AI Meets Pharma: Overcoming Barriers to Collaboration — Highlights from the 2018 MedTech World Forum
The application of AI in the pharmaceutical industry is regarded as a major avenue for the commercialization of artificial intelligence companies, with pharmaceutical firms considered the largest potential payers.
VCBeat (WeChat ID: vcbeat) has reported that, in terms of capital inflow, new investments in the “AI + drug discovery” sector exceeded $600 million from January to July this year—surpassing the total for the entire previous year and marking an explosive surge. More than 100 companies domestically and internationally have been labeled as operating in the “AI + novel drug development” space.
AI can be applied in numerous fields, including drug discovery, preclinical research, clinical trials, rational medication decision-making, pharmacovigilance, and drug repurposing.
On September 26, at the “2018 AI + Pharmaceutical Innovation Forum,” co-hosted by VCBeat, VBInsight, and the China High-Tech Industrialization Research Association, and co-organized by the China Medical Devices Industry Association and the Zhejiang Health Industry Federation, numerous guests—including experts and scholars in healthcare, representatives from healthcare innovation enterprises, investment firms, and medical institutions—shared their insights. They jointly explored the value of AI for pharmaceutical companies and hospitals, as well as strategies to overcome challenges in AI adoption.
Speaker: Ross Rothmeieris the United StatesMedidataVice President of Technical Solutions and Innovation Laboratory
He presented new data to the audience:2006Aged2015Annual10Over the years, the success rate of new drug development using conventional methods, from Phase I to Phase II, is63%, from Phase II to Phase III is28.8%，55%It can enter the stage of submitting a new drug application for market approval, but even at this stage, only83.9%can ultimately receive approval. If these percentages are aggregated, it becomes evident that only Phase I investigational drugs8.4%Success is achievable.
However, by leveraging AI to select biomarkers during the R&D process, the success rates for the aforementioned stages can be significantly increased to 76.7%, 46.7%, 76.5%, and 94.5%, respectively. The combination of these percentages implies that the overall success rate for drugs in Phase I clinical trials can reach 25.9%, more than triple the original rate!
It is precisely this remarkable efficacy that has prompted pharmaceutical companies to forge partnerships with AI firms.
MedidataCompanyWorldwideCustomers exceeding1000Home. On its software platform,13000Multiple studies,380Ten thousand patients can provide valuable data.
In China, Medidata supports 870 clinical trials and serves 146 clients, including leading Chinese pharmaceutical companies and contract research organizations (CROs) such as Hisun Pharmaceutical, Fosun Pharma, and WuXi AppTec.
Internationally, among pharmaceutical giants,GSK, Novartis, Johnson & Johnson, andAI company Insilico Medicine has initiated a collaboration,Merck, AbbVieisAtomwiseand collaboration; pharmaceutical companies such as AstraZeneca, Pfizer, and Takeda Pharmaceutical have also entered into partnerships with artificial intelligence firms.
The permeation of artificial intelligence will reshape every industry, and the healthcare sector is no exception. It is no longer a hyped gimmick but a fact that AI and deep learning can significantly improve efficiency. Currently, the application trends and models of AI are becoming increasingly clear, and their value to various stakeholders is beginning to stand out.
From the patient’s perspective, artificial intelligence (AI) is better equipped to address their healthcare needs and deliver superior medical services, earning widespread acceptance among patients. Fan Xiaolei, Co-founder of Suowen Boshi, stated, “Patients are increasingly embracing diagnostic and treatment decisions and recommendations derived from big data and AI-driven analytics. There is growing patient acceptance of participating in treatment management processes through new technologies. Existing data indicate that our patients hold a highly open and inclusive attitude toward new technologies and the evolution of treatment-related data—a finding that greatly encourages us.”
For pharmaceutical companies, the most direct benefit of artificial intelligence lies in new drug development. Mr. Li Yunfei, Director of the Integrated Product Development (IPD) Management Office at Tasly Pharmaceutical Group, commented on the role of AI in target discovery: “Our understanding of diseases is constantly evolving, and technologies, therapeutic pathways, and treatment methods are continuously changing. Drugs under development are becoming increasingly complex, demanding more advanced technologies. Informatics, including what is now referred to as big data and AI, will undoubtedly play a significant and crucial role in the future.”
In addition, Li Yunfei stated that artificial intelligence can serve not only as a tool for new drug development but also as a means to rediscover the value of traditional Chinese medicine (TCM). For instance, network pharmacology is an integrative tool that combines pharmaco-informatics, software informatics, molecular biology, big data, and artificial intelligence. It enables the prediction of TCM targets, identification of active ingredient groups in TCM, elucidation of TCM mechanisms of action, interpretation of prescription rationality and formulation patterns, and facilitates the discovery of new indications.
Beyond pharmaceutical companies, artificial intelligence can also alleviate challenges faced by hospitals. On the hospital side, institutions at different levels have distinct needs, and AI serves as a key technology to address these issues. For secondary hospitals, establishing medical consortia helps enhance diagnostic and treatment capabilities, thereby better meeting patient care demands. With the flow of big data, AI and deep learning can more effectively replicate the clinical experience and diagnostic-treatment workflows of tertiary hospitals, empowering healthcare professionals at the grassroots level. For tertiary A-grade (Grade 3A) hospitals, where physicians have substantial research needs, AI can better leverage real-world data.
Although AI boasts broad application prospects, its practical implementation requires substantial amounts of standardized, structured data for “training.” Furthermore, stakeholders in healthcare services must also shift their mindset and proactively embrace artificial intelligence.
In terms of data, if artificial intelligence is divided into three dimensions—algorithms, computing power, and data—the industry’s primary opportunities currently lie in the data and application layers, with competition centering on the quality and quantity of data. However, for Chinese medical AI enterprises, while there is a vast amount of potential data available in the market, it cannot be effectively organized or utilized. On one hand, although Chinese hospitals generate enormous volumes of data, more than 75% is unstructured, failing to unlock the value of “big data.” On the other hand, both modeling and machine training rely heavily on real-world clinical settings, yet most medical AI products in China lack integration into such clinical environments.
Fan Xiaolei explained, “Currently, the vast amount of existing data is non-standardized and heavily influenced by physicians’ seniority and individual practice styles. Furthermore, data incompleteness and superficiality are prevalent; current Hospital Information System (HIS) big data platforms only record basic information. Critical research-oriented data—such as molecular biology test results, genetic testing data, and out-of-hospital follow-up data—are not captured within hospital systems. The significant absence of such information creates substantial challenges for data utilization. Although natural language processing (NLP) technologies are available, the value derived from investing in the processing of this type of data remains relatively low.”
Li Yishi from Haoyue Capital proposed the criteria for “good” data: “We believe that high-quality data primarily resides within high-quality hospitals. It should reflect the expertise of clinical specialists, integrated with clinical guidelines and evidence-based medicine, encompassing more comprehensive, complete, and multidimensional data, and be generated with scientific research as its primary objective. Historically, data accumulated through clinical studies has been of a relatively high caliber.”
Technology can be caught up with, but the shift in mindset is the invisible barrier. Liang Yi, Chief Commercial Officer of Zai Lab, stated bluntly: “We can already roughly envision the future, where 5G technology and the Internet of Things (IoT) enable universal connectivity. This will disrupt the business models, production, and all other aspects of the entire healthcare industry. The IoT will bring about significant changes in the future. For companies across all industries, those that do not understand the IoT or ‘soft’ capabilities, focusing solely on ‘hard’ manufacturing, are destined to be eliminated. This is particularly true for laggard pharmaceutical companies in China.”
Liang Yi added, “When launching a new product, pharmaceutical companies still adhere to traditional thinking, which focuses solely on communicating with physicians. They conduct no R&D on the entire upstream and downstream ecosystem associated with the product. R&D is not limited to product development; it also encompasses business model innovation. Even marketing departments should engage in R&D, something pharmaceutical manufacturers have failed to do for decades.”
As a tool that significantly reduces labor costs and empowers every segment of the healthcare industry, artificial intelligence (AI) has made its presence felt across various sectors of medical health. AI can help alleviate the shortage of physicians in hospitals by providing patients with more precise and efficient services. On the pharmaceutical side, for AI to meet R&D needs and address the challenge of precise patient recruitment, the primary prerequisite is the generation of substantial amounts of data. In the future, AI companies will need to collaborate with big data firms or pharmaceutical companies that have already accumulated extensive data repositories. Furthermore, achieving precise patient recruitment also requires the involvement of specialized patient organizations.
Pharmaceutical companies can no longer operate without artificial intelligence, but overcoming data challenges is an urgent priority.