AI medical care has surpassed the "icing on the cake" stage

Author: Li Minger

**Source: **AI New Intelligence

Healthcare has become one of the most popular industries transformed by AI.

Today's AI is penetrating into all fields and links of the medical industry at an alarming speed and power. Recently, Sequoia Capital released an article titled "Generative AI in Healthcare" (Generative AI in the Medical Field), which also has an impact on He conducted a comprehensive and in-depth analysis of the application and development of AI in the medical industry and believed that it has "huge potential" in the future.

So, why does the medical field receive so much attention from capital in the current AI track?

Current status of AI medical care

In the report "Generative AI in Healthcare", Sequoia mentioned some important applications of AI in the medical field, including patient interaction, documentation, clinical decision-making, etc.

According to Sequoia Capital, current medical AI has leapfrogged the "icing on the cake" stage and begun to empower the core links of the medical industry. Such empowerment has greatly improved the efficiency and quality of the medical field. Reduce costs and manpower.

Specifically, the core links of the medical industry include six major links: patient interaction, documentation, clinical decision-making, pre-authorization, coding and revenue cycle management.

The main reason why the latest generative AI can empower these core nodes is that it can process large amounts of unstructured data and transform it into useful information and insights.

The core aspects of medical operations often involve multiple types of data, such as voice, text, images, videos, signals, etc. These data are often unstructured, that is, there is no fixed format or standard.

It contains a wealth of medical knowledge and value, but it is difficult to be effectively integrated or used by humans or traditional software systems.

In the traditional medical industry, the processing and integration of these data are costly but difficult to omit.

The U.S. medical coding market is worth approximately $21 billion and includes approximately 35,000 medical coders. Despite this large workforce, U.S. hospitals lose nearly $20 billion in revenue each year due to coding errors, leaving local providers to rely on cottage-style consulting firms to help them “find” missing information. income.

Similarly, in the process of interacting with patients, the medical industry always requires a large number of clerical workers to organize various medical documents.

According to statistics from Sequoia Capital, there are currently about 1 million clerical staff in the medical industry in the United States, and the average annual expenditure per clerical staff is US$40-50K, which means that the medical industry spends at least US$400 million on such positions every year the cost of.

Generative AI can use advanced algorithms such as deep learning and natural language processing to analyze, understand, generate and convert these data, thereby improving the efficiency and quality of medical operations, reducing costs and manpower, and adapting to different data sources and environment.

For example, in documentation, generative AI can be used to automatically convert conversations between doctors and patients into electronic medical records and coding; in clinical decision-making, generative AI can be used to convert multiple data sources and formats, such as medical images and medical records. Reports, etc. are transformed into unified medical knowledge and data.

This advantage is why Sequoia believes that AI can directly hit the core aspects of medical operations.

AI empowers healthcare

In addition to the advantages of processing unstructured data, AI at this stage also empowers the medical field in more aspects, including AI-assisted diagnosis, AI medical image analysis, AI precision medicine, drug research and development, and medical care. Robots, and many other subdivided tracks.

Specifically, in terms of AI-assisted diagnosis, AI can provide possible diagnostic suggestions by analyzing patient symptoms, signs, test results and other data, helping doctors make more accurate and timely decisions. For example, Alibaba Health's AI doctors can provide 90% accuracy within 1.5 seconds, and Baidu's AI doctors can already identify more than 900 common diseases.

AI medical image analysis uses machine learning, computer vision and other technologies to automatically analyze and diagnose medical imaging data, and combines it with genes, clinical and other factors based on a large number of quantitative features, such as morphology, texture, grayscale, intensity, etc. Correlation analysis is performed on the data to discover biomarkers and prognostic factors of the disease.

In terms of precision medicine, AI can mine and analyze large-scale biological data such as genomes, epigenomes, and transcriptomes to provide a basis for personalized prevention, diagnosis, and treatment. For example, Deep Genomics' AI platform can predict the impact of genetic variations on protein function and phenotype, and Flatiron Health's AI platform can use real-time clinical data to provide optimal treatment options for cancer patients.

In terms of drug research and development, AI can accelerate the drug discovery and development process by modeling and simulating data such as drug targets, drug structures, and drug action mechanisms. For example, BenevolentAI's AI platform can mine new drug candidates from massive literature, and Atomwise's AI platform can reduce experimental costs and time through virtual screening.

Judging from the current overall situation of AI medical care, AI medical technology, especially in some emerging and cutting-edge fields, such as genomics, immunomics, neuroscience, etc. Foreign AI companies often have more resources and experience.

For example, Google's DeepMind team, which specializes in disease gene searches, has used artificial intelligence systems to analyze the structures of almost all proteins in the human body.

This way, AJ can tell whether the letters in DNA will produce the correct structure. If not, it will be listed as a potential causative factor.

Similar examples include Paige.AI, which uses AI technology to help doctors analyze cancer pathology images and discover new treatments and drugs.

Paige originally used 1 billion pictures of 500,000 cancer medical pathology slides to create the world's first large-scale basic model. In cooperation with Microsoft, the two parties will develop the world's largest cancer picture AI model, with up to one billion parameters.

Although domestic AI medical technology has made breakthroughs in some fields, such as imaging diagnosis and intelligent consultation, there are still some technical difficulties and challenges, such as data islands and data quality.

At the same time, domestic AI medical application scenarios are relatively concentrated, mainly on the auxiliary side and the data side, such as CDSS (clinical decision support system), smart medical records, and medical data intelligence platforms.

Among the representative companies that have emerged are AI companies such as Lianyingzhi, which uses AI technology to perform imaging diagnosis.

Through CT cameras equipped with intelligent algorithms, deep learning convolutional neural networks and typical pattern recognition algorithms are innovatively combined to accurately identify the CT scan range.

Similar domestic companies include hypothetical medicine that uses AI technology for clinical diagnosis.

Its main technology is to imitate human cognitive processes through deep learning and convolutional neural network models, allowing AI models to automatically mine patterns in medical images.

Its AI product InferOperate performs deep learning on various types of neuroimaging data such as electroencephalography and brain functional imaging to extract image features and locate lesions, thereby providing doctors with intelligent surgical planning and fully automatic intraoperative positioning and navigation.

Trends and Opportunities

At present, although there is still a gap between domestic AI medical care and foreign countries due to industrial ecology, technical foundation, computing resources and other reasons, in terms of market growth rate and scale, domestic AI medical care development has a large market. space and growth potential, facing high medical demand.

According to data from the Huajing Industrial Research Institute, the market size of China's AI medical industry will be approximately 9.5 billion yuan in 2021 and is expected to reach 38.5 billion yuan in 2025.

Source: Huajing Industrial Research Institute

In the foreseeable future, domestic AI medical care will continue to make efforts in major fields such as AI drug research and development, AI+ pathology, AI medical imaging, and AI medical devices.

From the perspective of market demand and scale, AI medical imaging and AI drug research and development will become the main growth breakthrough.

Specifically, AI medical imaging applications are relatively mature, with a large number of products on the market. According to Global Market Insights data, the global AI medical imaging market accounts for 25% of the medical AI market, making it the second largest market segment after AI pharmaceuticals.

For the domestic medical industry, the current annual growth rate of medical imaging data in our country is as high as 30%, but the annual growth rate of imaging doctors is only 4%.

Considering that the training cycle for doctors is relatively long, the development of AI imaging medicine can effectively alleviate the shortage of medical talents, and the market still has great growth potential.

According to 36Kr analysis, the compound annual growth rate (CAGR) from 2020 to 2025 is expected to be 39.4%, and will exceed 30 billion yuan in 2025. Among them, AI medical imaging market share is the highest, reaching 50.6%.

In terms of AI drug research and development, AI can effectively solve the problems of high cost, low efficiency and high risk of new drug research and development.

The market size of my country's new drug R&D industry in 2020 is 1.2 trillion yuan, but the success rate of new drug R&D is only 11.3%. Even if it enters Phase III clinical success, the success rate is only 53.4%, and the overall cost of the clinical phase accounts for as high as 70%.

This shows that the research and development of new drugs requires a huge investment of money and time, but the benefits and risks are very uncertain.

Through the cognitive ability of artificial intelligence, we can accelerate target discovery, compound screening, drug design and other links, which can effectively improve the success rate and quality of new drugs.

In 2021, my country's AI pharmaceutical company Yingsi Intelligent cooperated with Zhejiang University to use a self-developed AI platform to optimize the design of the anti-cancer drug PD-1 antibody, and obtained clinical trial permission from the US FDA.

Such achievements show the potential of AI technology in the research and development of new drugs and also indicate the possibility of its large-scale growth.

According to a medical industry report released by DPI, the global market size of AI drug research and development is expected to grow from US$4 billion in 2020 to US$20.3 billion in 2027, with a compound annual growth rate of 26.5%.

In the current AI wave that is in full swing, the empowerment brought by large models is just the tip of the iceberg in the field of AI medical care. As the AI revolution continues, AI medical care, which has already gained momentum, will surely bring more opportunities and highlights.