Research as a way to understand and respond to Covid-19 and its impacts
Building resilience projects
Game theory-based approach to modelling optimal social distancing
With the Covid-19 crisis, social distancing has become a critical element of our lives. In this regard, the physical settings of the shared spaces - ranging from hospitals, healthcare clinics, public transport systems to meeting rooms - require significant re-planning. Due to the unpredictable nature by which the pandemic has been unfolding, the standard operating procedures also do change, and the protocols for physical interaction require continuous reconsideration. To overcome this challenge, we have developed a design optimisation methodology which takes the dimensions, as well as the constraints and other necessary requirements of a given physical space to yield optimal redesign solutions on the go.
Our porotype software, based on the optimisation algorithms, takes the input requirements on the social distancing criteria between people and the imposed constraints on the physical spaces such as the position of doors, windows, walkways and the variables related to the indoor airflow pattern. Thus, given the dimensions of a physical space and other essential requirements, the solution resulting from the automated optimisation algorithm can suggest an optimal set of redesign solutions from which a user can pick the most feasible option. We have been able to demonstrate the practical applicability of our prototype system by the way of examples.
QualDash for COVID-19
QualDash is a web-based dashboard for exploring National Clinical Audit data for the purposes of quality improvement, developed as part of an NIHR funded research project and introduced into five NHS hospitals. In light of COVID-19, some National Clinical Audits encouraged hospitals to upload data on a daily or weekly basis to enable close monitoring of patient outcomes in areas such as cardiology. This led us to explore new requirements that have emerged as part of the NHS’s response to the pandemic. We progressed the QualDash software into a new re-design and development iteration to further adapt the tool to specific service monitoring needs during this time of crisis.
A post COVID-19 recovery plan for small and micro businesses and entrepreneurs in Yorkshire
The COVID-19 outbreak has had a devastating impact, none more so than on small and medium-sized firms and the true impact of COVID-19 remains to be seen. Recognising this, this research aims to investigate the impact of and response to COVID-19 in small and micro businesses and entrepreneurs in Yorkshire and thereby support plans for local economic recovery post COVID-19.
A survey of SMEs, Microbusiness and Entrepreneurs (with over 650 responses) and nine focus groups with over 70 participants from SMEs and other stakeholders have been conducted and this work will help to model the socio-economic impact of COVID-19 on small business at local and regional levels, contribute to recovery plans, identify the support needed by these businesses post the current crisis and impact positively on the success of Yorkshire businesses in recovering from COVID-19.
Visualising risk factors for COVID-19: A web-enabled tool for feature engineering
Understanding risk factors of COVID-19 has quickly become a global demand to support decision making at an individual as well as national and international scales. Numerous hypotheses exist on pre-existing conditions that lead to different responses to the disease and diverse patient outcomes. Web-enabled collaborative platforms are needed to empower stakeholders to test hypotheses and share insights on disease comorbidities. In this project we explore the design and development of a tool called VisualREF, a web-based tool that guides iterative collaborative engineering of disease risk factors and comorbidities. The objectives of this project are to:
- Encode an ontology of disease families to keep the number of disease comorbidities at a human-perceivable level,
- Build a model that maps the ontology to stakeholder-defined tasks,
- Develop VisualREF a web-tool that offers a front-end for analysis, and allows experimenters to explore co-existing conditions while storing results of their exploration and sharing them with a wider community of researchers and practitioners to support collaboration.
COVID-19 Advocatus Diaboli: a responsible AI perspective for reliability ranking of data resources relevant to the COVID-19 pandemic
Daniel Neagu, PB Anand, Amr Abdullatif, Krzysztof Poterlowicz, Dermot Bolton, Kalyani Mambakkam (RA)
Also team of 5 Erasmus+ students from ENSEEIH Toulouse, University of ABC (Sao Paolo, Brasil), Plekhanov University (Russia), University of Cairo (Egypt), and 2 University of Bradford BSc and MSc student volunteers)
This interdisciplinary research prototypes a transparent big data quality ranking methodology for evaluation of highly visible public data resources, relevant data-driven computational models, and their public perception, that critically reflect reliability, consistency, flaws, contradictions and misinterpretations of Covid-19 records. The team carried an extensive review assessing resources, models and social media with an inclusive Responsible AI perspective of complementarity for data, models and their perception features during pandemic times, recorded in a an open-source code and benchmark data set, technical report and new research and knowledge transfer outputs.
QReLU and m-QReLU: A novel quantum-inspired paradigm for deep learning to aid diagnostics from medical images
Luca Parisi, Daniel Neagu, Renfei Ma (The Chinese University of Hong Kong - Shenzhen) and Felician Campean
Dr Luca Parisi is also affiliated with the University of Auckland Rehabilitative Technologies Association (UARTA), New Zealand (NZ), and Parkinson's UK. Dr Renfei Ma is also affiliated with UARTA, NZ.
This research contributed to significant improvement of the accuracy and reliability of diagnoses of COVID-19 and Parkinson's Disease from medical images by solving the long-standing 'dying ReLU' problem, introducing and extensively validating a new quantum-inspired paradigm. With codes made freely available in TensorFlow and Keras on GitHub, and a preprint uploaded on arXiv, this approach can be expected to have an impact on the deep learning community worldwide, also paving the way for impact on application domains within medical imaging and beyond.