How Numerate’s AI tackles translational questions with tiny data sets
With three new partnerships formed since June, Numerate Inc. is picking up steam and demonstrating that industry and academic groups alike are finding utility in its AI platform, which can build robust predictive models for preclinical development using the tiniest of data sets.
According to President and CEO Guido Lanza, the primary feature that sets Numerate apart from other artificial intelligence (AI) companies is that it is not dependent on accessing large mountains of data.
“This ability to work with very, very small data sets -- we’ve not seen anybody else crack that,” said Lanza.
That focus on small data sets stands out from the trend among AI companies in drug development, which are mining large databanks for various purposes. For example, the WuXi NextCode Genomics Inc. subsidiary of New WuXi Life Science Ltd. told BioCentury in June the differentiating feature of its system is that it has been trained on arguably the largest data set amassed by any AI company.
According to CTO Brandon Allgood, after ten years in business, Numerate’s in-house database contains data on 2,000-2,500 targets and over 10 million compounds against those targets. “I would not call this big data; I generally