navigation

3 Ways Machine Learning is Changing the Pharma Industry

3 Ways Machine Learning is Changing the Pharma Industry

by
July 14, 2021

The Pharma Industry is developing rapidly nowadays. Artificial Intelligence (AI), Machine Learning (ML) and Big Data all have crucial roles in this. Now, let us see 3 of the fields that ML has a huge impact on the digital transformation of the Pharma Industry.

Drug Discovery

One major progress that ML has made is in the field of drug discovery. This is really exciting because the process of discovery and development becomes shorter, cheaper, and more effective. The reason lies in the usage of massive and high-quality data and the improvements of the algorithms. Let’s explore some instances of applying machine learning algorithms in the process of drug discovery. 

One of the very first steps is to analyze the genomic DNA sequence. Several ML algorithms are involved in the DNA sequence data. One of them is sequence classification. It is used to predict the identifications of genes in the DNA molecules. In oncology, to sequence, a patient’s tumour used to cost around 10 million dollars. Today it has dropped down to hundreds and the tendency is to keep getting cheaper. These results are very valuable for AI and the pharmaceutical industry.

Here, we can see a graphical representation provided by NHGSC – National Human Genome Sequencing Consortium. They are tracking the falling costs per genome through the years: compared to Moore’s Law, which predicts that every 2 years the computing power doubles.

A great example of how Machine Learning is changing the Pharma industry significantly is AlphaFold. This is a piece of software developed by Google’s team DeepMind. Its mission is to predict the structure of the protein with high accuracy. This is important because predicting its 3D structure defines the protein’s functions and role. The latest version of the software showed amazing results last year. This breakthrough will help scientists understand diseases and design drugs easier and faster. It is said that AlphaFold is a solution to a 50-year-old major biology challenge.

Clinical Trials

Another field in the Pharma Industry that Machine Learning is making changes in is Research and Development (R&D). Most specifically, we will focus on the Clinical Trials. Clinical trials are research studies that investigate if a new treatment is safe and effective for humans. They observe both focus groups of patients and healthy people to observe how the new treatment or drug will affect them. The goal is to have tolerable side effects or none at all. 

There are a few steps that should be done for clinical research to begin and the first one is submitting an application IND (Investigational New Drug) to the FDA (Food and Drug Administration).  After that, the clinical trial starts. It contains 4 phases that need to be passed successfully and with positive results of the research. They usually take around 10 years and involve more than 3000 patients. The whole process from drug discovery to the marketplace is difficult and requires lots of time and financial resources. To shorten the time to market (TTM) and decrease operational costs, many pharmaceutical companies are turning to AI and ML. Artificial Intelligence is mostly used in the discovery stage, which happens from medicine discovery to R & D, for instance, in finding new disease applications for already existing drugs.

Machine Learning makes it possible to analyse and interpret data faster than ever before, which is why it is really useful for clinical trials. Imagine how much data is produced during all of the aforementioned phases. And how much time and money is needed to process and analyse it. It is also important to monitor all of the patients during all of the stages and make sure that they are running the experiment in the same way. Thus, the investigator and expert committee design special Clinical Trial Protocols with essential instructions on what and how all the required steps should be done. It is distributed to all of the participants in the trial. Machine Learning is shortening the process of analysing the entire documentation of all of the people involved in the experiment. Q2 Lab Solutions – a laboratory service organisation for clinical trials – uses AWS ML tools to do this. Using Amazon Textract for converting the pdf protocol to raw text. Amazon Comprehend Medical is a natural language processing (NLP) service that is extracting medical and clinical terms. They extract only the lab tests from the documents that the lab needs to run for a clinical trial. 

Epidemic Outbreak Predictions

Now let’s focus on the topic that touched everyone’s life for more than a year and a half. The Covid-19 pandemic shifted our lives drastically. But can we learn to predict a future virus outbreak or prevent it from reaching pandemic levels? When will it be possible for ML to detect if a new disease will emerge so that we can prepare and take precautions? BlueDot is a software company that uses AI, ML, and Big Data to forecast outbreaks and the spread of infectious diseases. The data they gather comes from media articles, news reports, and satellites. Then, specialists perform data validations before public report releases. 

BlueDot is the first company that published a scientific paper for the detection of the Coronavirus. That happened on the 30th of December in 2019 when “a report of a cluster of pneumonia of unknown aetiology was published on ProMED-mail, possibly related to contact with a seafood market in Wuhan, China”. This is actually not the first outbreak prediction that BlueDot has discovered. In 2014 they detected that the Ebola virus could leave West Africa and in 2016 predicted the possible spread of the Zika virus in Florida.

ProMED-mail, mentioned above, stands behind the Program for Monitoring Emerging Diseases. The system performs a global reporting of infectious diseases that affect humans, animals, and plants. The ML algorithms that are specified in the outbreak predictions should work with the latest data to be more accurate. This is why Internet service like ProMED-mail is very useful. There are a lot of factors that need to be considered for improving the precision of the algorithms, such as climate, climatic events, the lifestyle of countries, and population. 

Predicting epidemic outbreaks using ML could save plenty of lives and reduce the economic loss they are causing.

It is amazing how we could use ML and Big Data in our favour. Tons of data could be analysed and the most valuable of it to be extracted with no human interference through all of the steps. Still, let us not forget that one AI system is powered by quality data. And we can definitely acclaim its inseparable and invaluable part in Pharma’s evolution. 

Do you want more great blogs like this?

Subscribe for Dreamix Blog now!