Health care using AI is bold, but much caution first
India cannot jump into AI-driven health care without first addressing the foundational issues within its health system.
India’s health care system is on the verge of a digital revolution, with AI-powered health solutions making headlines. Recent discussions about introducing “AI-powered primary care for every Indian, available 24/7” are ambitious. However, before diving headfirst into AI-driven health care, India must address several foundational challenges within its health system. While the potential of AI to transform health care is exciting, it also raises critical concerns about feasibility, ethics, and the readiness of India’s health infrastructure for such a leap.
(i) The Role of Primary Health Care
Primary health care (PHC) serves as the backbone of a functional health care system, ensuring that essential health services reach communities. PHC is designed to address the broader determinants of health, such as social, environmental, and economic factors, while empowering individuals to take charge of their well-being. By integrating services into local settings, PHC enhances access to care and ensures equitable health outcomes. The prospect of replacing or heavily relying on AI in PHC, however, could undermine this foundational principle. AI, being inherently impersonal, might transform patients into passive recipients of care rather than active participants, eroding the patient-centered approach central to health care.
(ii) Limitations of AI in Health Care
AI’s strength lies in its ability to process large amounts of data quickly and automate repetitive tasks. In health care, these capabilities can be useful for managing logistical tasks like predicting hospital supply needs, optimizing drug procurement, or even managing biomedical waste. However, AI lacks some essential attributes of human intelligence that are crucial to medicine, such as emotional intelligence, empathy, and cultural understanding. These human qualities are vital in patient care, where understanding the full scope of a patient’s condition goes far beyond pattern recognition.
Medicine is not just about diagnosing and treating diseases; it requires an understanding of the social and cultural contexts in which people live. AI systems, though capable of analyzing data, lack the consciousness and moral reasoning required to make sensitive decisions in health care. This makes it difficult to replicate the nuanced, ethical decision-making process that human health professionals use daily. Consciousness and awareness of the real-world environment drive human decision-making, a factor that AI cannot replicate.
Moreover, medical data, unlike other types of data, is often fragmented and difficult to access. This scattered nature of health data makes it harder to train AI models effectively. Developing AI systems that can fully understand and process health care data requires vast quantities of detailed and accurate information. But in health care, patient data is private and sensitive, raising ethical concerns about its use in AI systems.
(iii) Case Study: Naegele’s Rule and AI Challenges
A historical example from obstetrics — Naegele’s rule, which predicts the date of birth based on the last menstrual period — highlights the challenges AI faces in health care. Despite being widely used for over 200 years, Naegele’s rule is only 4% accurate. This is because it fails to account for critical factors like maternal age, nutrition, height, or race. To develop a better predictive model, AI would need vast amounts of personalized data. However, collecting such data presents ethical and privacy issues, as it belongs to patients and must be handled carefully.
This illustrates the inherent tension between improving AI accuracy and protecting patient privacy. Health care is not a field where errors can be easily overlooked, as the consequences of mistakes can be life-threatening. This further complicates the development of AI in health care, where both accuracy and ethical considerations are paramount.
(iv) The Cost and Complexity of AI in Health Care
The costs involved in developing AI systems for health care are substantial. Building the infrastructure to collect and analyze health data requires significant investment. Moreover, AI models need continuous updates and adjustments to stay relevant, especially in areas like reproductive health where patterns can shift over time. The complex nature of health care data adds to the challenge, as it is difficult to standardize across diverse populations like India’s.
India’s vast cultural and regional diversity further complicates AI implementation. The country’s varied population means that AI models must account for a wide range of health behaviors, environmental factors, and genetic predispositions. Generating data for AI models that accurately reflect this diversity is a huge challenge, and collecting the personal and behavioral information needed to do so raises ethical concerns.
(v) AI’s Potential in Health Care
Despite these challenges, AI has the potential to make significant contributions to health care in specific areas. Narrow AI, for instance, can be highly effective in performing specialized tasks such as screening medical images, predicting supply needs, or managing hospital logistics. Diffusion models, which are designed to detect patterns in complex datasets, could help in areas like histopathology or medical imaging, where precise analysis is crucial.
Large Language Models (LLMs) and Large Multimodal Models (LMMs) also have promising applications in medical education and research. By simulating patient interactions, these models can help train health care professionals and provide access to vast amounts of medical knowledge. In medical education, AI tools can offer personalized learning experiences, enhancing the training of health care professionals. However, while AI can be useful in educational settings, it still lacks the human touch that is so important in direct patient care.
(vi) The Black Box Problem and Ethical Concerns
A major concern with AI in health care is the so-called “black box” problem. This refers to the lack of transparency in how AI algorithms make decisions. In health care, understanding the reasoning behind a diagnosis or treatment plan is critical for building trust between patients and providers. If AI systems make decisions that are not easily understood, it could undermine confidence in the care process. Health care providers may be reluctant to rely on AI recommendations if they do not understand how those recommendations were reached, and this lack of trust could ultimately harm patients.
An example from the world of AI development is Google DeepMind’s AI algorithm, which defeated world-class players in the board game Go. While such achievements are celebrated in the gaming world, they raise concerns in health care. In games, the consequences of an AI mistake are minimal, but in health care, errors can have serious, even fatal, consequences. This highlights the need for caution in deploying AI in medical settings, where the stakes are much higher.
(vii) AI Governance and Data Privacy in India
A recent petition in the Kenyan Parliament by content moderators against OpenAI’s ChatGPT has brought attention to the ethical issues surrounding AI development, including the exploitation of underpaid workers in AI training. This has raised concerns about the potential exploitation of vulnerable populations in AI training and development, particularly in countries like India. Given that patient data is required to train AI models, safeguarding the interests of Indian patients is critical.
India currently lacks comprehensive regulation for AI, unlike the European Union, which has developed the Artificial Intelligence Act. In the absence of such legislation, it is crucial that AI tools in health care are developed and deployed with core medical ethics in mind, particularly the principle of “Do No Harm.”
(viii) Conclusion
While AI has the potential to revolutionize health care in India, it is not a solution that can be implemented overnight. The challenges of patient care, the need for high-quality data, and the ethical implications of AI demand a more measured approach. Before India can fully embrace AI-driven health care, it must address the foundational issues within its health system. Investing in data infrastructure, ensuring ethical use of AI, and preserving the human-centric nature of health care are critical to realizing the benefits of AI without sacrificing patient care.
