There’s a medical revolution headed this way. Artificial Intelligence (A.I.) is positioned to sweep the world of drug development off its feet; we just need to show it how to dance first. Predicted to save large sums of money, time, and effort, the biggest trend in clinical research right now is AI. So is the commotion justified? Read on to see how AI is expected to help treat severe and rare diseases much faster than anticipated
Normally, researching just a single proposed drug for treating illness takes around 12 years. The new drug goes through several distinct phases of discovery, from the original idea, through to creation, and testing on humans in clinical trials.
Only 33% of proposed drugs make it to the phase just before approval (Phase III). A new drug costs between $43 million to $4 billion to get to market, roughly. It takes a lot to develop, for example, a new pill that helps people with untreatable psoriasis, or a new therapy for a previously untreatable cancer.
The whole time there’s someone out there waiting patiently for an end to the pain and discomfort. Clearly, there’s a need for faster drug development, but how will AI help us?
Before Trials Begin
In the modern age of drug development doctors know they are looking for a molecule that is selective (meaning it doesn’t interact with anything in the body), non-toxic, and that works in some way to manage the symptoms of a disease. AI simulations can take a lot of research and review work out of the equation by using historical data to determine which molecules are likely to work best. Advanced forms of AI are able to understand behaviors of molecules and the context with which we intend to use them, guiding us to better decisions before we’ve even applied to start a clinical trial.
Clinical Trial Recruitment
It is well known that one of the most difficult parts of running a clinical trial is getting the volunteers required for testing. AI can look through medical data to find patients who may be receptive to a trial. It can search through social media support groups and potentially identify regions that will have more of a particular disease required for testing. This could be done by hand, but takes valuable time from already busy medical staff.
Another area of critical influence for AI here is in the design of inclusion and exclusion criteria. Traditionally, doctors have defined these criteria using their best judgement, but recent studies have shown that in many cases criteria were too restrictive. AI has shown potential to define these criteria without impacting the safety of the trial while also including more patients. This could reduce the amount of failed trials and speed up drug development.
Expanding Trials for Non-City Residents
Another great challenge with clinical research is typically low diversity of volunteers, with the standard volunteer being a healthy white person. In recent years much effort has been made to expand diversity and include those who live outside the city. This can be done through the internet or with a phone, with the volunteer self-reporting their experience. Artificial intelligence can examine pictures taken by a volunteer for quality and suggest ways to improve the pictures if necessary. This improves the quality of the clinical data and makes “decentralized” trials much more attractive.
Data Analysis and Prediction
Beyond this, AI is phenomenal at data analysis and prediction. In recent years doctors have used AI to scan and compile masses of handwritten notes and turn it into usable data. It has been able to diagnose disease and even spotted unique signs of disease that have evaded the eyes of trained clinicians, expanding our understanding of the visual cues that represent illness. Hospitals are beginning to develop disease detecting algorithms. It sounds like a lot, but it appears to be working fairly well.
Challenges Ahead
There are still significant challenges blocking the full integration of AI into the drug development pipeline; differing formats of data; lack of access to more data; training biases; the development of specific tools and contextual understanding. The industry will likely go through an AI adoption phase where leaders figure out how to effectively integrate these tools into different touchpoints of the clinical trial process.
We believe that once the adoption of AI into clinical research is complete the impact will be nothing short of massive. It’s difficult to make forward looking statements on such a topic, but it does seem clear that we could be looking at a brave new world of drug development.
