Artificial intelligence revolution: new possibilities through use in health
The development of AI-based technologies has created the notion that AI can solve any problem. However, there are many challenges and limitations associated with using AI10/05/2023
Along with its use in tasks such as facial recognition, Global Positioning System (GPS) navigation apps, and consumer behavior analysis, artificial intelligence (AI) has been increasingly used in the field of health. The use of AI in health promotion includes technological innovations, methods, and devices used in all components of patient care, from the treatment of illnesses to improving individual or community rehabilitation. According to the World Health Organization (WHO), AI shows great potential in improving the delivery of health services worldwide and can be used to improve the speed and accuracy of diagnoses and disease screening, assist in clinical care, and strengthen health research and drug development. According to this organization, AI can also support various public health actions such as disease surveillance and system management.
According to Rosália Morais Torres, a physician and professor at the Federal University of Minas Gerais (UFMG) and the coordinator of the Center for Health Technology (CETES) at the Faculty of Medicine in UFMG, AI can play a crucial role in dealing with tropical diseases, which prevail in regions with limited resources and infrastructure. She suggested detecting tropical disease outbreaks in advance using AI algorithms capable of analyzing large sets of patient data by associating them with environmental factors. She stated, “This can help health authorities to take measures to prevent the spread of disease.”
According to doctors, AI can diagnose tropical diseases such as dengue and malaria more quickly and accurately, supporting health professionals in making timely and effective treatment-related decisions. Dr Torres highlighted “AI can be of great value in epidemiological studies, by helping to identify risk factors associated with tropical diseases, given its ability to analyze large data sets obtained from patient records and environmental and geographic data.” She believed that this epidemiological information could be used to develop effective prevention strategies and guide public health policies.
Another relevant aspect of AI highlighted by the CETES coordinator was its potential use in streamlining processes for discovering new drugs to treat tropical diseases, considering its ability to analyze large amounts of data and quickly select potential drug candidates. She noted, “Interestingly, we also discuss the role that AI can play in the development of vaccines for neglected tropical diseases, identifying possible antigens, and predicting their effectiveness. In short, AI has the potential to transform the diagnosis, treatment, and prevention of tropical diseases, helping to improve public health outcomes in regions where these diseases are prevalent.”
AI applications in health promotion and ethical challenges
AI can empower people to be more independent in their self-care, enabling resource-poor countries and rural communities, where patients often have restricted access to healthcare professionals or medical professionals, to close existing gaps in access to healthcare services themselves. New AI-based technologies hold great promise for improving diagnosis, treatment, research, and drug development, as well as supporting governments performing public health functions, including providing responses to surveillances and outbreaks. Importantly, ethical practices and the well-being of people should be prioritized when using and deploying these technologies.
Dr. Torres recognized that while AI had the potential to transform healthcare, there were challenges that needed to be carefully considered and addressed to ensure its ethical and responsible use. She mentioned that AI needed to be trained and fed with a large amount of data and emphasized that algorithms could be biased if trained on data that were not representative of a given population or obtained without transparency or prejudice, potentially leading to inaccuracies in relation to diagnosis and medical treatment. She stated, “An extremely relevant aspect is the issue of data privacy and security, as AI algorithms require access to large amounts of patient information, and it is necessary to develop policies and practices to protect such data.” She highlighted the pressing concerns about legal liability in case of any issues, considering the lack of clarity regarding the adoption of responsibility for mistakes made by the AI algorithm, especially if there was no human oversight. Then, she concluded, “Finally, we must remember that the development and implementation of AI technology is costly and could lead to an ever-increasing gap in the quality of healthcare between developed and developing countries.”
Researchers from Stanford University in the United States published the new edition of the report “Artificial Intelligence Index Report 2023” in April, which contains an annual study of trends in AI with interesting findings. For example, the number of incidents related to AI misuse has rapidly increased. According to the Algorithmic and Automation Incidents and Controversies (AAIC) database, which tracks incidents related to the ethical misuse of AI, the number of AI-related incidents and controversies has increased 26-fold since 2012, drawing attention to raising awareness regarding its potential misuse. The AI, Algorithmic, and Automation Incidents and Controversies Repository is a public, open, and independent dataset that documents recent incidents and controversies related to AI, algorithms, and automation.
Opinion of WHO regarding the use of AI in healthcare
The benefits of AI in health do not negate the high complexity of the AI technologies that employ personal data from patients and public health to generate intelligent solutions. Therefore, it is critical that institutions commit to the effective use of AI at a global level. To limit the risks and maximize the opportunities intrinsic to the use of AI in health and considering the high potential of AI technologies in medicine, the WHO has released guidelines regarding their ethical use. These principles should be followed in AI use to ensure that the full potential of AI in healthcare and public health is exploited for the benefit of all. The following six points have been highlighted.
- The control of healthcare systems and decision-making needs to be performed by humans and not entirely by AI.
- Developers responsible for deployed technologies need to monitor and ensure the full operation of all tools in addition to meeting all security standards.
- Developers are also responsible for publishing data and information regarding the development and handling of products, ensuring transparency at all stages.
- Only trained and qualified professionals should use health systems that employ AI tools.
- To promote diversity and avoid the exploitation of “addictive” algorithms, AI technologies should be trained with a database containing data that considers different nationalities, genders, and races.
- Based on their performance, AI tools should be continuously evaluated so that any type of problem can be quickly identified and solved.
Health promotion and medical community trust
In the opinion of Dr. Torres, AI still needs to be viewed as a safe technology to gain the trust of the medical community, considering that the quality of the information that is fed into medical algorithms defines the quality and safety of the results and the efficiency of the algorithms, which needs to be validated. She points out, “On the other hand, the medical community needs to get to know AI better and learn to use its potential within medicine.” The UFMG CETES coordinator highlighted the importance of understanding the potential of AI in changing the way medical work is performed and noted that it was not necessarily a risk to medical work. She stated, “AI technology can automate certain tasks leading to improved efficiency and accuracy in diagnosis and treatment. For example, it can analyze X-ray or magnetic resonance images and detect patterns or anomalies that are eventually more difficult for humans to evaluate. It can also be used to analyze large amounts of patient data and identify trends and patterns relevant to medical diagnosis or treatment.”
However, doctors are categorical in stating that although AI algorithms can automate some tasks currently performed by medical professionals, they are unlikely to replace human work in healthcare, since medical professionals play a key role in interpreting the results generated by these algorithms. Professionals involved in patient care can be supported by the unique skills and knowledge provided by AI. Dr. Torres noted, “We can say that it has the potential to change the nature of medical work, shifting the focus of professionals to higher-level tasks, such as decision-making and care coordination,” while emphasizing that AI could lead to new opportunities for professionals and physicians to utilize their experience and expertise in new and innovative ways. Then, she exclaimed, “We must always remember that medical activity is, in essence, the human taking care of the human and, in that, he can never be replaced.”
Regarding the effectiveness of AI in providing medical prescriptions that could lead to more appropriate treatments, Dr. Torres clarified that there was currently no concrete evidence that an AI-guided prescription was more effective than a human prescription. However, she noted that in exceptional cases in which AI algorithms had been fed with quality data or machine learning (ML) had been well-developed, AI could become a great adjuvant in medical practice, increasing the effectiveness of diagnoses and supporting improved choices in treatment. She concluded, “It must be fed with data and the better the quality and quantity of this data, the better the result obtained through AI algorithms.”
First global report on AI in healthcare
In 2021, the WHO published the first global report on AI in health, listing six guiding principles for its design and use. The document, titled “Ethics and governance of artificial intelligence for health”, was the result of two years of consultations carried out by a panel of international experts from different areas, such as law and digital technology, as indicated by the organization. The document identified the biggest challenges and potential ethical dilemmas in managing AI in healthcare and provided valuable guidance for countries on maximizing the benefits of AI while minimizing its risks and avoiding the associated pitfalls.
The report warned of the dangers in overestimating the benefits of AI in health, especially at the expense of investments and strategies essential for achieving universal health coverage. According to the document, opportunities in the use of AI are linked to challenges and risks, including the unethical collection and use of health data; biases encoded in algorithms; and AI risks to patient safety, cybersecurity, and the environment. She also emphasized that systems built primarily on data collected from individuals in high-income countries may not work well for individuals living under low- and middle-income conditions.
Report of WHO on potential of AI in health financing
Despite an abundance of publications on the application of AI in different areas of health, the issue of health financing has received less attention. The WHO recently published the report “ The implications of artificial intelligence and machine learning in health financing for achieving universal health coverage – Findings from a rapid literature review” about the specific contributions of AI and ML to aspects of health financing. According to this document, there is little evidence of the effects, risk, and challenges of these technologies in relation to health financing. The survey was compiled from a review of 38 studies (carried out between 2000 and 2021) on the implications of AI and ML in health financing, which were published on platforms such as Google Scholar and PubMed. Most of the reviewed studies analyzed the application of technologies in supplementary health.
The report addresses topics such as cost forecasting, risk management, fraud detection, and the identification of opportunities for targeted policies, highlighting the potential positive impact of AI compared, mainly in terms of the speed and accuracy of data analysis compared with traditional statistical methods, in addition to the increased ease of applying AI techniques to large volumes of data. The researchers reiterated the limitations of their study and concluded that the use of AI and ML could help improve universal health coverage.
The reviewed articles indicated that in health expenditure forecasting, using ML could improve the efficiency and equitability of resource allocation, in addition to improving risk adjustment and population management to meet the needs of specific groups. However, if not properly used, ML-translated information could create the opposite effect of excluding the groups most in need and increasing the benefits received by the operators instead.
According to research by Accenture Consultancy, it is estimated that in the coming years, the value of the world market for AI applications to medical assistance will exceed US$ 7 billion. In 2011, the International Business Machines Corporation (IBM) launched a supercomputer to process health data; this computer became a reference database in the field of oncology and one of the most innovative applications in AI. Following the same path, Google launched the DeepMind supercomputer, gathering data from thousands of patients to increase their knowledge of various pathologies, including their symptoms and evolution. These applications, which were focused on research centers, formed the basis for a new phase of AI in medicine, supporting its popularization.
Alphabet, which is the parent company of Google, is the major investor in health-related AI research worldwide. The Massachusetts Institute of Technology (MIT), Stanford University, and Harvard University in the United States and Oxford University and Cambridge University in the United Kingdom also stand out for their contributions to this field. Additionally, the University of São Paulo (USP), State University of Campinas (Unicamp), and Federal University of Minas Gerais (UFMG) in Brazil are among the universities most dedicated to this field of research.
The report “Artificial Intelligence Index Report 2022” pointed out that the private sector worldwide would invest more than US$11 billion for research and innovations in the field of AI in healthcare in 2021. In the last five years, related resources have totaled US$28.9 billion, positioning the healthcare sector as the largest recipient of private investments in AI, surpassing traditional sectors such as finance and retail that are using AI for information technology. Computer vision, which involves segmenting images of organs, lesions, or tumors, has aroused tremendous interest in the medical community. These numbers reinforced the importance of AI in technological developments and health advances.
Currently, AI is evolving rapidly. A few years ago, the development of applications such as OpenAI’s ChatGPT and Google’s Bard was limited to far-fetched discussions in technology roundtables. However, the potential of AI technologies has greatly changed in recent months, with them developing at a breathtaking speed, creating a mixture of both awe and apprehension worldwide. With the increased impact of AI on our lives, there is a clear need to examine current regulatory guidelines regarding its use and implementation.
Regarding Brazil, it is a country that faces social inequality and has limited Internet access, with 28.2 million Brazilians (15.3% of the population) still not having access to an Internet connection at home in 2021, according to the Brazilian Institute of Geography and Statistics (IBGE); in this case, is AI the solution or the problem? Dr. Torres believed that it was certainly a part of the solution, considering the incorporation of AI into public health policies and its role in helping improve and streamline the diagnosis, treatment, and prevention of diseases. She concluded, “At an individual level, Internet access is important for the implementation of telemedicine and, in this regard, there is still a long way to go.”