Japanese pharmaceutical company Sumitomo Pharmaceutical And AI pharmaceutical UK company Exscientia to have joined together to launch the first phase of clinical trials of a new experimental drug using AI artificial intelligence technology on 30 January 2020. 

This new experimental drug called "DSP-1181" is a therapeutic drug developed for patients suffering from obsessive-compulsive disorder. During the development of this new drug, the developer used AI to significantly shorten the "exploratory research (research and discovery of new drugs)" period that would have taken an average of 4 and a half years, and completed the research stage within 12 months.

The so-called exploratory development refers to research to produce new pharmaceutical compounds (= compounds used as raw materials for pharmaceuticals). The production of this compound must be assumed to be a safe and easy-to-take medicine. In other words, while the pharmaceutical manufacturer actually conducts design and development, it also needs to confirm through experiments whether the compound itself can maintain the above-mentioned quality.

But why does an "exploratory development" to make predictions or recommendations from available data take an average of four and a half years? In fact, developing new drugs is a series of "trial and error". Even if a drug manufacturer has completed a drug with good efficacy, the risks will still need to be identified on a case-by-case basis if the drug itself poses safety concerns to human health. Therefore, in the process of this research, even experienced researchers also need to spend corresponding time to conduct.

Certainly, AI dramatically shortens the period of exploration and research, it wonders people that how AI can be used to create this benefit? Could the same benefit be applied to the development of COVID-19 vaccines or specific drugs?

First of all, the answer to the first question is that when dealing with huge medical information, intelligence or data, AI can effectively learn and provide highly accurate predictions and Suggestions on the efficacy, safety, and other necessary conditions of drugs. Of course, this does not mean that AI can completely replace the experience and wisdom of researchers, but at least in the efficiency of research and development can provide a sufficient benefit. During the development of the new drug compound dSP-1181, Exscientia used AI technology to prioritize the design and deployment of the compound in the exploration research.

When the research of new drugs officially enter the era of AI application, what reform must be taken to the human clinical trials of drugs go through in the future before they can be regarded as the new drugs that have been certified? It worth our attention.

Once a drug has gone through an "exploratory development", it is followed by a "pre-clinical trial" stage, which usually takes between a year and a half and three years to confirm its efficacy. When the new drug also successfully passed the "human clinical trial stage", it can enter the new drug certification stage. About the certification stage time-consuming, according to Japanese medical statistics of industrial policy institute report, "2015 approved drugs, clinical development during an average of 41.5 months (about 3 years and 5 months), when developing a medicinal product containing a new effective ingredient (NME) project would cost 54 months (about four and a half years) on average, other drugs except NME items would take about 32.2 months (about 2 years and 8 months) to develop," DSP - 1181 "which belongs to the NME drug types.

Is there any possibility of  AI application for human clinical trials of new drugs?

According to the current observation, pharmaceutical companies in various countries have begun to think about and put forward countermeasures on how to use AI technology to improve the efficiency of clinical trials. For example, AI can be used to summarize and analyze patient data and trial results, quickly select suitable patients for clinical trial candidates, and select specific patient groups that respond to drug efficacy. In addition, AI mobile applications (APPS) can also be used to observe and monitor the use of drugs, mainly in the direction of improving the accuracy of clinical trials.

To return to the second question above, can the application of AI technology fully assist the pharmaceutical manufacturers in the development of vaccines or specific drugs for COVID-19?

Unfortunately, due to the self-handicapping of AI technology at the present stage, AI cannot give full play to its advantages of rapid analysis to put forward appropriate suggestions in the case of scarce relevant intelligence information. If we can fully understand the genetic composition of COVID-19 and other relevant information, we may be able to imagine the design approach of the therapeutic drug with the help of

AI in a short time. But in fact, since we currently knows little about the Coronavirus over the worldwide, there is virtually no advantage in using AI to develop drugs.

In a word, AI technology that can greatly shorten the period of new drug development and indeed bring great help to the development of medicine, and this technology is expected to continue applying to the research and development and production of new drugs in the future, and can be used more effectively and fully.

Back to the top