Generative AI Illuminating Healthcare: How Big Language Models Are Energizing Digital Healthcare

21st Century Business Herald reported by Zhao Na Shanghai

GPT has revolutionized various industries.

In April, Microsoft and Epic Systems announced that they would bring OpenAI's GPT-4 AI language model to healthcare to help healthcare workers respond to patient information and analyze medical records. Integrating generative AI into daily workflows can increase caregiver productivity, allowing them to focus on clinical responsibilities that really require their attention.

"The advent of GPT is as revolutionary as the personal computer, and it will revolutionize the way we work, learn, communicate, and even disrupt healthcare." Bill Gates said that AI has already played an important role in improving disease detection and diagnosis, and will help more research breakthroughs in the future, so that those who do not have access to medical services can also obtain accurate and reliable medical advice.

In China, Fang Weihao, chairman and CEO of Ping An Health, revealed at the March earnings conference that this year it will continue to use ChatGPT to help specialists cover diseases outside the template, which can play a very strong auxiliary role in the health management and health consultation dialogue between family doctors and patients.

According to data, since March this year alone, more than 20 domestic companies have entered the large-model track. On May 31, Baidu announced the establishment of a fund with a scale of 1 billion yuan, focusing on investing in incubating high-quality start-ups in the field of large models.

"Artificial intelligence has always been widely used. After generative AI comes out, there will be a wider range of opportunities in the field of AI. Zhang Lu, founding partner of Fusion Fund, said in an interview with the 21st Century Business Herald that in the eyes of CEOs and CIOs of many artificial intelligence giants, generative AI brings new opportunities to many different industries, and the biggest benefit will be medical care.

GPT promotes the implementation of medical innovation

When ChatGPT is frequently out of the loop, the landing of generative AI in vertical fields has gradually become a hot spot.

"In the large language model of ChatGPT, there may be tens of billions of nodes, which is actually simulating the human brain. The human brain has 20 billion to 30 billion neurons, and ChatGPT is simulating the knowledge and thinking ability of the human brain. Shang Hua, vice president of Gobo Medical Group, said that at the 2023 Life Valley Digital Healthcare Innovation Forum held in Beijing in April this year, he proposed that the ultimate form of digital medical development in the future will be a true digital human, "This is the avatar of people in the virtual world, which can show various conditions of the human body and help diagnose treatment, clinical research, and surgery at low cost." ”

In a recent interview, the reporter found that although entrepreneurs and investors with different backgrounds see different medical futures, "subversive", "empowering" and "change" have become common keywords for the changes that generative AI can bring to the medical industry.

"There's a lot of money going into this space right now." Zhang Lu analyzed that from the perspective of large language models, the data foundation, data quality and data professionalism in the medical field are the highest, which makes investors pay special attention to the application of generative AI in this field.

"Although ChatGPT does not have many direct applications, and it is difficult to directly solve medical problems (in the short term), the new scenarios and opportunities it can bring to medical AI are certain." Gong Enhao, founder and CEO of Subtle Medical, believes that generative AI will bring more innovative applications to the medical field, such as helping to improve the patient experience and the efficiency of doctors' diagnosis.

For example, GPT's multimodal AI can realize comprehensive processing applications based on multimodal data such as text, speech, images, and videos. In the chat scenario, patients can enjoy more humanized AI consulting services; In non-chat scenarios, multimodal data analysis can bring more efficient data feedback to doctors.

Deep Penetration Medical is an AI medical imaging company founded in 2017, whose core business is to use AI to accelerate the speed and improve the imaging quality of MRI and PET. This process itself uses generative AI to process raw data to obtain synthetic data, and then reconstruct MRI and PET images based on the synthetic data.

Digital health companies embrace generative AI

As an interactive artificial intelligence model, ChatGPT can summarize the core information about the disease in concise language, and further communicate according to follow-up and information supplementation, the main value is to save time and cost, and improve the efficiency of consultation.

When the scope of attention is expanded to the application of GPT models in the medical field, scenarios including the interpretation and generation of electronic health records, pre-diagnosis and condition analysis, patient care and health education, medical image interpretation, health management and electronic medical records can be seen.

Take medical imaging AI as an example. Medical imaging provides rich and effective information for doctors and is one of the important bases for clinical disease evaluation. The difficulty of its practical application is that, on the one hand, the picture quality of imaging equipment will restrict the judgment of doctors, on the other hand, limited by the time and energy of doctors, large-scale diagnosis is difficult to achieve in the clinic.

The aforementioned pain points are transmitted to the patient, which is the pain points such as the shortage of doctors, the imbalance in the allocation of medical resources, and the difference in diagnosis results.

Let's look at the size of the market. According to the report of Pacific Securities, China's AI medical imaging solutions market is expected to enter a period of explosive growth, increasing from less than 1 billion yuan in 2020 to 44.2 billion yuan in 2025, with a compound annual growth rate of 135.9%.

Gong Enhao told reporters that unlike the product idea of AI diagnosis, AI imaging companies empower doctors with high-quality images. For example, in the solution of deep medical care, the process of improving imaging speed and image quality with AI technology includes using generative AI to process raw data to obtain synthetic data, and then reconstruct MRI and PET images based on the synthetic data.

Further, the essence of generative AI is to learn from input data through deep learning models and generate new data. This means that it can have three typical applications in the field of medical imaging, the first is to generate synthetic data based on raw data to bring image enhancement effect to imaging equipment imaging; Second, through model training, it is possible for R&D personnel to open up some scenarios where data is missing; Third, future projections can be made based on existing data, i.e. estimates of health status and disease risk.

The industry expects more AI models to emerge

AI medical care is an important field for the application of artificial intelligence industry, and it is also a topic that academia, industry and capital circles continue to pay attention to.

Talking about the future, Peter Lee, senior vice president of Microsoft and head of Microsoft Research, wrote in "Beyond the Imagination of GPT Healthcare", "We are more likely to see ChatGPT or GPT-4 as a disruptive single point. But in the near future, there will be more and more powerful AI models. ”

Both Li and his two co-authors believe that while GPT-4's role in medicine and healthcare may be limited, subsequent AI systems will gradually approach and surpass human capabilities in medicine. The most important thing is to figure out how methods in the medical field fit into the evolution of AI systems in order to maximize the benefits of human health.

Zhang Lu's team has invested in projects such as Huma.ai, Proscia, Rhino Health, and Deep Penetration Medical in the direction of the combination of artificial intelligence and medical care. She told reporters that many investment institutions with scientific and technological backgrounds and medical backgrounds are paying attention to the direction of AI medical treatment. VCs with different backgrounds bring different resources to portfolio companies, so she recommends that entrepreneurs accept investment institutions with diversified backgrounds.

Zhang Lu also believes that more new models and technologies will emerge rapidly in the coming years. Many technologies that currently seem impractical will become more practical and enter the commercialization phase after research in the coming years.

It is worth noting that VCs in China and the United States, as well as different investors in the same region, have different project preferences. Some medical investors from domestic VC institutions have reminded that in the context of the cold financing in recent years, VCs prefer projects with application scenarios and open up closed-loop projects, "It is best for companies to have real income to prove themselves." ”

Of course, regardless of the heat of the capital market segment, entrepreneurs and investors believe in the opportunities that generative AI can bring to the healthcare industry.

"It's an exciting time." Zhang Lu said that generative AI will promote the digitalization of the whole industry in many industries. Medical care not only has a huge market, but also is related to people's livelihood and social development, and will continue to usher in more funds and talents.

For more information, please download 21 Finance APP