Turing AI Fellowship: Machine Learning Foundations of Digital Twins

A digital twin is a computer model of a real-world object—like a bridge, a power plant, or a patient’s heart—that stays constantly updated with live sensor data from its physical counterpart. This research aims to build the machine learning and artificial intelligence foundations needed to make digital twins far more powerful than they are today. Current digital twins are often static or limited in how they learn from data. The core problem is that we lack the fundamental algorithms to let a digital twin autonomously update itself, test "what-if" scenarios in real time, and feed those insights back to improve the physical system. Without this, decision-makers cannot reliably simulate the consequences of a change before making it. If the programme succeeds, it will transform how we manage complex systems that quietly underpin modern life. A city’s water treatment plant could test the effect of a sudden drought on supply before it happens. A hospital could simulate how a patient’s condition might respond to different treatments. A manufacturing line could predict a breakdown hours in advance. The work is primarily fundamental science—establishing the mathematical and computational principles for next-generation digital twins—but its practical applications could eventually touch nearly every engineered system that society depends on.