Bringing a new medicine to the market is surprisingly inefficient: approximately 90% of the new medications fail in clinical trials, development times are 10-15 years and costs can obtain more than $ 2 billion dollars. It is difficult to think about a more effort that needs an impulse of AI, and the technology industry, intoxicating recent advances, is immersing.
But what took us here to take us there?
History teaches us that the correct equation at the right time can change everything. Einstein E = Mc2 He helped mark the beginning of the nuclear era. Neuronal networks, with sufficient computing capacity and training data, lit the current explosion of AI. And at the end of the 90s, when it was difficult to find something on the web, Sergey Brin and Larry Page invented the Pagerank algorithm that turned Google (now alphabet) into one of the most valuable companies in the world.
Pagerank and other called “centrality algorithms” may not have the world still transforming. In fact, they can be the key to the next advance of drug discovery promoted by AI.
When applied to websites, centrality algorithms identify which pages are more left and, therefore, are more relevant to a consultation. When applied to biomedical data, they can identify the most linked answers to scientific questions, highlighting which findings have the stronger experimental support. Crucially, centrality algorithms can be applied to relatively raw data, including massive data sets generated by modern high performance approaches, so they can connect points that have never been connected before, between the data points distributed in innumerable databases and other data sources. New connections can mean new discoveries. And multi -agents AI systems are revolutionizing thesis capabilities even more than in the past.
Lots or data, very few ideas
By design, scientific publications tell stories, and only a handful of stories can fit in each article. Therefore, modern studies, with their massive data sets that accompany, leave thousands or even millions of stories incentrated. When combined with other studies, the number of unscathed stories increases, perhaps exponentially.
This is both a tragedy and a great opportunity. Some of these stories can be new strategies to cure cancer, or rare diseases, or to counteract important threats to public health. And we miss them simply because we cannot use the data that are already in our virtual hands.
A rapid setback calculation of the other dent of how many data we are talking about: a 2022 survey found approximately 6,000 publicly available biological databases. One of these databases, the gene expression Ominibus (GEO), a public repository housed by the NCBI, currently has about 8 million samples. If we assume that each sample has about 10,000 measurements (half of the approximately 20,000 genes in the human genome) we obtain approximately 80 billion measurements. Multiplying through 6,000 databases takes us to approximately 500 billion points of total data. That is without chemistry databases, patented data sources or large -scale data sets that port It has been deposited in central databases. Whatever the real number, there is no doubt that it is great and is growing fast.
The opportunity
The effective use of this data treasure could drastically increase the ability of AI approaches to offer significant biomedical advances. For example, by combining centrality algorithms with a construct called “Focal Graph”, AI agents can take advantage of these data to deliver experimentally supported findings from traceable sources. MoreOver, cuando se combinan con modelos de idiomas grandes (LLM) como el chatgpt de Openi o los enfoques de Anthrope’s Claude, los enfoques focales basados en gráficos pueden ejecutarse autónomos, generando ideas sobre los impulsores de la enfermedad y las formas potenciables o de aterrizaje hacia la treealización de atada a los que atienden a los que atienden a un aliento y el wear of the true and the landing of the true and the landing of the true and the wear of the true and the landing of the true and the wear of the realization of a breath and the wear of the true and the wear of the realization of the realization of a breath and the wear of the realization and the wear of the true and the wear of the true and the wear of the true. In fact, appreciating realingjete aea that deals and is a transpared realingtarjete aea, there is a lot of talk about a slide of the “peak of inflated expectations” in the “minimum of disappointment” of the Gartner Hype cycle. Such pronouncements are understandable, but almost surely premature. In fact, we can be on the eve of the next advance: a new combination of “ancient” algorithms that radically promises the discovery and development of new medications. Such advance is very necessary, and when using the complete amplitude of tools and notable data, it can finally be available.
Photo: MF3D, Getty Images
As an early pioneer of microarrays technology, Doug Selinger authorized some of the first publications that describe experimental and computational approaches for large -scale transcriptal analysis. After completing your Ph.D. In the Laboratory of the Church of George in Harvard, he joined the Novartis institutes for biomedical research, where his 14 -year career covered the entire drug discovery pipe, including a significant work in ID/Validation of Target, Sauring Sauring and preclinicaling and preclinicaling.
In 2017, Doug founded Plex Research to develop a novel form based on search engines algorithms. The unique Plex platform has helped the diets of biotechnology and pharmaceutical companies to accelerate
This publication appears through Medical influencers program. Anyone can publish their perspective on business and innovation in medical care in Medcity News through influential people of Medcy. Click here to find out how.