Early drug-discovery at this pharma-giant will now have help from brains of a different kind. Novartis is tapping Intel in the area of deep-neural networks for adding speed to high-content-screening in the discovery process. Interestingly, this collaboration team is cutting time-to-train image-analysis models from 11 hours to 31 minutes – an improvement of greater than 20 times.

What surfaced at AI DevCon has been bubbling at Novartis since last few months. The pharma-major has been betting strongly on AI (Artificial Intelligence) for some time now with numerous AI-pilot studies underway. From the desk of global head of drug development to CEO’s chair, Vas Narasimhan has been walking the talk enthusiastically with endeavours like ‘NIBR 2.0’; more open sharing of data in early research; and greater automation in early-stage research, including AI, as per some media reports. These have been purported to catalyse Novartis’s forays in areas like CRISPR/Cas9 for radical drug discovery and development as well as in strengthening rational evidence-based decisions around clinical trials (using 10 years of company data to dig patterns, for instance).

Nerve, a software system for tracking patient data on 550 clinical trials using Novartis drugs, is a case in point. What’s encouraging to see here is how these innovators are open to the idea of learning from bad decisions for going to good decisions with machine-learning and AI capabilities. The company is also seriously embracing telemedicine, automation and quantum computing for goals like efficiency and safety.