IRLAB presents application of deep learning on multidimensional CNS drug efficacy data

January 12, 2021

IRLAB (Nasdaq Stockholm: IRLAB A) presents results from a collaboration between the Department of Mathematical Sciences at Chalmers University of Technology, Gothenburg, the specialist artificial intelligence (AI) company Smartr and IRLAB. Based on multidimensional in vivo phenotypic dose response profiles generated by IRLAB’s ISP technology, it is found that in particular multilayer perceptron networks perform well for prediction of therapeutic indication and classification of drug candidates. The methodology further holds promise to discover new or expanded indications for both drug candidates and approved treatments within the central nervous system (CNS).

IRLAB (Nasdaq Stockholm: IRLAB A) presents results from a collaboration between the Department of Mathematical Sciences at Chalmers University of Technology, Gothenburg, the specialist artificial intelligence (AI) company Smartr and IRLAB. Based on multidimensional in vivo phenotypic dose response profiles generated by IRLAB's ISP technology, it is found that in particular multilayer perceptron networks perform well for prediction of therapeutic indication and classification of drug candidates. The methodology further holds promise to discover new or expanded indications for both drug candidates and approved treatments within the central nervous system (CNS).

“This application of AI methodology to our ISP database yielded stable results supporting the use of deep learning as a valuable addition to the machine learning methods we use in our proprietary research platform ISP, Integrative Screening Process. The ISP technology is key to the rapid and successful discovery and development of our clinical candidates mesdopetam and pirepemat. Enhancing the precision in our methodology further improves quality and contributes to increased competitive advantage for IRLAB and our drug candidates," says Dr. Susanna Holm Waters, M.D., Ph.D., Director of Biology and Biostatistics at IRLAB.

The scientific abstract, part of a master thesis in engineering mathematics & computational science, was submitted to and accepted by the Society for Neuroscience (SfN) and will be presented at one of the most prominent conferences globally, SfN Global Connectome: A virtual event.

Find the poster presentation at the event:

Authors: K. Granbom, F. Wallner, P. Svensson, S. Holm Waters, J. Kullingsjö, N. Waters, A. Andersson

Presentation time: Tuesday, January 12, 2021, at 10:00 -10:30 EST (16:00-16:30 CET)

Presenter: Klara Granbom, MSc, on behalf of IRLAB

Presentation number: P383.09

Abstract: Clinical predictions in CNS drug discovery based on in vivo systems response profiles and non-linear machine learning methodology

Session title: Techniques in Neurodegenerative and Neuropsychiatric Diseases' Research

The poster will, after its publication, be published on IRLAB's webpage under "scientific publications" (www.irlab.se/research-platform/scientific-publications).