Is drug development failure a technical problem?

Glen de Vries is the co-founder and co-CEO of Medidata. Medidata was founded in New York in 1999 to develop and sell software for clinical trials. In 2019, it was acquired by France Dassault Systèmes.

Big data technology has changed many industries, including the most data-sensitive drug clinical trials. For example, modeling organs and even human bodies is seen as a disruptive technology for realizing precision medicine in the future. However, Glen de Vries, co-founder of Medidata, believes that the success or failure of drug development is not entirely a technical issue. He founded Medidata in New York in 1999 to provide pharmaceutical companies with clinical trial data collection and management tools. According to his observation, the failure of project management may also prevent excellent drugs from entering the market.

In October 2019, the French industrial software giant Dassault Systèmes acquired the company for US$5.8 billion, which is Dassault’s largest acquisition since its establishment. The main business of Dassault Systèmes is to manage aircraft research and development, which seems completely unrelated to the field that Medidata is engaged in. However, in the eyes of both parties to the transaction, the work of the two companies is actually the same: doing complex events Effective management-long cycle, high investment, high risk, these three problems faced in the research and development of aircraft are also faced by the research and development of new drugs.

Yi: Many companies are collecting data on wearable devices or patient medical consumption records, calculating incidence probability, establishing social risk management models, and finally making product recommendations. What is the difference between Medidata’s data collection and business model?
V: Very different. We do not serve customers through business model innovation. Our work first is to collect evidence for drugs to verify the availability of drugs. Secondly, most of our cooperation with customers is to implement a new product, including new drugs, new compounds, new molecules and some medical devices. Relevant data is not only collected from wearable devices, but also produced from a large amount of traditional medical data, such as patients, doctors and nurses who take care of them, medical blood laboratory, imaging department, and even from updated tests, such as the results of genetic testing. We will combine various data and build models accordingly. This model is different from the business model in that it expresses which medicine is suitable for which type of patient, so that customers can develop the most accurate therapeutic drugs based on this model. We initially only collected data on “a certain specific effect of a certain disease produced by a certain drug.” After adopting the new model, we can collect data on a larger scale, which includes not only the data in the current clinical trial, but also a large amount of public data that can be obtained.

Yi: Does precision medicine need to model each patient? Does the cost increase as a result?
V: To some extent, precision medicine represents a new model of global new drug development and biotechnology development. Nowadays, new drugs basically contain a so-called decision support system during the launch process. The so-called decision support is what kind of medicine is suitable for patients; for doctors and patients, which medicine can bring the greatest benefits; and which corresponding standards can be judged to help them make decisions. In this sense, all new drugs have a system to explain how they can be most effective.

The time and economic cost for a traditional medical product from clinical trials to market launch is huge. The time is generally about 10 years, and it is normal for the economic cost to reach 1 billion US dollars. Now, we have actually helped customers successfully reduce the development time and expenditure in this area many times, get the evidence they need at a faster speed, establish a specific model for a certain patient, and determine which medicine is suitable for the patient. Realize precision medicine. In fact, what precision medicine can bring to customers is more cost savings rather than increased costs.

Yi: Is the “precision” of precision medicine accurate to the individual or just to a smaller group?
V: Vaccine is a kind of “treatment”, it is aimed at a very large population. There are also targeted anti-cancer drugs, which provide targeted therapy for certain subdivided populations. In the future, more and more precise medical treatment will appear, and the research on the development of new drugs and treatment methods will also undergo great changes. Compared with the large group of people originally targeted, medical care now pays more attention to each individual. For example, we have two types of products, one is for doctors, which are used to obtain doctors’ data; the other is for patients. These systems can be connected with patients’ smartphones and sensors to directly obtain a series of data, so as to better help clinical trials. The sponsors find the standards and the most suitable treatment methods that adapt to certain more subdivided populations.

Yi: How has the development of AI technology changed the data acquisition and modeling work that Medidata is engaged in?
V: New drug development is a very difficult area to advance. Relevant data and insights can only be obtained from subjects undergoing clinical trials. This is the complexity and challenge of new drug development. To develop a new drug in the past, it was necessary to go through various challenges. For example, it was difficult to find patients, the diseases that might be targeted were very rare, or the patients were so sick that they did not meet the enrollment requirements, it was difficult to carry out effective clinical trials. So our AI team developed SCA (synthetic control group). The principle is to integrate other clinical trial data and historical data to reduce patient enrollment in clinical trials and reduce costs at the same time. The data of the control group is real, from another clinical trial, it is not artificially synthesized. With the further development of precision medicine, the original practice of recruiting patients on a large scale will change, and the scope of drug research will become smaller and more precise. In the future, in clinical trials of developing new drugs, we can not only obtain data from real patients, but also obtain controlled data from other clinical trials, but it is also very possible to obtain data from virtual trials. These rich sources of data are so exciting. look forward to.

Yi: Dassault Systèmes and Medidata are in different fields. What is the synergy between the two?
V: In addition to providing research tools for researchers, we also do data management and analysis, which is equivalent to providing a life cycle that proves the effectiveness of drugs. After the drug is successfully listed on the market, it will enter the production, logistics, and distribution links in the future. Related digital virtual content and technology are part of the Dassault Systèmes platform that can provide services. The whole process from the emergence of the treatment concept to the final effect in the human body can be covered and included. We call this extended service process the “full life cycle management” of drugs. In fact, out of the 10 new drug developments, only two new drugs can finally reach customers, largely because there is no good lifecycle management platform to support them to the end.

Yi: To what extent is the failure of drug development a theoretical error, and to what extent is it caused by the lack of management tools?
V: Actually, there are two reasons. What affects the development of new drugs is not only the problem of the drug itself, but also a large part of the problem of project management, such as cost. There are also various factors that will make good drugs eventually find suitable subjects, especially It is the future precision medicine drug, because it only targets a very small part of the object. The failure of the drug trial is largely due to the project management. If the patient cannot be found, the drug is unsuccessful. What we can do is to help new drug developers accurately find the right patients based on our data, so as to save a lot of time and money spent on finding suitable test subjects in the traditional clinical trial process.

Yi: Medidata was founded in the 1990s. Which of the technologies, ideas, or products developed today were not available 10 years ago?
V: The problems facing human beings are not less and less, but more and more. Not only the 2020 epidemic, compared with 2019, (of course, to a large extent because of the epidemic) some people stopped going to see a doctor, or chronic diseases, or even life-threatening diseases. Pay attention. Even after the epidemic is finally over, countless people will continue to get sick. However, I am still optimistic about the future of this industry. Because people place unprecedented emphasis on medical health and life sciences. In the past twenty years or so, when my friends asked about my career-related content, they did not necessarily understand my explanation, but today almost everyone knows what a clinical trial is, and everyone knows it. What does it mean. In the past, only professionals in the industry knew that better clinical trials could be made through the analytical capabilities of big data, but they did not have a strong sense of urgency. However, now, whether it is the public or professionals, their needs and sense of urgency have increased.