COVID-19: Patient-specific lung models to guide interventions prior to clinical application
Lead Research Organisation:
Swansea University
Department Name: College of Engineering
Abstract
This project will deliver computational models of the lung, to support the development of
patient-specific treatment strategies for the COVID-19 pandemic. The models will i)
automate analysis of the damaged lung, providing additional quantitative data to support
more reliable and rapid conclusions about the presentation of the virus, ii) provide predictions
of how the lung will perform in response to different management strategies (supplemental
oxygen, mechanical ventilation, fluid balance) and potential future treatment strategies
outlined in the RECOVERY/REMAP-CAP trial (e.g. steroids, anti-inflammatories, antibiotics
and plasma from recovered patients); innovatively factoring specific parameters such as
weight, height, age, general fitness and ethnicity - which unquestionably have acute
relevance for recovery.
COVID-19 is heterogenous - affecting everyone differently. Therefore, rapid and
appropriate medical responses to individual cases are critical. Presently patients can remain
on ineffective treatment pathways for 4-6 hours before alternative treatment strategies are
employed. This project reduces waiting times, enabling prioritisation based on quantitative
tools. The models deliver heightened understanding of individual lung mechanics, enabling
clinicians to quickly make better informed treatment decisions to optimise COVID-19 survival
rates.
The model will use patient CT data, patient-specific calibration factors (age, sex, size) and
risk factors (comorbidities, clinical frailty score, exercise tolerance, APACHE-II, ethnicity),
state-of-the-art image analysis and computer simulation, in collaboration with 3DLifePrints
to build human lung models. Patient data will be accessed via ICNARC and the SAIL
databank. The model will mimic lung structure and mechanical function, accounting for the
effect of tissue damage and providing dynamic feedback of lung health.
patient-specific treatment strategies for the COVID-19 pandemic. The models will i)
automate analysis of the damaged lung, providing additional quantitative data to support
more reliable and rapid conclusions about the presentation of the virus, ii) provide predictions
of how the lung will perform in response to different management strategies (supplemental
oxygen, mechanical ventilation, fluid balance) and potential future treatment strategies
outlined in the RECOVERY/REMAP-CAP trial (e.g. steroids, anti-inflammatories, antibiotics
and plasma from recovered patients); innovatively factoring specific parameters such as
weight, height, age, general fitness and ethnicity - which unquestionably have acute
relevance for recovery.
COVID-19 is heterogenous - affecting everyone differently. Therefore, rapid and
appropriate medical responses to individual cases are critical. Presently patients can remain
on ineffective treatment pathways for 4-6 hours before alternative treatment strategies are
employed. This project reduces waiting times, enabling prioritisation based on quantitative
tools. The models deliver heightened understanding of individual lung mechanics, enabling
clinicians to quickly make better informed treatment decisions to optimise COVID-19 survival
rates.
The model will use patient CT data, patient-specific calibration factors (age, sex, size) and
risk factors (comorbidities, clinical frailty score, exercise tolerance, APACHE-II, ethnicity),
state-of-the-art image analysis and computer simulation, in collaboration with 3DLifePrints
to build human lung models. Patient data will be accessed via ICNARC and the SAIL
databank. The model will mimic lung structure and mechanical function, accounting for the
effect of tissue damage and providing dynamic feedback of lung health.
Publications
Arora H
(2024)
Lung disease characterised via synchrotron radiation micro-CT and digital volume correlation (DVC)
in TrAC Trends in Analytical Chemistry
| Description | The lungs are made up of 1000s of pathways for air to get into the lung. At the end of these airways there are small sacs (alveoli), where gas exchange occurs. One challenge in covid is the loss of integrity and volume of healthy lung for gas exchange to occur efficiently. We have shown the mechanical effects of how airway pressure is distributed around the lung in health and those effected by covid. Moreover, we have shown how these airways can become over-pressurised (to levels where the lungs will get damaged). This is a common risk/fear when using mechanical ventilators. Now we can study in detail how different patients may respond to mechanical ventilation. We have a nice visualisation tool to enable clinicians to hypothesise about what pathologies might be present and how they may effect the mechanics of different regions of the lung when subject to different mechanical treatment pathways. |
| Exploitation Route | The model framework is highly transferable to other lung diseases - therefore scope to support long term consequences of covid, as well as other common lung diseases. Work can also guide the development of new testbeds for mechanical therapies for lung health and potentially integration with other stimuli (drugs etc.). Experimental researchers are now using the model framework to design optimal printed structures to approximate the details in complex lung structures in terms of their compliance and resistance. We have been engaged to embed the model framework onto existing medical devices to expedite interpretation of measurements in clinics - pilot grants on this theme have been submitted. Work is disseminated at multiple conferences and workshops. |
| Sectors | Healthcare,Manufacturing, including Industrial Biotechology,Pharmaceuticals and Medical Biotechnology |
| URL | https://www.wcb2022.com/,https://www.compbiomed.net/2021/,https://esbiomech.org/conference/esb2023/,https://biomedeng.org/biomedeng23/ |
| Description | The project works closely with clinical staff (NHS). Awareness and information exchange between the two different sectors (computational mechanics and applied medicine) has increased. Degree of engagement and openness to pursue new opportunities is growing. In the long-term, this will be beneficial to improve chances of technology transfer. Not all interactions have progressed to further collaborative work but the volume of new engagements is increasing due to positive peer-to-peer discussions happening in hospitals as part of this project. Sustained collaboration and long-term impact is at the heart of these discussions moving forward. We are even looking at the infrastructure for ease of collaboration and integration of modelling in the NHS. The project has facilitated the potential to discuss such technology transfer, which otherwise may not have arisen, and further open doors for discussion. An example interaction involves the image processing aspect of the project. We have activities in machine learning and linking in contacts developed here, we are now obtaining large datasets for a complementary project (PhD student on non-covid work) to take on similar analysis developed in this project. This grant facilitated the pilot work to show the capability in a short time scale, which opened doors to new posibilitites on similar threads of research. |
| First Year Of Impact | 2020 |
| Sector | Healthcare |
| Description | An active lung model for novel disease insight |
| Amount | £15,609 (GBP) |
| Funding ID | 2600440 |
| Organisation | Engineering and Physical Sciences Research Council (EPSRC) |
| Sector | Public |
| Country | United Kingdom |
| Start | 09/2021 |
| End | 09/2025 |
| Title | Image-based model frameworks |
| Description | The ability to move from medical image to personalised model was delivered in this project for the lungs. Methods were also demonstrated as feasible for cardiovascular system to complement existing suites of cardiovascular model frameworks. Automation of geometry extraction (segmentation and statistical analysis) plus the use of such data to create personalised computational models was delivered. Use in future population scale work will be valuable as well as for studying other physiology e.g. animal and improving translation to human. |
| Type Of Material | Improvements to research infrastructure |
| Year Produced | 2022 |
| Provided To Others? | No |
| Impact | The core methodology is in submission for publication and has minor revisions. The second phase of automation is in draft ready for a demonstration of the techniques on a reasonably large scale i.e. 10-50 patients. Other engagements have resulted from presentation of this work including potential new applications in other pathologies and demographics. |
| Title | Model outputs generated from anonymised patient data |
| Description | The simulations run on mechanical ventilation and other forced or spontaneous breathing cycles are being populated here. The models generated from patient images will be published here but the raw images are readily available from national databases e.g. NCCID - therefore these will not be duplicated. Model algorithms and codes to create the models are to be published in open literature. This manuscript is in submission and will be linked to this database once published. |
| Type Of Material | Computer model/algorithm |
| Year Produced | 2022 |
| Provided To Others? | Yes |
| Impact | Interest from the machine learning community - via networks such as the advanced data driven engineering design CDT that combines computational models, real world data and machine learning methods. Currently local hospital data is used to generate the first datasets/results published here but by the end of the project, larger datasets derived from bigger patient databases will also be published, which would be suitable for study in more data centred approaches. |
| URL | https://zenodo.org/record/6120866#.YjHzPYnP1oY |
| Description | Respiratory and cardiovascular network collaboration |
| Organisation | Indian Institute of Technology Kanpur |
| Country | India |
| Sector | Academic/University |
| PI Contribution | We have supported the development of an international collaboration on cardiovascular and respiratory research. We are providing technical insight and expertise in advanced computational mechanics methods and machine learning. I visited IITK with 2 co-Is from this proposal to present modelling methods and other approaches to study lung and cardiovascular physiology. We have held two workshops (one online, one in person) and are inviting the IITK contingent to visit us in Swansea later this year. |
| Collaborator Contribution | IITK hosted the in person workshop and facilitated key discussions with all levels of management and researchers to enable all avenues for future collaborations to be explored. In particular, pathways from concept to commercialization were discussed as well as real-time clinical feedback opportunities for computational research. This collaboration effort will leverage key infrastructure investments ongoing at IITK in medical technology and healthcare. |
| Impact | - Two workshops (one online, one in person). - An outline for knowledge exchange through online training program and research seminars. - Grant application strategy for a joint venture in this theme of cardiovascular and respiratory health. |
| Start Year | 2022 |
| Description | Ventilator manufacturer engagement |
| Organisation | Koninklijke Philips Electronics N.V. |
| Department | Philips |
| Country | Global |
| Sector | Private |
| PI Contribution | Engaged Philips to understand the interaction of the patient with a mechanical ventilator. We have developed a simplified model of the patient bedside care on a mechanical ventilator. We have tuned compliances of the physical tubing/valves/connections typically seen to provide high confidence in the basic/classical arrangement in a hospital. This model was integrated with the multi-scale lung model to screen patient response to mechanical ventilation. |
| Collaborator Contribution | The partner provided a loan of a mechanical ventilator to enable benchmarking tests to be performed. Access to the device significantly enhanced the learning of the team into how waveforms are created and why in terms of ventilator outputs. This enhanced the interpretation of real data from hospitals, ensuring a sensible model framework is produced for the patient lung model plus ventilator. |
| Impact | Conference Abstracts at https://www.wcb2022.com/ and https://www.compbiomed.net/2021/ have been accepted to present the framework. Further abstracts related to this work have gone into ESB2023. This is a multidisciplinary collaboration that will be presented at these meetings - engineers, mathematicians, medical staff and industry linked in to support these outputs. Other outcomes are wider in terms of control systems and actuation of respiratory devices. Students are developing new devices based off this ventilator for studying other elements of respiratory mechanics - also directly following on from work packages in this grant. |
| Start Year | 2021 |
| Description | Imaging Exhibition by Oriel Science |
| Form Of Engagement Activity | Participation in an activity, workshop or similar |
| Part Of Official Scheme? | No |
| Geographic Reach | National |
| Primary Audience | Public/other audiences |
| Results and Impact | A new exhibition was awarded in the city centre focused on imaging, which is due to launch in April 2023. Images and physical models are provided for a showcase on how medical imaging, laboratory work with optics and image-based computational models are used to study the lungs and cardiovascular system. Building on the success of the last exhibit on COVID-19 where 1000s of people engaged, more interactive models are proposed here to increase our reach on this topic. |
| Year(s) Of Engagement Activity | 2023 |
| URL | https://www.orielscience.co.uk/ |
| Description | Oriel Science Exhibition |
| Form Of Engagement Activity | Participation in an activity, workshop or similar |
| Part Of Official Scheme? | No |
| Geographic Reach | Regional |
| Primary Audience | Public/other audiences |
| Results and Impact | An exhibit on COVID-19 research innovations was organised as part of a medium-long term exhibit in Swansea by Oriel Science. We applied and were granted an exhibitor space and the exhibit was due to go live to the public in the Winter. However COVID-19 restrictions meant the exhibit still hasn't opened and is now looking to go live in Easter. We have prepared a 3D printed visual model of the early work on COVID-19 image analysis supported by a presentation highlighting the workflow for modelling the disease. The focus of the presentation is the interdisciplinary nature of the work and the need for collaborative work across engineering, science and medicine to tackle such research challenges. The target audience are school visits as well as the general public but numbers will be updated once visits begin in Spring/Summer. Some of our exhibit has been posted on LinkdIn by the industrial partner and this gained significant traction, resulting in ~3000 hits within a week and requests to collaborate/support the manufacturing aspect of the project. DOI etc. will be updated when the event goes live after lockdown measures are eased. This has been enhanced with further content since the last submission. The final output was more interactive before project close and resulted in media coverage on BBC Wales to discuss the impact of the research and how the exhibit showcases this research. |
| Year(s) Of Engagement Activity | 2021 |
| URL | http://orielscience.co.uk/ |
| Description | Radio interview for BBC Wales Science Cafe |
| Form Of Engagement Activity | A press release, press conference or response to a media enquiry/interview |
| Part Of Official Scheme? | No |
| Geographic Reach | National |
| Primary Audience | Public/other audiences |
| Results and Impact | Discussed the research activity, the impact of modelling on how we understand the disease and how we are communicating the research impact via the exhibition at Oriel Science. Emphasis on multidisciplinary approaches were made to highlight the variety of challenges in this research and how the team delivered. |
| Year(s) Of Engagement Activity | 2022 |
| URL | https://www.bbc.co.uk/programmes/m001cxkg |
| Description | School visit talks (various) |
| Form Of Engagement Activity | A talk or presentation |
| Part Of Official Scheme? | No |
| Geographic Reach | Local |
| Primary Audience | Schools |
| Results and Impact | A series of webinars and in-person talks have been delivered by the Department on the role of Biomedical Engineers in society. A focus on the COVID-19 response has been driven, to highlight what Swansea University did in response to the pandemic and the multidisciplinary approaches used. This project accounted for >1/3 of the content of these talks typically, with complementary studies on viral dynamics/spread, the vaccine and mechanical ventilators having crossover for the remainder. Webinars had 20-50 students typically and in person events were restricted in numbers but 3 were held with 10-20 in the audience. The key outcomes were further discussions and questions about how much engineering is involved in these various areas of medicine. The questions also sought to discover the translation of lessons from the current pandemic, which is certainly true for the lung models presented in our study. They are highly transferrable to other diseases and so opened up discussions on wider issues affecting families as well as recreational habits (e.g. mechanics of breathing in different exercises). |
| Year(s) Of Engagement Activity | 2021 |