By Martina Sordi
DOI: 10.57912/27932148
In 2020, the COVID-19 pandemic shocked the world. In the months following the outbreak, Italy suffered more than most. According to the World Health Organization, as of December 2023, Italy had over 26 million confirmed cases and approximately 193 thousand confirmed deaths, with cases still being reported today. Italy has the eighth-highest COVID-19 death toll in the world. Although vaccines were able to slow down transmissions and deaths, the image of military trucks transporting bodies from hospitals in Bergamo and other cities to local cemeteries will be hard to erase from the minds of the Italian people.
One of the main criticisms of the Italian National Healthcare System (INHS) that arose during and after the height of the pandemic, was the low number of beds and medical staff compared to the number of patients. As of 2021, there are around 40,000 registered doctors, a 1.45% decrease since 2020. It is predicted that by 2025 at least 29,000 doctors and 21,000 nurses will retire. While other countries within the Organization for Economic Co-operation and Development (OECD) show an upward trend in hiring new nurses, Italy’s is only decreasing. The number of students graduating with a nursing degree in Italy is 17.2 for every 100,000 people, versus an OECD median of 42.8. Between 2021 and 2025, all but two regions will see a decline in general practitioners. An alternative proposal– pioneered by experts in the field and political leaders– hoping to mitigate the effects of this decline is the use of artificial intelligence (AI). The implementation of AI in healthcare is a recent innovation. Advocates see its potential to address staffing shortages in middle-low income communities. Detractors resort to ethical implications, however, AI can help improve the performance of healthcare facilities, reduce the workload for health workers, and use a diverse approach to identifying possible diseases.
Looking at AI in public health, healthcare systems have lagged behind other industries in adopting and developing AI technologies. AI has been praised as a tool for diagnostics. AI can hasten diagnoses while improving accuracy when compared to the existing methods. AI algorithms have the capacity to analyze large data sets, including lab testing and medical imaging (X-rays and CT scans) accurately. Natural language processing (NLP) methods can extract various data from medical texts, like a doctor's personal notes and reports, to show patterns that may also help predict diseases. These specific NLPs have to be designed in a way that ensures the AI results are as trustworthy as those conducted by doctors. Lastly, Artificial Intelligence can aid doctors in identifying recommendations for patients, while providing possible recommendations on how to improve clinical outcomes.
While Italy has started developing NLPs, little progress has been made. The lack of NLP adoption is due to the lack of datasets available that could be domesticated for AI. A database takes time and resources to craft, and with not enough data available, there is not enough to build a proper NLP. A structural roadblock for the success of NLPs is a linguistic one. The Italian language does not always guarantee complete comprehension when it comes to all its grammatical structures and declinations, not to mention the various Italian dialects in Italy. In contrast, NLPs in English have already been established and further developed, thanks to an ample language database.
With the marked decrease in medical staff, ensuring that patients are moved to the correct department is critical. AI would also ensure that staff is assigned and reassigned to efficiently deal with the ups and downs of patient flow. While AI cannot specifically help with moving patients physically, it can help with estimating the number of patients in advance, therefore enabling smoother operational planning within the emergency departments. By doing this, there is a decrease in patient overcrowding especially in crucial departments, which can help healthcare workers in their daily tasks and management. One of those daily tasks is reviewing patient records.
However, there is always the risk that AI implementation in public health could result in potential failures. If an algorithm is not programmed correctly, there is a chance of false negatives resulting in the missing life-threatening disease. Likewise, there could also be a false positive, due to AI errors, and lastly, there could be unnecessary intervention in emergency departments on the part of medical staff because of imprecise diagnosis. In addition to AI presenting errors in the field of diagnostics, the European Parliament report on Artificial Intelligence in Healthcare also outlines possible other risk factors that could be impacted by AI. For example, the report outlines the risks of bias, lack of transparency and privacy, as well as gaps in AI accountability. This serves in contrast to the collective of European experts mentioned previously.
The AI Act published in 2024 (EU regulation 2024/2689), is a legal framework that examines the risk of AI across EU countries. The proposal classifies AI tools according to four main levels of risk: unacceptable, high, limited, and minimal. The framework lays out that while there are some ‘systemic risks’ of AI in public health, the technology can be developed to safeguard public interests, in the fields of disease detection, diagnosis prevention, control and treatment, and the overall improvement of health care systems. Regulation 2024/2689 harmonizes the rules of AI, while at the same time amending previous regulations. EU regulation 2017/745, adopted in 2021, was the first to give the go-ahead for the use of AI in the medical field for all EU countries. A report published in 2021 by the Italian Ministry of Health, states that per these regulations, Italy could draw inspiration and implement AI into their institutions. In 2024 the ‘Agenzia per L’Italia Digitale-Agency for a Digital Italy’ (AGID) produced the 2024-2026 Italian strategy for AI. The AGID lays out a plan to develop AI in four major groups: research, public administration, businesses, and advanced training in AI.
Artificial intelligence could benefit healthcare to a considerable degree. Particularly in Italy, the upgrades in diagnostics and patient flow can greatly benefit the currently struggling public health system. While AI is not at the level to take the place of doctors, the development of advanced NLPs can significantly aid the remaining healthcare workers. If Italy decides to implement AI in the field of diagnostics in the next few years, it could substantially transform the speed and efficiency at which doctors operate. Eventually in the future, if AI shows considerable improvements, reliability, and a low number of errors, the debate could be opened for the potential use of AI to function as an active assistant to doctors. However, for now, countries like Italy should begin applying AI to their basic everyday public health routines, to advance their healthcare systems and ease the burden of medical personnel.
Yorumlar