Asia Pacific, where costs continue to grow and both government and individuals are finding it difficult to keep up with demand and the costs incurred. In many regional markets, populations are ageing at a rapidly and older people tend to be the heaviest users of healthcare services. With a more desk-based workforce, lifestyles are becoming more sedentary with higher calorie intake, leading to a rise in non-communicable diseases such as diabetes and cardio-vascular disease. These long-term diseases are important to manage effectively but tend to be more expensive to treat. Finally, as consumers become wealthier, they tend to spend more on healthcare and have higher expectations of health services, which drives up costs. Many parts of the region suffer from healthcare professional shortages and rural populations find it particularly difficult to access healthcare services. To meet these challenges, AI may assist with diagnosis, treatment plans and patient accessing healthcare professional support.
Developments of AI in Asia Pacific
China, a market dealing with many of these issues, wants to be the global leader in AI by 2030. A leading Chinese online healthcare company, Ping An Healthcare and Technology Company Limited, has developed mobile AI-powered health services known as Ping An Good Doctor. With over 289 million registered users, its online AI programme can provide medical diagnoses and treatment plans based on symptoms and medical history of patients, with decisions reviewed by healthcare professionals but not made by them.
Ping An Good Doctor has even launched the One-Minute Clinic, the world’s first AI-powered healthcare clinic unveiled last year just outside Shanghai. Here, patients can physically attend a small booth, be diagnosed, provided with a treatment plan, and can receive medicine from an attached vending machine which houses 100 of the most common medicines.
Japan, a market with a public health system and universal health coverage, is also struggling with a shortage in healthcare professionals and rising healthcare costs. The government is investing in AI development in the hope that the technology will enable healthcare professionals to see more patients for longer more cost effectively. The Japanese government aims to develop 10 AI-enhanced hospitals by 2022, where AI will help with time consuming administrative tasks, like patient records management and data management, along with highly skilled decisions like diagnostics and optimising treatment. AI-assisted programmes will parse MRI and other diagnostic imaging, and analyse blood tests and other information, making treatment recommendations to inform healthcare professional decision-making. AI will collect data and it is hoped that through this accumulation, healthcare professionals will be able to refine diagnostic capabilities and better identify the best treatment for each patient.
Another example is a South Korea government-backed initiative that aims to release the potential of full personalisation of healthcare. After reformatting patient data, including scans and genomic data, AI will identify patterns in how patients react to treatment and enable more effective treatment while eliminating unnecessary treatment. This has benefits for patients by reducing unnecessary treatment and potential side effects, while reducing costs.
Cybersecurity and Maintenance of Integrity of Data
Maintaining public trust in how data are being managed and used is important, particularly when express consent may not be provided for its collection. Despite much being anonymised, the data being collected are personal and sensitive by nature including information about diagnosis, treatment, medical history, prescriptions, genetic information and background information like income, age, ethnicity. Data breaches, like the 1.5 million patient records stolen in Singapore last year, can undermine public confidence in how this highly sensitive and personal information is being safeguarded and highlight the value of these data.
Safeguarding the quality of the data being used by AI is imperative. Disparities in how different people collect and enter data or take scans can change how the AI learns and could have unintentional consequences on how it interprets the data. For example, it can unintentionally create or reinforce bias against particular population groups potentially, leading to over or under treatment.
Intentional meddling, ‘adversarial attacks’, where manipulations of data can change the behaviour of AI systems is also a risk. A recent Harvard research paper found that by changing a small number of pixels in an image of a benign skin lesion, a diagnostic AI system could be tricked into identifying the lesion as malignant. They have found that simply rotating the image could also have the same effect.
Regulatory and Ethical Considerations
These technological developments in healthcare can also raise issues around regulation and ethical oversight, which have not developed at the same rate as the application of AI itself. What should the role of AI be in healthcare services? How are patients being safeguarded? Who should be the decision-maker in patient management and treatment? What patient consent is needed? What rights do patients have over their data and their treatments? What limitations should be put on the use of data? How does AI impact the healthcare workforce? These are just some of the questions that need to be answered and defined in order to maintain the trust required to continue to develop AI’s full potential.
Although data management and cyber security are being regulated in varying ways globally, AI remains largely unregulated. Governments and healthcare companies have prioritised investment in technology and systems in order to compete in a heated global race and to unlock its potential. However, less thought has been given to regulation to safeguard patients, data, systems and services, with little action. This is complicated by how the developments of AI tends to transcend borders. There is a need for a regional, if not global, conversation about what the role of regulation should be and how to harness the power of AI while safeguarding patients and populations.
This regulatory vacuum poses reputational risks to companies and health systems using AI as they may be blamed when things go awry. In response some multinational companies recognise the need to lead discussions in this space. Google, in 2018 developed nine ethical principles that will guide their AI work and a need to act “with deep responsibility, care and humility”. This has been in part due to the reputational damage already inflicted and trust already lost. It is important for companies developing and using AI programmes to define what their guiding principles, safeguards and responsibilities to mitigate against this critical issue.
Into the Future With AI
Professionals having trust in the technology is also important for systems success. A recent poll in Singapore, for example, found that one in five healthcare professionals feels their long-term job security is threatened by advancements in healthcare technologies including AI. This personal concern is unfounded as healthcare professional oversight remains vital, but this fear may still undermine its use and its potential success.
Over the next 20 years, healthcare solutions, services and systems are likely to be revolutionised by AI. However, we must mitigate (where possible) against unintended consequences and adversarial attacks. Governments, public bodies and the healthcare industry are part of this solution and debate around this will be vital to unlock the true potential of AI. [APBN]
References
- Nikkei staff writers (2018, August). Japan plans 10 ‘AI hospitals’ to ease doctor shortages. Retrieved from: https://asia.nikkei.com/Politics/Japan-plans-10-AI-hospitals-to-ease-doctor-shortages
- Energias Market Research (2018, December). Global Artificial Intelligence (AI) in Healthcare Market to Witness a CAGR of 48.7% during 2018-2024. Retrieved from: https://www.globenewswire.com/news-release/2018/12/11/1665026/0/en/Global-Artificial-Intelligence-AI-in-Healthcare-Market-to-Witness-a-CAGR-of-48-7-during-2018-2024.shtml
- Accenture Analysis (2017). Artificial Intelligence: Healthcare’s New Nervous System. Retrieved from: https://www.accenture.com/sg-en/insight-artificial-intelligence-healthcare
- Rei Kurohi (2019, June). Singapore ranked No. 3 in survey on use of AI in healthcare. Retrieved from: https://www.straitstimes.com/singapore/health/singapore-ranked-no-3-in-survey-on-use-of-ai-in-healthcare
- MIT Technology Review Insights (2018, December) Asia’s AI agenda: The ecosystem. Retrieved from: https://www.technologyreview.com/s/612654/asias-ai-agenda-the-ecosystem/
- Samuel G. Finlayson et al. (2019, March) Adversarial attacks on medical AI: A health policy challenge Retrieved from: https://cyber.harvard.edu/story/2019-03/adversarial-attacks-medical-ai-health-policy-challenge
- PR NEWSWIRE ASIA (2019, April) Ping An Good Doctor Launches “One-Minute Clinic” at Shanghai Jiao Tong University. Retrieved from: https://www.asiaone.com/business/ping-an-good-doctor-launches-oneminute-clinic-at-shanghai-jiao-tong-university
About the Author
Saskia Kendall, Head of Health, Asia Pacific, MHP Communications
Saskia is a policy and public affairs specialist. She oversees the health practice in Asia Pacific at MHP Communications, working for clients including pharmaceutical companies and health insurers.