Artificial intelligence (AI) holds significant promise for improving health care delivery and patient outcomes. While AI has seen widespread adoption in sectors like finance and information technology, its integration into health care has been slower. This delay is primarily due to the ethical, regulatory, and safety concerns associated with human health. Additionally, AI must gain social acceptance to be fully embraced in the health sector.
Strategic research in AI is crucial for advancing health systems and digital health. This includes enhancing the collection, monitoring, and management of information, as well as improving hospital and government information systems.
AI: a few definitions
Artificial Intelligence is a field dedicated to creating intelligent computer systems which are capable of simulating human thought and behaviour.
The definition proposed by the High-Level Expert Group on Artificial Intelligence suggests AI encompasses both technology and study.
‘Artificial intelligence systems are software (and possibly also hardware) systems designed by humans that, given a complex goal, act in the physical or digital dimension by perceiving their environment through data acquisition, interpreting the collected structured or unstructured data, reasoning on the knowledge, or processing the information, derived from this data and deciding the best action(s) to take to achieve the given goal. AI systems can either use symbolic rules or learn a numeric model, and they can also adapt their behaviour by analysing how the environment is affected by their previous actions.’
‘As a scientific discipline, AI includes several approaches and techniques, such as machine learning (of which deep learning and reinforcement learning are specific examples), machine reasoning (which includes planning, scheduling, knowledge representation and reasoning, search, and optimisation), and robotics which includes control, perception, sensors, and actuators, as well as the integration of all other techniques into cyber-physical systems.’
Most contemporary AI systems fall under artificial narrow intelligence (ANI) or Weak AI, designed to perform specific tasks. In healthcare, ANI applications range from image recognition and natural language processing, to complex clinical decision-making, such as medical diagnostics. Despite their promising performance, ANI models are limited in scope. For instance, an ANI model trained to diagnose diabetic retinopathy from fundus images cannot be used to detect pneumonia from chest x-ray images.
In contrast to ANI, artificial general intelligence (AGI) or Strong AI aims to replicate true human intelligence, capable of learning and performing any task. However, no AI systems have yet achieved AGI capabilities.
AI in health domain
The use of AI in healthcare is gaining significant attention across various sectors, including organ, cell, and functional tissue diagnostics, care robotics for interventions assistance, rehabilitation and patient support, biomedicine from genetics to modelling, and personalised biomedicine.
Benefits
AI-powered systems in healthcare can perform a wide range of tasks autonomously or semi-autonomously, offering numerous advantages. AI can collect and disseminate information beyond the constraints of time and place, enhancing medication adherence through AI-driven apps and supporting diet and exercise routines via virtual health assistants.
Studies have shown that AI can outperform human capabilities in certain areas, such as analysing chest x-ray images by radiologists. In medical imaging, AI-assisted diagnoses can significantly increase workflow efficiency by processing over 250 million images daily. This is said to not only improve the quality of care by reducing human errors but also frees up time for clinicians and healthcare workers. AI accelerates processes like diagnosis, reducing waiting times, enhancing communication with care teams, providing decisional support, and automating routine tasks (e.g. progress monitoring). Consequently, clinicians and healthcare workers can focus more on complex tasks, such as managing rare diseases and interacting with patients, which helps reduce burnout, job dissatisfaction, and workforce shortages.
Limitations and risks
Despite the increasing availability of regulatory guidelines from organisations like the World Health Organisation and the European Union, the use of AI in healthcare remains controversial due to challenges in ensuring data privacy and proper data usage.
AI in medicine and healthcare carries risks, including the potential for errors that could harm patients, data privacy and security issues, and the perpetuation of social and health inequalities. AI systems may incorporate existing human biases and patterns of discrimination or reinforce social inequalities in healthcare access. For example, an AI-driven pulse oximeter developed with incomplete or biased data overestimated blood oxygen levels in patients with darker skin, leading to the undertreatment of hypoxia. Similarly, facial recognition systems have been shown to misclassify gender more frequently in individuals with darker skin tones. It has also been shown that populations subject to discrimination are often underrepresented in datasets used to train AI solutions, potentially denying them the full benefits of AI in healthcare.
Conclusion
AI offers significant benefits in healthcare, including round-the-clock availability, ease of use, and increased efficiency in service delivery. However, there are numerous ethical concerns raised when it comes to AI, such as data security risks, changes in the patient-physician relationship, and the potential for increasing social inequalities. To address these issues, strategies such as enhancing transparency in predictive accuracy and information sources have been suggested. Future research and development in AI should focus on these areas to further its adoption and improve healthcare service delivery.
Véronique Ropion, MD
Strategic Projects Director, Pharmalys Ltd
June 2024
References:
- Chew HSJ, Achananuparp P. Perceptions and Needs of Artificial Intelligence in Health Care to Increase Adoption: Scoping Review. J Med Internet Res. 2022 Jan; 24(1): e32939. Published online 2022 Jan 14. doi: 10.2196/32939: 10.2196/32939
- Delipetrev B, Tsinaraki C, Kostic U. Historical Evolution of Artificial Intelligence. Luxembourg: Publications Office of the European Union; 2020.
- Ethics guidelines for trustworthy AI. European Commission. [2021-12-28]. https://ec.europa.eu/futurium/en/aialliance-consultation.1.html
- Federspiel F, Mitchell R, Asokan A, et al. Threats by artificial intelligence to human health and human existence. BMJ Global Health 2023;8:e010435. doi:10.1136/ bmjgh-2022-010435
- Giansanti, D. Artificial Intelligence in Public Health: Current Trends and Future Possibilities. Int. J. Environ. Res. Public Health 2022, 19, 11907. https://doi.org/10.3390/ ijerph191911907
- Murat Civaner M et al. Artificial intelligence in medical education: a cross-sectional needs assessment BMC Medical Education (2022) 22:772. https://doi.org/10.1186/s12909-022-03852-3