"The AI + Healthcare Favored by the Nobel Prize Has a New Solution in China"

In 2024, the Nobel Prize in Chemistry was awarded to David Baker and others to recognize their research on protein design and structure prediction using AI technology, which has brought unlimited imagination to pharmaceutical research and development. In recent years, with the penetration of artificial intelligence technology in various fields of society, the medical industry is also expecting a digital product that "subverts" the traditional diagnosis and treatment methods, while improving the clinical treatment effect and reducing the burden on medical insurance and patients.

This also means that the greatest potential of the combination of AI and healthcare will not be limited to the application in a certain treatment scenario, but to participate in, optimize, and even rewrite all aspects of the entire medical service process.

According to the IDC report, it is expected that by 2025, the total global market value of artificial intelligence applications will exceed hundreds of billions of US dollars, and the Chinese medical industry will account for 1/5 of the total market size. When the tide comes, more and more enterprises, medical institutions, and universities and research institutes have begun to join the exploration of smart medical practice.

As a global leading pharmaceutical company, Roche is naturally an important participant in the "AI + Healthcare" wave. At this year's CIIE, Roche not only showcased several AI-driven digital products in collaboration with its partners, but also launched the "Digital Healthcare Incubation 2.0" program to continue to empower local smart healthcare innovation.

"The CIIE has always been a 'weathervane' of innovation. Roche has long been deeply engaged in the field of smart healthcare and has rich industry-leading experience. At the same time, Roche has experienced three decades of development and layout in China, and has the ability to link various resources, catalyze the acceleration of scientific and technological landing into clinical application products, and ultimately benefit every patient. Roche is willing to be a bridge connecting the academic and industrial sectors, bringing valuable and meaningful innovative achievements to medical research and patient treatment." said Dr. Li Bin, Vice President of Medical Affairs of Roche China.

Medical services are moving towards intelligence and individualization

Having gone through the early stage of "AI + Image Recognition" and the basic stage of medical informatization, the application of artificial intelligence in the domestic medical industry is developing rapidly both in terms of conceptual perception and technological iteration. The "Smart Healthcare Era" empowered by AI is coming.

Wang Jingjing, Executive Director of the Global Health Industry Innovation Center, recalled that when AI first entered the medical field in the past, everyone held a skeptical and cautious attitude. With the further iteration of AI technology, people have shown a more accepting and embracing attitude towards AI, expecting that "the development of technology can promote the entire industry to move towards smart healthcare."

Wang Jingjing, Executive Director of the Global Health Industry Innovation Center, delivers a speech (Source: Roche)

So, back to the original concept, what exactly is smart healthcare? Compared to the simple extension of medical informatization, it actually emphasizes more on intelligent decision support and automated service supply. By integrating and analyzing a large amount of medical data, it provides intelligent decision-making tools such as auxiliary diagnosis and treatment plans for doctors.

In recent years, on the hospital side, the demand for individualized, diversified, and multi-level medical services has been increasing, which is difficult to cope with only by traditional diagnosis and treatment methods. As a systematic "solution", smart healthcare can cover all aspects of individualized diagnosis, treatment, disease monitoring, and disease course management, improving medical efficiency in many aspects.

However, the development of digital products that meet the needs of smart healthcare is not a simple matter. From a technical perspective, this not only requires the AI model to analyze and learn multi-modal data, and understand the deep-level principles such as disease-related driver genes and abnormal protein structures. At the same time, it is also necessary to pay attention to the real performance and needs of doctors and patients in the actual clinical process, so as to achieve a perfect balance between "science" and "experience".

Shen Xiafang, Senior Director of the Hematology Field and Innovative Health Solutions in the Medical Affairs Department of Roche Pharmaceuticals China, mentioned that overseas, the global headquarters has laid out many "innovative projects developed by AI technology companies in the health field through mergers and acquisitions". Relatively speaking, although these projects are effective, their perspectives are still somewhat "peripheral". But in China, enterprises can transform valuable innovative projects by jointly studying a clinical problem with doctors, and such a link is "closer and more efficient".

Shen Xiafang, Senior Director of the Hematology Field and Innovative Health Solutions in the Medical Affairs Department of Roche Pharmaceuticals China, delivers a speech (Source: Roche)

Over the past few years, Roche has been continuously attempting to apply AI technology to various links in the pharmaceutical value chain, such as drug research and development, diagnosis, production, and patient management, through various forms of external cooperation and incubation, thereby improving overall productivity and efficiency, and reducing the burden costs of the medical insurance system and patients.

Through these practices, the company has found that the key to truly achieving the "balance" between science and experience and ensuring the practical application of smart healthcare products lies in three major factors: meaningful data on a large scale, advanced analysis technology, and valuable application scenarios.

Take data as an example, this is the "basic production factor" for the development of AI applications. "In China, the interconnection and sharing of medical data has always been a pain point. But in the past two years, we have also seen that the country is paying more and more attention to these data. We have established the National Data Bureau, and more than 200 data products from the medical end have been launched, and the transaction volume is considerable. This also means that in the future, when developing AI innovative applications, the limitations caused by data will be greatly improved, and data can be better combined with the increasingly breakthrough technologies and applied to the medical scenarios we need." Shen Xiafang said.

Specifically in clinical practice, currently, Roche is focusing on two major areas: clinical decision support and remote patient monitoring and management, and developing a relevant digital product portfolio. These product portfolios will not only bring more treatment options for doctors and patients, but also be an important engine for developing new-quality productivity and promoting high-quality development.

Taking cancer as an example, Roche's smart healthcare "China Answer Sheet"

The digital assisted decision-making model in the field of liver cancer is one of the cases of Roche's smart healthcare "China Answer Sheet".

China is a big country with a high incidence of liver cancer. In recent years, various new immunotherapies represented by PD-1 have greatly improved the overall survival period of patients. However, in actual treatment, due to factors such as differences in the condition and diagnosis and treatment capabilities, many patients cannot quickly find a suitable treatment plan, resulting in low effectiveness and a prominent adverse reaction rate, and the trial and error cost increases accordingly.

Some experts have described that using the same treatment plan for different types of liver cancer patients is equivalent to "opening a blind box", with a success rate of less than one-third. And the identification method for the effectiveness of liver cancer treatment is still blank, and domestic patients have few opportunities for trial and error. Therefore, the top priority in the diagnosis and treatment of liver cancer is to accurately identify the differences between patients and then select the best treatment plan.

In order to solve this problem, Roche, in collaboration with top tertiary hospitals in China, used radiomics to perform "digital biopsy" on tumors and developed a digital assisted decision-making model in the field of liver cancer. Its advantage is that it does not need to establish a new detection method, or use puncture to obtain additional blood samples, tumor tissue samples, etc., to achieve precise targeting of the target population. For example, this model can analyze the characteristics of the patient's lesion location, size, and number, and recommend a more beneficial treatment plan.

Professor Sun Huichuan, Deputy Director of the Liver Cancer Institute of Fudan University and Deputy Director of the Liver Surgery Department of Zhongshan Hospital Affiliated to Fudan University, said that the model has entered the clinical verification stage and can provide digital evidence for the effectiveness of the drug for patients, thereby helping doctors make decisions, increasing the certainty in the diagnosis and treatment process, bringing clinical benefits to patients while saving medical costs.

Professor Sun Huichuan shares the application of the digital assisted decision-making model in the field of liver cancer (Source: Roche)

The relevant application results show that the accuracy rate of this model in predicting the effectiveness of a certain method in treating patients exceeds 80%, and the clinical benefit of patients receiving treatment is expected to increase by 2.6 times, and the model has been shown to be effective in the practice of multiple hospitals.

In addition to the treatment link, Roche's exploration of digital product development is also reflected in multiple application scenarios such as diagnosis and prognosis management. In this regard, the lymphoma digital prediction model and the lymphoma digital remote management tool are also important practices of Roche in incubating and co-creating local smart healthcare innovation projects.

In the diagnosis and treatment process of lymphoma, there are problems such as high aggressiveness, high heterogeneity, and high incidence of chemotherapy resistance. Taking diffuse large B-cell lymphoma (DLBCL) as an example, relevant data shows that 40% of patients will still experience recurrence after undergoing standard immunochemotherapy, and they are accompanied by high-risk prognosis problems, facing greater risks and a heavier economic burden.

In response to this, Roche, relying on its accumulation in the PET/CT algorithm field and combined with the real-world data accumulated by cooperative hospitals, has developed a DLBCL recurrence prediction model that meets the clinical needs of China. It is understood that this model mainly fuses multi-modal information to guide the discovery of new biomarkers to match lymphoma patients with a better treatment plan.

The results displayed at this CIIE show that based on the training and verification of the model with the clinical data of more than 1,400 patients in more than 400 centers, the model can quickly and automatically outline and evaluate FDG-PET images, and output quantitative and visual results for imaging doctors within 3 minutes, improving the speed and quality of DLBCL clinical image assessment.

In addition, in the long-term follow-up and monitoring process of the disease, Roche's digital products are also playing a role. Also for lymphoma, Roche has also launched a lymphoma digital patient remote management tool, which opens up the diagnosis and treatment paths inside and outside the hospital, simultaneously realizes the standardized treatment of doctors and the disease course management of patients, thereby improving the overall cure rate of lymphoma patients.

It is also understood that these products are actually only Roche's initial "trial" in the field of smart healthcare. In this process, Roche not only emphasizes "originality", but also "innovation". This also means that only on the basis of a deep understanding of the pain points and needs of domestic doctors and patients, and insisting on co-creation and co-construction with hospitals and doctors, can more products that change the health of patients and have clinical value and feasibility be developed.

Looking forward to the joint construction of industry, research, and healthcare to continuously launch more smart healthcare products

The digital solutions under the guidance of the smart healthcare concept bring unlimited imagination to future medical practice. However, from the perspective of product development, there are still many difficulties and challenges from clinical ideas to outcome transformation, such as unclear clinical value and market positioning, fragmented and low-maturity technology, insufficient productization capabilities, and lack of commercial resources.

Shen Xiafang has a deep feeling about this. She mentioned that, for example, in the research and development stage of digital products, the non-uniformity of data standards, examination methods, and time nodes for clinical efficacy evaluation in different hospitals may affect the data verification process, and in turn may affect the efficiency and accuracy of the model. Another example is that the concept of the medical system that "recognizes hardware but not software" may also lead to obstacles in the promotion of digital products in the hospital.

In order to solve these problems, in the process of the in-depth integration of basic scientific research and local innovation, "government, industry, academia, research, healthcare, and enterprises are all indispensable". In the past, under the guidance of this concept, Roche has built a complete "Digital Healthcare Incubator", aiming to achieve an efficient and seamless link from concept verification to landing through providing support for the entire industry chain. The successful practices in scenarios such as liver cancer and lymphoma have benefited from this.

At this CIIE, under the background of the new upgrade in the field of smart healthcare, Roche has also released the "Digital Healthcare Incubator 2.0", hoping to connect the stages of clinical idea verification, product development, and registration and listing to form an "integrated digital tool incubation and transformation platform to accelerate the transformation of innovative achievements".

Specifically, the 2.0 version of the digital healthcare incubator will emphasize more on the "concept verification" and "scalability and sustainability" capabilities. Shen Xiafang introduced that the former focuses on finding medical scenarios that truly have market landing value. "Doctors know the needs of patients best, but whether this need is very common and serious requires a more comprehensive perspective at the national and even global level. From this dimension, the industry side is needed to help supplement the part of concept verification, so that the AI product can really accurately target the urgent needs and transform into a greater influence."

The latter refers to the fact that the product transformation process needs to go through a series of complex links and requires the input of professional talents. To this end, Roche, in collaboration with scientific research units such as the Global Health Industry Innovation Center (GHIC) of Tsinghua Institute of Engineering, has established the "Elite Academy", hoping to attract more talents to join, and to spread the capabilities, knowledge, and standards that digital medical products should have on the transformation chain to a wider range, and truly achieve the scalability, sustainability, and systematization of digital products.

At present, the Global Health Industry Innovation Center has established an ecosystem with doctors and professors as the core, and with the collaborative empowerment of large enterprise partners, a complete set of achievement transformation models has been formed. "In the process, we will classify innovative projects. Relatively mature projects will directly enter the transformation channel, and those that are slightly lacking will first enter the concept verification center. If it is earlier, we can enter the cultivation center for further precipitation. When the scientific research achievements are transformed into products, the innovative projects gradually mature and can move forward in their own way, and the Global Health Industry Innovation Center will provide various resources to help them accelerate their development." Wang Jingjing introduced.

In addition, in order to improve the transformation efficiency of clinical achievements, Roche is also actively exploring the use of AI tools to improve scientific research efficiency, so that "scientific researchers can invest their time and energy in the source of innovation". For example, in response to the actual needs of doctors in scientific research, clinical and other work scenarios, Roche has developed functions such as literature interpretation, intelligent writing, and intelligent Q&A using generative AI technology.

At present, although these tools are mainly applied within Roche's medical department. But in the future, as the generation quality of the tools continues to improve during the actual use process, they will be gradually promoted to external users.

These explorations are also the embodiment of Roche's firm innovation and the application of AI technology to more product areas in the era of artificial intelligence technology empowering the entire healthcare industry. According to Roche, the company's existing product pipeline covers five core areas such as oncology/hematology and neurology; between 2020 and 2029, more than 20 innovative drugs will be launched globally. It can be said that as an innovation-driven veteran pharmaceutical company, Roche has sufficient motivation in exploring the issue of using AI to enhance the efficiency of existing businesses and develop digital products.

"Nowadays, the application of AI in our internal has actually penetrated into all aspects from research and development to daily work processes, and we have deeply felt the improvement and optimization of work efficiency." Xia Tian, Vice President of Finance, Strategy and Operational Optimization of Roche Pharmaceuticals China, mentioned.

At the CIIE, Roche stated that behind the practice of smart healthcare in China, the support of all sectors such as industry, academia, and research is indispensable. In the future, Roche will continue to bring more transformative innovative solutions and smart healthcare innovation practice solutions, striving to enable every Chinese patient to benefit from the medical experience brought by innovative technologies.