From the Nobel prize winner in Economics Robert Solow comes this quote: “You can see the consequences of the computerization everywhere but in the productivity statistics.” This also applies to the pharmaceutical industry. According to the consulting firm Deloitte, the productivity of their research activity. Thus, the cost for the development of a new drug – including the cost of failures – have, since 2010, almost doubled to 2.2 billion dollars.
The large corporations rely on Big Data and artificial intelligence in order to accelerate the research of new drugs. Example of Novartis : The Basel pharmaceutical giant has launched the project for Data42. Here medical professionals and data engineers work together, the existing Data from several decades research according to new interrelationships between these factors to sift through.
25 characteristics are “male”
project Manager Achim Plückebaum and his Team of around 150 employees are first in front of a huge mountain of data Alone, the existing clinical data from the research and development comprise 20 petabytes. This corresponds to around 40’000 years of music on an MP3 Player. “In a first step, we need to pull the data together, harmonize and correlate,” says Plückebaum. “There are alone for the Variable “sex” 20 to 25 different characteristics.”
We have to teach the machine that “m” means the Same as “male”.Achim Plückebaum, Novartis,
Sometimes there is only “m” for male, sometimes “times” in English written out. “We have to teach the machine that both characteristics mean the Same to interpret the data correctly, and to make them comparable,” says Novartis Manager.
Is managed this step, analyzes the data using artificial intelligence to new questions. In this way, Novartis wants to find new active ingredients or new areas of application for existing drugs to discover.
New effect for an old medium
example: Currently, the Team is trying, from the data of the psoriasis-bestseller Cosentyx an indication extension to derive, so to find out, whether the funds can also be used in other diseases. The experts look for by means of new associations to the Point that Cosentyx could be in rheumatoid Arthritis effectively.
This used to work next to the studies, Cosentyx also other research, such as that for heart medication, where diseases of the skin occur as a side effects. May be evidence for the Thesis that it is find.The analysis of evidence should provide that Cosentyx is also effective in rheumatoid Arthritis, waving Novartis more revenue.
U.S. tech companies are part of the game
Currently, there is a real Hype about automated data analysis to the pharmaceutical research. Also US tech giants like Google or Apple are working on it. The consulting firm Everest Group, believes that investments in artificial intelligence technologies in the healthcare sector are expected to increase from $ 1.5 billion in the year 2017, to over $ 6 billion in the coming year.
Gisbert Schneider, Professor and head of the Department of Computer-Assisted Drug Design, ETH, warns, however, against the euphoria. The automated analysis of imaging systems such as x – ray or mammography is at an advanced stage. “This procedure but not so simple on the pharmaceutical research, because we know only rarely, all of the relevant contexts. We often know too little about the molecular causes of diseases”, says the expert.
the question it comes to
Before the data could bring a new personalized compounds is shown, should find the analysis of the right biomarkers. A Biomarker is a biological characteristic that can be measured, for example, in blood or tissue samples, and evaluated, and the pathological changes showing. An example: The blood sugar level is a Biomarker for Diabetes. Automated data analysis could help to find new biomarkers, explains the ETH Professor Schneider.
A big challenge is in programming the search algorithms, the data for the right questions to sift through. Anyone who asks the wrong question, you get nonsensical answers. Therefore, the Novartis Team Data42 means, based on the Science-Fiction spoof “the hitchhiker’s guide to the galaxy” by Douglas Adams.
Who can shorten the cycles of Learning that is at the front.Gisbert Schneider, Professor for computer-assisted drug Design at ETH Zurich
In the history calculated by the Supercomputer Deep Thought to a response to the “question of the meaning of life, the universe, and all the Rest”. After Thousands of years of Deep Thought spits out a stunning answer: 42. “The number in the name of the project constantly reminds us that we need to understand the question in order to get by means of artificial intelligence and super computers also actionable responses,” says team leader Plückebaum.
“I expect that the data analysis can assist with new methods of artificial intelligence, the pharmaceutical industry, so, for example, early on the right drug candidates with better properties,” says ETH-expert Schneider. Thus, fewer drug candidates fail in the clinical Phase, which is really expensive. “Who here can shorten the learning cycles, which is at the front.”
Created: 09.12.2019, 22:24 PM