How the science of data is transforming life
This is the age of data science. The revolution in information technology that has swept the world these past decades has reached the point where there are massive amounts of data being collected and circulated continuously, powerful computational platforms capable of processing that data, and extraordinary benefits awaiting those who can make sense of it all.
Enter data science, an interdisciplinary field that uses scientific methods, processes, algorithms and systems to extract knowledge and actionable insights from data, and then apply the knowledge and insights across a broad range of application domains. Data science relies on big data — huge volumes of data that grow exponentially with time, leading to datasets so large that traditional data management tools cannot process the data efficiently, or make accurate predictions based on the complex, multidimensional, nonlinear dynamics.
This situation calls for artificial intelligence (AI) — the simulation of human intelligence in machines that are programmed to think like humans and mimic their actions, and that exhibit traits associated with a human mind, such as learning and problem-solving. Machine learning (ML), in turn, is a branch of AI based on the idea that systems can learn from data, identify patterns, and make decisions with minimal human intervention.
The way it works is that we collect more data, and by analyzing it, we identify trends and insights and make better predictions. For example, big data and AI/ML are widely used to power self-driving cars, Google Maps, ridesharing apps like Uber and Lyft, and commercial flights that rely on AI autopilot.
AI has been particularly robust for COVID-19 prediction work. For instance, my team developed a general framework for building a trustworthy, data-driven epidemiological model, consisting of a workflow that integrates such elements as data acquisition, event timelines, model development, and forecasting amid uncertainties in different scenarios.
Using this framework, we created a model to evaluate the effects of different control policies to mitigate the spread of the coronavirus pandemic. Then we applied the framework to a modified susceptible-exposed-infectious-recovered (SEIR) model in New York City. We found we could accurately predict daily new infection cases, hospitalizations, and deaths that matched the available data from the NYC website. In addition, we employed the model to study the effects of vaccination and timing of reopening indoor dining in NYC, and analyzed epidemiological models via physics-informed neural networks.
AI and ML are crucial to all domains today. For example, our Purdue Data Science Consulting Service provides hands-on consulting support for data analysis and business analytics to overcome data science challenges in research, education, and business and organization management. We will be collaborating with government agencies, national labs, and corporate partners. To date, we have filed two patents with the Purdue Research Foundation Office of Technology Commercialization — one regarding using AI/ML algorithms to predict lithium-ion battery microstructure properties, and the other involving an ML-driven contouring system for high-frequency, four-dimensional cardiac ultrasound and photoacoustic imaging.
AI/ML algorithms will continue to spur technological innovation across many fields. In manufacturing, for instance, AI/ML-powered robots will work alongside humans to perform a limited range of tasks like assembly and stacking, and sensors feeding AI-powered predictive analytics will help keep equipment running smoothly. In transportation, AI will reduce commutes via self-driving cars that lead to fewer accidents. In healthcare, AI will play a vital role in disease detection and drug discovery.
In short, AI, often involving ML, is changing the way we live — and we can expect that trend to intensify.
Guang Lin, PhD
Professor, School of Mechanical Engineering, College of Engineering; and Departments of Mathematics, Statistics (by courtesy), and Earth, Atmospheric, and Planetary Sciences (by courtesy), College of Science
Director, Data Science Consulting Service
University Faculty Scholar
Faculty Council Member, Purdue Engineering Initiative in Data and Engineering Applications
Purdue University
