The potential of generative AI in healthcare – empowering patients and clinicians

Despite early rhetoric that artificial intelligence (AI) would replace the need for clinicians and lead to job losses in the sector, AI is now firmly embedded as a valuable tool in clinical practice. It’s helping radiologists spot tumors that could be easily missed, helping ICU nurses to identify patients at imminent risk of deterioration so they can intervene early, and helping hospitals predict patient demand in advance so they can plan staff and equipment resources accordingly.

Still, today AI is only delivering a fraction of what it can potentially do to improve patient care. Generative AI – artificial intelligence capable of generating text, images, speech, or any other media – has the ability to transform healthcare in ways that were previously thought impossible. Why and how you may ask? 

To answer that question, let’s look at one important example. Generative AI’s ability to take data, present it in a human-friendly form: natural language, text, graphics. Even more importantly, Generative AI allows users to ask questions from the data, thereby facilitating a meaningful two-way conversation with data. And meaningful conversations are fundamental to our ability to make well thought-out decisions, learn from one another, explore new ideas, and even explore contrarian perspectives.

In healthcare, it is vitally important that those conversations center not only on the clinical data but also on the patient. The complexity of disease and the personalization of medicine often leave today’s clinicians and patients with several different therapy options to choose from, each with its own advantages, such as efficacy, and disadvantages, such as unwanted side effects. In practice, these difficult conversations already take place on a daily basis, but often under stressful conditions.

Generative AI offers the ability for both parties to be better informed and better prepared for an open discussion about the best course of action. By facilitating natural language interaction between the digital world and people, it will allow clinicians and patients to have deeper interactions with clinical data at a time and place of their own choosing.

Supporting clinician decision making, engaging patients in their care

With access to the patient’s clinical results and medical history, generative AI will empower clinicians to explore the data in greater detail with far greater speed and efficiency than sifting through disconnected paperwork, PDF files, or computer screens. It will turn vast amounts of data into actionable insights, summaries, AI-based predictions, and intuitive visualizations, allowing them to evaluate and explore different therapy options. Deeply embedded into clinical workflows, it will support doctors and nurses in providing better patient care rather than further complicating their daily workload.

At the same time, patients will be empowered to ‘talk to’ their medical records, asking questions relevant to their own unique circumstances and perspectives and getting answers in an easily understood form. By allowing clinicians and patients to have meaningful conversations with the data, both are able to extract meaningful insights in the moment, and at a literacy level commensurate with their knowledge base. Such meaningful interactions will enrich the patient-clinician face-to-face interactions and support better decision making. Generative AI has the potential to enable healthcare providers to treat more patients with more complex conditions, do it more efficiently, and achieve better patient outcomes. 

The ability for people to ask questions of their medical record need not be restricted to people who become ill – the point at which they become ‘patients’. Healthy individuals will equally benefit from generative AI that provides them with highly personalized health education and personalized coaching, allowing them to ask those highly personal ‘what-if’ questions that lead to health-improving behavior change and the opportunity for patients to actively engage in their medical journey and live healthier lives.

Nevertheless, there remain significant issues that need addressing to truly harness the transformative power of generative AI in healthcare. On the clinical side, it must be deeply and seamlessly integrated into clinical workflows, so that it intuitively assists caregivers at the point of decision-making rather than being yet another application they have to deal with. The interoperability of data must be standardized and rigorously applied, and security and privacy guaranteed. And we must make sure the output of generative AI is fair and unbiased, which means fine-tuning it with diverse data sets and rigorously validating its performance for specific use cases and patient cohorts to avoid perpetuating or amplifying existing healthcare disparities.

Only by addressing these key issues, while keeping the focus firmly on the caregiver and the patient, will generative AI achieve the level of trust required for widespread clinical and patient acceptance. This is why the safe and responsible use of data and AI, for example via data and AI principles and via an Artificial Intelligence Code of Conduct for healthcare, is so important.

If we steadfastly start with clinician and patient needs, embed generative AI deeply into the clinical workflow, and keep caregivers and patients firmly in the loop to build trust, the transformative potential of generative AI to enhance diagnostic capabilities and improve access to care is truly immense.