Dr Stephen Green

Research Interests

Stephen's main interest is in Artificial Intelligence with considerable knowledge of Classic Computer Vision and Deep Learning. Stephen has worked in AI for over a decade in topics primarily in health research spanning from pain management, anaesthesia, cancer research, motor function classification, stroke detection and functional near-infrared spectroscopy. Stephen also has a background in Cryptography and Cryptanalysis which infers a lot of his current work and remains a topic he continues to research in parallel with his current interests. 

Strategic Themes

Artificial Intelligence and the way techniques can be used in real-life situations. 

Machine Learning and how its gradual integration into modern life can be eased and simplified. 

Classic Computer Vision and how to judge which problems can be solved with simpler means than using Deep Learning. 

Research Project

Digital Roads of the Future (DRF) initiative to develop a connected physical and digital road infrastructure system that is sustainable. For more information, see https://drf.eng.cam.ac.uk/research. The current project is to derive RRL2 handcrafted road asset condition forecasting & forensics maintenance processes, followed by automatically deriving stepwise optimal maintenance processes from Digital Twins for human use and subsequent robotic maintenance tasks for AMP use. 

Stephen is part of the Data Science work package for the project, which, among other things, focuses on developing Bayesian and machine learning methods for data integration of digital twins data, as well as assessing the quality of the data via uncertainty quantification. For more information, see https://drf.eng.cam.ac.uk/research/data-science. 

Biography

After completing an MMath degree at the University of Leicester, Stephen did his PhD there in Methods for Large-Scale Data Analysis & Machine Learning for Intelligent Image Processing. After this, he worked for Harvard Medical School in Boston Children’s Hospital’s anaesthesia division as a Research Fellow, leading their mathematical modelling division. This work translated ideas that had been considered for MRI scans into function Near Infra Red Spectroscopy (fNIRS) readings which had greater practicality and less intrusive than all other available measures. These fNIRS readings identified which brain cortices were activated during harmful stimuli and how surgeons can react in the operating room. 

Stephen then worked as a Research Associate (RA) at MIT’s mechanical engineering department creating a gross motor function classifier for metachromatic leukadystrophy and cerebral palsy. Parents could upload videos of their child to an Android app which would provide a fast and accurate diagnosis of what stage of the disease the child was in. Stephen also contributed to the 77 Lab’s digital twin project, creating a model excavator for the Sumitomo Corporation that could be operated remotely by a non-expert for long periods without on-site fatigue. 

Continuing his position as an RA at the Institute of Manufacturing at the University of Cambridge, he developed an algorithm that could distinguish signs of long COVID, Myeloma and strokes from a control group using non-intrusive wearable data which could diagnose patients in real-time. 

Stephen is now the Senior Research Associate for the Digital Roads Of The Future’s Data Science division and director of Artificial Intelligence and Machine Learning. He is currently handcrafting road asset condition forecasting and forensics maintenance processes, creating prototypes alongside optimising algorithms for decision-making. He is also conducting an intensive literature review of state-of-the-art processes for highway maintenance treatments for spatial modelling for the distribution of defects.