Significant changes are taking place in the healthcare and life sciences world, spurred by the emergence of generative artificial intelligence (genAI) and advancements in genomics.

The healthcare industry has faced some setbacks in the past, such as electronic health records (EHRs) not living up to the hype or genomics failing to deliver on its promise. However, the healthcare industry has reached a turning point where it can now handle and make sense of vast amounts of data in ways that were not possible before, revolutionizing how information is used to help patients.

Recently, a group of healthcare experts explored this topic at Nvidia’s GTC conference. Kimberly Powell, general manager and vice president of healthcare and life sciences at Nvidia, facilitated a thought-provoking panel discussion on how generative AI enhances patient care, accelerates research and improves efficiency.

Evolving from genomics to multi-omics

There’s a shift happening from genomics to multi-omics to understand human biology better. This involves examining DNA, RNA, proteins, methylation and epigenetics. By combining multi-omics sequencing with AI, researchers are on the verge of gaining unparalleled insights into health and disease, which marks a significant moment in medical research.

Historically, the cost and complexity of sequencing a human genome presented significant challenges. However, technological advances have drastically reduced these obstacles, enabling researchers to identify disease indicators more efficiently. Such progress, akin to lowering guesswork in advertising, represents a leap forward in healthcare, explained Cathie Wood, chief executive officer/chief investment officer at ARK Investment Management.

“In 2003, the first whole human genome was sequenced, and we learned later that it wasn’t exactly 100 percent sequenced. But that took about $2.7 billion and 13 years of computing power. Today, we’re down to $200 and a few hours of computing power. The convergence between sequencing and AI will cause the most profound results, transforming human life, plant life, animal life, and so forth,” said Wood.

ARK’s investments are driven by a decade of research focusing on the convergence of various technologies, emphasizing multi-omics sequencing and AI. Wood is optimistic about the future of healthcare, where AI can help identify and treat diseases at their genetic roots, marking a significant departure from traditional approaches.

Redefining EHRs and patient care

The challenges associated with EHRs in healthcare have long been discussed. One primary concern is how EHRs affect the patient-doctor relationship. While EHRs have many benefits, they have also placed an additional administrative burden on clinicians. As a result, doctors spend more time on data entry than directly caring for their patients. This shift has raised concerns about losing meaningful human interactions in healthcare.

Despite the setbacks, the potential for AI to help reduce the workload of healthcare professionals is becoming more apparent. Specifically, genAI models can streamline the often time-consuming task of entering data into EHRs. Hence, genAI has the potential to improve patient care by simplifying tasks like data synthesis and clerical work, according to Eric Topol, professor and executive vice president at Scripps Research.

Moreover, an intriguing and unexpected benefit of genAI in healthcare is fostering empathy. Despite machines' lack of emotional understanding, large language models (LLMs) developed through AI have demonstrated an ability to support empathetic communication. Such an innovation could revolutionize how patient care is delivered and reinforce the importance of empathy.

“Because of its ability to promote empathy, I would suspect that in the future, every clinician will have to undergo coaching by an LLM to be maximally empathetic and communicative. It’s yet another exciting dimension of the future of healthcare,” said Topol.

Embracing LLMs

Peter Lee, president of Microsoft Research, recalled his experience with the early adoption of OpenAI’s latest multimodal LLM, GPT-4. Lee noted both excitement and concern within the clinical community. Remarkably, GPT-4 has facilitated personal connections between clinicians and patients, suggesting a future where AI enhances the humanity of healthcare.

One of the most surprising and meaningful outcomes has been the ability of GPT-4 to enhance personal connections between clinicians and patients through “reverse prompting,” where the AI suggests personal touches to add to communication based on the patient’s life. This feature prompts clinicians to consider these details, fostering a deeper patient connection.

Moreover, LLMs are transforming EHRs. Epic Systems partnered with Microsoft a year ago to incorporate GPT-4 into its EHR software. Today, hundreds of thousands of clinical notes have been written using GPT-4, and over 70 health systems have integrated it into their operations. This rapid adoption contrasts with the usual slow pace of change in healthcare, highlighting the impact of generative AI.

Revolutionizing drug discovery

In drug discovery, genAI has created a wave of optimism and confidence among researchers. This excitement stems from AI’s ability to analyze vast amounts of data while decoding the complexities of proteins and molecules. Using AI to produce accurately labeled data, researchers can simulate complex physical processes and fast-track drug discovery.

Nvidia’s CEO, Jensen Huang, has demonstrated AI advancements for drug research, indicating that AI methods could soon become widespread. However, there are challenges to fully realizing AI’s potential in biology, particularly in genomics. The guarded nature of biotech and pharmaceutical data and concerns about AI’s disruptive impact may impede progress initially.

Nevertheless, AI’s benefits encourage a more open data-sharing approach, leading to a significant leap forward in drug discovery. An example of AI’s potential in medicine is BenevolentAI’s successful repurposing of Baricitinib, a drug used to treat arthritis, as a viable medication for COVID-19. The FDA has acknowledged this achievement, a testament to how AI can help streamline the drug development process.

“When I first got to Scripps Research more than a decade ago, it would take two or three years to develop the crystal structure of a protein. Now, it takes minutes. It was the first antibody nonexistent in nature. This is where hallucinations are great. You want to hallucinate things, giving us new insights into the biology of diseases that we never had before - not just in cancer but across all the other major chronic conditions,” said Topol.

Addressing regulatory challenges

Establishing regulatory frameworks in healthcare is essential to fully exploit AI’s potential without compromising patient safety or hindering innovation. There is a growing push to create basic guidelines that will help shape the rules around this technology. The National Academy of Medicine and the American Medical Association are already working on guidelines, a positive step towards establishing these rules.

“We need regulation to provide clarity to companies like ours [Microsoft]. People might react negatively to a new technology in medicine. You would never expect your obstetrician to listen to a pregnant woman’s belly and then pronounce the health of the baby; you expect them to use an ultrasound. In the near future, it would be just as foolhardy to think that a doctor might practice medicine without the assistance of AI,” said Lee.

Governments also have an essential role to play in fostering innovation. Everyone involved in healthcare, from medical professionals to lawmakers and tech developers, must work together to ensure AI is deployed to benefit both the industry and the patients.

AI has the power to change almost every aspect of our lives. There is so much data in healthcare but far too much for people to connect the dots – but machines can. Many people attend GTC for the technical information, but this is an example of an industry session that shows what’s possible with AI.