The article is also available in French and Spanish.
Digital health: friend or foe?
The unprecedented impact of the COVID-19 pandemic on even the simplest routines in our lives, like going to the doctor, has pushed digitalisation and technological innovation to the top of the agenda for global policymakers. The potential enhancement that AI can bring to healthcare is unquestionably significant. However, without the correct framework, it can also lead to the incorrect use of a very powerful technology. Sensitive data protection, fragmented regulatory frameworks, and the still-too-slow-but-needed cultural transformation are just a few of the challenges.1
While progress in the development of AI as an enabler rather than a hindrance has primarily focused on developed economies up to now, applications in the health sectors of developing countries are proving to be just as important, and perhaps more urgent.
Best practices from AI experts: how to fast-forward innovation in healthcare.
1. From reactiveness to proactiveness: bringing technological innovation into people’s homes
2. Speeding up cultural transformation: building trust locally
3. Implementing AI solutions in healthcare to help doctors make better decisions
1. From reactiveness to proactiveness: bringing technological innovation into people’s homes
The application of AI solutions to traditional medicine could lead to an important shift from reactive to proactive healthcare. Being able to collect medical data and forecast an illness before it takes hold is one of its many advantages. The AI4Leprosy project did exactly that: globally, there are between 2 and 3 million people living with leprosy (despite it having been officially eradicated), with almost 200,000 new cases diagnosed every year. Microsoft, the Novaris Foundation and the Oswaldo Cruz Foundation joined forces and developed an AI-powered diagnostic tool capable of detecting leprosy via images: “The main goal of the project was to increase accessibility to an accurate method of classifying leprosy to assist clinicians, especially in remote communities. The data was used to train an open-source AI powered model able to assess the probability of people having leprosy. The AI models implemented had a satisfactory accuracy of 96.4%.”2 How were the images collected? With a very simple smartphone application.
“In Africa, many women who look after their children have to walk one or two days to go to the doctor. Take cervical cancer as an example: because of the stigma, these women will not seek medical assistance. But if you provide the right tools to the local community workers who work in the village or in the school where the mothers pick up their children, basic first examinations can easily be performed. In Africa, we empowered a very broad network of telemedicine so the community worker can connect with a specialist that is two or three days away and can provide a diagnosis. You take a photo with your phone and you get your diagnosis. And that triggers an action, i.e. you go to the doctor.” Elena Bonfiglioli, General Manager Healthcare, Microsoft
2. Speeding up cultural transformation: building trust locally
According to a recent study conducted by the European Parliament on the applications, risks and ethical and social impact of artificial intelligence on healthcare, the increasingly widespread use of AI technologies is still frowned upon by many. The lack of transparency and understanding as to how artificial intelligence really works, privacy and confidentiality concerns, and a fragmented regulatory framework are just some of the controversial issues impacting trust in AI. Although it has been proven that AI enables the processing and analysis of huge amounts of data in a very limited time, we need to build trust and establish robust governance.
“Data is collected and administered as an essential resource through a governance model that creates sufficient trust within the society, with the patients, the doctors, and the regulators so that we can truly rely on this data to make decisions. And when it comes to developing countries especially, you need to work with the local stakeholders and involve them in decision-making. You need to hear their needs and then you make sure you measure the outcomes that matter to them.” Meni Styliadou, Vice President Health Data Partnerships, Data Science Institute – Takeda
Trust, then, is one of the key issues in guaranteeing that local communities and administrations, stakeholders, regulators and doctors are part of a process that is sustainable and rooted, and not dependent on grants that may or may not materialise. In order for AI to be effective, data is needed. And collecting data can only happen if we make sure that the process is safe and understood by all those involved, particularly the patients:
“Often when we start an AI project, be it in healthcare, environment, climate, or agriculture, the need for data arises, and it is often challenging to find sufficient data. At times, data exists but, at the end, it needs to be harmonised, processed, cleaned, transformed and ultimately shared. For us, the difficulty lies in creating a culture that values collecting, governing, and sharing data.” Stefano Sedola, Partner, StratejAI
3. Implementing AI solutions in healthcare to help doctors make better decisions
The World Health Organization estimates that an additional 10 million healthcare workers will be needed by 2030, primarily in low- and lower-middle-income countries, to guarantee the proper functioning of health systems worldwide. In addition to investing in the training of more healthcare professionals globally, they will also need to be provided with the tools that will enable them to take care of a greater number of patients. This is where AI solutions come into play. “Doctors will have an additional tool to help them make better, evidence-based decisions,” explains Meni Styliadou. “At the same time, we will have the opportunity to work with more healthcare professionals, to empower and enable community workers and nurses, particularly in the developing world,” she concludes.
Overall, AI solutions have the potential to reduce the so-called administrative burden doctors face every day and help them provide better care for their patients. Technology will never replace medical knowledge, but it will have an impact on overall productivity.
Final thoughts
While progress has been made in recent years to speed up digitalisation efforts in all sectors of society, healthcare still lags behind due to a variety of challenges, ranging from data access barriers to safe data sharing, all the way up to a generalised lack of trust. The latter is probably one of the most important elements in the equation: advancing AI capabilities depends on the quality and quantity of data we are able to collect. Without data and algorithms, machine learning cannot occur.
In addition, the success of data collection and the consequent creation of models and algorithms that can help save lives globally depends on the availability of resources. As Miriam Stankovich, Senior Digital Specialist at the Center for Digital Acceleration puts it, “the AI that can identify tuberculosis from chest X-rays in India could save time, money, and lives in South Africa, particularly in rural areas where there are no specialists to examine such images. However, to obtain images in the first place, communities need X-ray machines and people to operate them. Failure to provide these resources will mean that AI tools will simply serve those already living near better clinics and thus exacerbate the already existing digital divide.”
1 Healthcare Data Innovation Council – White paper
2 https://healthdatainnovation.eu/ai4leprosy/
Recommend & share this article if you think it will be helpful to your peers
Click on the play button below to watch our video
Additional information
• Panel on AI impact on society: futurism & practice. AI and data innovation for the future of healthcare, European Development Days, 2022
• A European approach to artificial intelligence | Shaping Europe’s digital future (europa.eu), European Commission, 2023
• UNESCO adopts first global standard on the ethics of artificial intelligence, UNESCO, 2022
• The OECD Artificial Intelligence (AI) Principles, OECD
• Policy on artificial intelligence, European Commission, 2022
• Artificial intelligence in healthcare: Applications, risks, and ethical and societal impacts, European Parliament Think Tank, 2022
• The Fourth Industrial Revolution, World Economic Forum, 2023
Log in with your EU Login account to post or comment on the platform.