Expert Intelligence (EI) announced, at the 2023 American Society for Mass Spectrometry conference in Houston, Texas, its AI App Builder for scientific analysis. The EI App Builder is a fully automated platform that replaces the need for AI engineering to create AI applications for Mass Spec and other types of scientific analysis. EI AI Applications augment existing software analysis tools by giving scientists a new method to look inside complex and noisy signal data with an easy-to-use, visual AI App Builder.
"AI has become the "go to" method for extracting hidden patterns in data or creating human-like interactions. To extend AI into scientific analysis we've designed an entirely new way of building AI models with Ultra Low Data AI technology that removes complexity and cost.", said Lalin Theverapperuma, Ph.D, CEO of Expert Intelligence. "The Expert Intelligence App Builder opens up AI to any scientist by eliminating the need for massive datasets, deep AI knowledge, or programming tools. Any scientific expert can quickly and easily create AI models solely with their domain knowledge."
Ultra Low Data AI App Builder to Create AI Models for Science
Mass Spectrometry and other analysis equipment are widely used for discovering the components of a material across a wide range of industries including drug manufacturing, industrial process control, or anti-counterfeiting.
EI's AI App Builder uses AI modules to replace the discrete steps in traditional AI engineering. The EI App Builder supports common equipment data formats and directly imports data into a Data Visualizer. The scientific expert simply identifies data regions of interest in the AI-assisted "Data Canvas" and launches the AI App Builder. AI modules behind the scene, create and test the AI model - no AI engineering required.
To automate the entire AI pipeline, EI developed new AI technology to create models with Ultra Low Data and train the network with reinforcement learning techniques. Unlike traditional AI techniques that require massive data sets, Ultra Low Data AI can be trained with as few as 50 data points, opening AI to virtually any analysis.
To ensure ongoing quality of the AI inference results, EI created the EI Inference App to host the AI models. The EI Inference App continuously monitors the quality of the AI results to ensure inference accuracy.
A New Research Tool for Scientific Analysis
The new AI framework is designed for scientists such as chemists, biochemists, and materials scientists to directly create their own AI models for analysis.
Examples of initial projects include a Mass Spectrometer-based breathalyzer from Breathtech for Covid detection and identification of e-cigarette Brand and Flavor Classification with 99% and 98% accuracy, respectively. The EI App Builder has also been used to successfully identify blood cancer cells (CTCs vs WBCs) in a research environment.
Training in the Cloud With Cloud Portal or Local App Deployment
EI's App Builder runs in the AWS cloud as a SaaS service. The Expert Intelligence framework removes the complexities of cloud storage and managing cloud computing resources. From the latest GPUs to inexpensive CPUs, the virtually unlimited capability of the AWS cloud provides the capacity to handle any size project.
Once a model is ready to release, the AI model is accessed through either the EI Inference Portal or, optionally, exported into the locally-hosted EI Inference App.