The role of Artificial Intelligence in the pharmaceutical industry.
Can machines think? This question that Alan Turing was concerned about 70 years ago gave scientists a nudge to invent a perfect robot with human thinking.
Today, when Artificial Intelligence has come into play, the great Turing’s hypotheses are becoming a reality.
AI has tremendously transformed the pharmaceutical industry; particularly, healthcare software development services.
How can we develop new medicines with the help of AI? How promising are these technologies?
The two tasks of AI in the pharmaceutical industry
AI helps us drive cars, make purchases, and choose music, and also protects us against cyber attacks and performs many other functions.
However, the above technology has made the greatest sensation in the pharmaceutical industry, where scientists are using it to develop new methods of treatment and more advanced medicines.
Abraham Gilbert, the Head of the Machine Learning Department at Brainpool AI, notes that AI is transforming the pharmaceutical industry by shortening the time for the development of medicines.
The traditional way of finding the right compounds is too complex, and AI helps people perform this task in a much more efficient way.
AI has two key missions in the pharmaceutical industry.
Firstly, it helps scientists find new treatments for complex diseases and develop better medicines.
Secondly, it makes the process of developing new medicines as efficient as possible by carefully selecting candidate molecules at the earliest stages of the research, which extremely reduces the development costs.
This sounds like science fiction – how can an algorithm solve such issues?
Invention of new medicines
In the course of numerous laboratory tests, scientists examine from five to ten thousand chemical compounds, the best of which are further optimized.
Only four or five potential medicines get to the stage of clinical trials, with just one substance being approved and becoming available to doctors and patients.
Previously, studying possible candidate molecules and their combinations took a lot of time and money because each compound had to be examined in vitro and tested at least on animals.
This constrained the process of developing medicines and increased their cost.
Modern technologies can solve these issues.
AI-based systems test the physical and chemical properties of molecules under computer simulations and improve the ingredients of medicines, minimizing potential side effects.
Virtual screening helps to select suitable molecules for further testing.
A human would need to make billions of decisions to do that. It usually takes years for scientists to perform these kinds of research and testing.
AI helps us analyze data, speeding up the process and minimizing associated costs. Statistics say the cost of bringing a new medicine to market is about $2.6 billion.
Reducing the cost of developing new medicines is crucial not only for businesses but also for society.
Thus, patients with rare diseases are more likely to be cured, as medicines for them can’t be mass-produced and hence are not widely affordable now.
According to Steve Oliver, Professor of Systems Biology and Biochemistry at the University of Cambridge, AI helps us test more compounds with improved accuracy and reproducibility. That is why pharmaceutical giants have cooperation programs with healthcare software development companies.
For example, Pfizer uses the IBM Watson system to invent immuno-oncology medicines.
By using the Exscientia platform, Sanofi studies treatments for metabolic diseases. Genentech partners with GNS Healthcare on cancer treatments.
Approval of medicines
Another challenge hampering the pharmaceutical industry is the exclusion of medicines at the stage of clinical trials performed on people.
At this point, the drug candidate has already gone a long way from a molecule to a chemically pure drug that has been tested on animals.
Thus, it may turn out that years of work (sometimes it takes up to ten years to develop a medicine) and invested funds are wasted.
According to the Massachusetts Institute of Technology, only 13.8% of medicines successfully pass clinical trials.
Companies can use AI to determine the effectiveness of a particular medicine before this stage to prevent later failures.
For example, Aladdin, the UK medical technology company, launched an AI platform called AIDD in October last year.
The platform is a belt system aimed at accelerating the development of medicines that includes all stages of work: virtual viewing, lead generation, and lead optimization.
Its developers say that their solution can shorten the non-clinical stage from ten to three or six years.
Examples of successful developments made with AI
NLP and artificial neural networks help AI perceive the world, analyze data, and generate text and speech.
These capabilities of AI have sparked a wave of drug development projects based on finding patterns in large amounts of data.
Below are examples of some successful startups.
- BERG’s experience in fighting cancer.
Scientists at BERG, the US biotechnology company, have found a method for detecting previously unknown mechanisms of cancer. To do this, they tested about a thousand diseased and healthy human cells.
They mimicked cancer cells while altering sugar and oxygen levels and studied their lipid, metabolic, enzymatic, and protein profiles.
Scientists used an AI platform that generates and analyzes a vast amount of biological information and data on the results obtained.
Thus, specialists are trying to find the core differences between diseased and healthy cells and identify potential treatments.
This search method is unique, as researchers collect patient-centered biological data instead of turning to standard trial-and-error searches.
By using this innovative approach, BERG’s team figured out what molecules are crucial for cancer metabolism and developed a drug called BPM31510.
The company is also using the AI-powered platform to improve treatments for people with diabetes and Parkinson’s disease.
- BenevolentBio and treatment for sclerosis.
UK-based BenevolentBio also uses healthcare software solutions to discover medicines. The researchers upload data from dedicated papers, patents, clinical trials, and case records to a special platform. Thus, AI links a medical condition to the genes or compounds that affect it.
The system was tasked with developing new treatments for amyotrophic lateral sclerosis (ALS). AI found about 100 potential compounds, and scientists selected five of them to test on patient cells.
Four compounds showed promising results.
- Moderna’s COVID-19 vaccine.
Moderna is working on their COVID-19 vaccine, using Machine Learning algorithms and cloud-based technologies, and on the new type of messenger RNA medicines.
Their researchers are focusing on messenger RNA – an information-based molecule that extracts genetic code from DNA. The human body produces billions of messenger RNAs every day.
Scientists send mRNA with instructions to cells that produce certain proteins able to cure or prevent diseases.
Having obtained the genetic sequence of the virus, scientists identified proteins on its surface.
They use the vaccine to send instructions through messenger RNA to produce proteins that have exactly the same signature as the virus. Thus, your immune system can respond.
- NVIDIA, AstraZeneca, and a new model of medicine development
NVIDIA, an IT giant, combined the efforts of AstraZeneca and the University of Florida to find new AI-powered models for developing medicines.
The result of this joint work is a new technology called MegaMoIBART, which works on neural network transducers and is focused on reaction prediction, molecular optimization, and generation of new molecules.
The program is currently undergoing training. The first stage includes the study of large unlabeled data sets, and the second stage includes familiarization with a smaller amount of labeled data to solve specific narrow tasks.
The platform is trained on a dataset of the chemical composition using NVIDIA DGX SuperPOD.
Conclusion
According to estimates, by 2030, more laboratories will turn to healthcare software development services, the development of medicines and the testing stage will be shortened from five years to several months, and the development will cost less.
Today, AI systems form the backbone of precision medicine and provide scientists with the means to combat diseases.
AI and Machine Learning are ushering in the era of fast, cost-effective, and efficient ways to develop safe medicines.
Perhaps, in the near future, these technologies will be able to save people from currently incurable diseases.