Periodic Reporting for period 3 - ENVISION (Intelligent plug-and-play digital tool for real-time surveillance of COVID-19 patients and smart decision-making in Intensive Care Units)
Período documentado: 2023-02-01 hasta 2023-07-31
ENVISION aimed to provide an artificial intelligence (AI)-based tool for intensive care unit (ICU) personnel to treat COVID-19 patients according to best practices and experiences learned from previous COVID-19 patient data from the ICU, known as the Sandman.IC.
To achieve this goal, we initiated the development of the Sandman.IC an Apple iPad-based application built upon the pre-existing Apple iPad application, Sandman.MD which was specialized for surgery wards. The plan was to collect data from 14 ICUs located in hospitals across Europe, and the Sandman.IC app was successfully installed in all planned ICUs. With the support of our health prediction tool and its alerting system, ICU staff could respond promptly to any patient's health deterioration. Furthermore, the AI-based decision support tool provided clear guidance to ICU personnel, enabling them to take appropriate measures, potentially halting patient deterioration and initiating recovery. These were the overarching objectives of the project.
However, due to the limited number of available COVID-19 patients when the systems were initially installed, we were only able to collect data from 41 patients. As a result, we gathered retrospective data from three major university hospital ICUs in Germany and the UK, amassing a total dataset of over 1500 patient records. With this extensive database, we were able to train our AI models and address multiple relevant ICU use cases.
To achieve this goal, we initiated the development of the Sandman.IC an Apple iPad-based application built upon the pre-existing Apple iPad application, Sandman.MD which was specialized for surgery wards. The plan was to collect data from 14 ICUs located in hospitals across Europe, and the Sandman.IC app was successfully installed in all planned ICUs. With the support of our health prediction tool and its alerting system, ICU staff could respond promptly to any patient's health deterioration. Furthermore, the AI-based decision support tool provided clear guidance to ICU personnel, enabling them to take appropriate measures, potentially halting patient deterioration and initiating recovery. These were the overarching objectives of the project.
However, due to the limited number of available COVID-19 patients when the systems were initially installed, we were only able to collect data from 41 patients. As a result, we gathered retrospective data from three major university hospital ICUs in Germany and the UK, amassing a total dataset of over 1500 patient records. With this extensive database, we were able to train our AI models and address multiple relevant ICU use cases.
The AI-based ECMO admission risk prediction algorithm was developed and integrated to the overall system. This prediction method can be used with the data collected by the Sandman.IC system. The performance of the method was tested. The Area Under the Receiver Operating Characteristic (AUROC) value was found to be 0.63. Although the AUROC value was not very high, it could be improved by introduction of a small number of additional variables to be collected by the Sandman.IC system.
The paper "Mechanical ventilation as a major driver of COVID-19 hospitalization costs: A costing study in a German setting" was submitted to Health Economics Review by the UMCG team in March 2023 and is currently under review to be published.
During this reporting period, the ENVISION project was represented at the Euroanaesthesia 2023 (EA23) Congress, a prestigious event for target stakeholder groups. The project team showcased their achievements through a poster and a presentation, providing an overview of the AI-based tool development, data analysis, and publications. This participation allowed ENVISION to engage with key stakeholders and gather valuable feedback and insights. In addition, the ENVISION and COVend projects consortia joined efforts to organise a joint Cluster and Policy Event at the Representation of the State of Hesse to the EU in Brussels, Belgium, bringing together policymakers, health authorities, clinicians, and researchers to review the scope and benefits of COVID-19 actions and get prepared for future pandemics while strengthening our healthcare systems on July 4th. The Highway to Health Cluster & Policy Event was attended by over 50 stakeholders, including policymakers, health authorities, clinicians and researchers. The aim was to review the scope and benefits of COVID-19 interventions, strengthen preparedness for potential future pandemics and identify ways to improve health systems. The project has laid a strong foundation for the development of advanced AI capabilities to support ICU personnel in their treatment of COVID-19 patients.
The paper "Mechanical ventilation as a major driver of COVID-19 hospitalization costs: A costing study in a German setting" was submitted to Health Economics Review by the UMCG team in March 2023 and is currently under review to be published.
During this reporting period, the ENVISION project was represented at the Euroanaesthesia 2023 (EA23) Congress, a prestigious event for target stakeholder groups. The project team showcased their achievements through a poster and a presentation, providing an overview of the AI-based tool development, data analysis, and publications. This participation allowed ENVISION to engage with key stakeholders and gather valuable feedback and insights. In addition, the ENVISION and COVend projects consortia joined efforts to organise a joint Cluster and Policy Event at the Representation of the State of Hesse to the EU in Brussels, Belgium, bringing together policymakers, health authorities, clinicians, and researchers to review the scope and benefits of COVID-19 actions and get prepared for future pandemics while strengthening our healthcare systems on July 4th. The Highway to Health Cluster & Policy Event was attended by over 50 stakeholders, including policymakers, health authorities, clinicians and researchers. The aim was to review the scope and benefits of COVID-19 interventions, strengthen preparedness for potential future pandemics and identify ways to improve health systems. The project has laid a strong foundation for the development of advanced AI capabilities to support ICU personnel in their treatment of COVID-19 patients.
Most AI-based COVID-19 development projects appeared to have focused on predicting mortality. While this was a relevant concern during ICU bed shortages, ENVISION had aimed to reduce mortality instead of merely predicting it. The use cases developed in ENVISION were designed to suggest the best-known treatment options to ICU personnel, surpassing the state of the art.
The retrospective data had been input into various types of AI models with the goal of developing the treatment decision support tools envisioned in the use cases. The initial results from sepsis prediction had shown a high level of promise in terms of accuracy. The models were planned for further refinement and integration into the commercial Sandman.IC system in the future. These decision support methods were likely to result in more COVID-19 patients surviving in the ICU or experiencing faster recoveries compared to the scenario without the project. With improved treatment options leading to reduced COVID-19 mortality, the Sandman.IC had the potential for a significant impact on the number of lives saved in Europe. Furthermore, thanks to shorter patient recovery times, shortages of ICU beds could be mitigated or even prevented.
The retrospective data had been input into various types of AI models with the goal of developing the treatment decision support tools envisioned in the use cases. The initial results from sepsis prediction had shown a high level of promise in terms of accuracy. The models were planned for further refinement and integration into the commercial Sandman.IC system in the future. These decision support methods were likely to result in more COVID-19 patients surviving in the ICU or experiencing faster recoveries compared to the scenario without the project. With improved treatment options leading to reduced COVID-19 mortality, the Sandman.IC had the potential for a significant impact on the number of lives saved in Europe. Furthermore, thanks to shorter patient recovery times, shortages of ICU beds could be mitigated or even prevented.