Periodic Reporting for period 1 - LungQ-Care (Facilitating personalised Lung Treatment Decisions through a Deeptech AI Clinical Decision Support System)
Reporting period: 2023-10-01 to 2024-09-30
The escalating global burden of respiratory diseases, highlighted by the World Health Organization as the third leading cause of death and disability, underscores an urgent need for innovative healthcare solutions. Chronic Obstructive Pulmonary Disease (COPD) and other pulmonary conditions continue to challenge healthcare systems, with a demand for precise and efficient diagnostic tools that far exceeds the current capacity. The advent of artificial intelligence (AI) in medical imaging presents a transformative opportunity to address these challenges. THOR, developed by Thirona, is positioned to revolutionize the lung care workflow through its advanced AI-driven clinical decision support system (CDSS).
Strategic Imperative:
Respiratory diseases not only degrade the quality of life, but also impose significant economic burdens on healthcare systems worldwide. Current diagnostic processes, particularly in imaging like chest CT scans, are hampered by time-consuming manual analyses prone to variability. This often leads to non-personalized and sub-optimal treatment plans. The introduction of THOR aims to disrupt this status quo by automating the analysis of chest CT scans, delivering rapid, accurate, and personalized medical reports to lung doctors, thereby facilitating tailored treatment plans.
Objectives of the Project:
THOR is designed to integrate seamlessly into existing hospital infrastructure. The primary objectives of THOR are to:
1. Provide lung doctors with structured, quantified reports that include disease-specific biomarkers and visual data, enabling personalized patient care.
2. Enhance the accuracy of these analyses by ensuring robustness against factors such as variation induced by disease manifestation, demographics, and scanners variability.
3. Reduce the time required for radiologists to analyse chest CT scans from an average of 15-40 minutes to under 5 minutes using THOR.
These objectives address the critical needs within the lung care value chain by significantly improving the efficiency and effectiveness of medical diagnostics and treatment planning.
Scale and Significance of Impact:
The implementation of THOR is expected to have substantial impacts across multiple dimensions:
• Clinical Efficiency: By reducing analysis time and improving report accuracy, THOR will allow for more patients to be seen and accurately diagnosed, thus potentially improving patients' quality of life through timely and appropriate treatment interventions.
• Economic Benefits: More efficient diagnostic processes will reduce the operational costs for healthcare providers and potentially decrease the financial burden on the healthcare system through more effective management of respiratory diseases.
• Strategic Healthcare Innovation: By setting a new standard for the integration of AI in clinical workflows, THOR aims to lead a shift towards more data-driven, precise, and patient-centered healthcare.
The deployment of THOR represents a strategic innovation in the management of respiratory diseases, particularly at a time when global healthcare systems face unprecedented pressures. Its ability to integrate advanced AI technology within the existing hospital workflows promises not only to enhance the efficiency of medical personnel but also to elevate the standard of care provided to patients suffering from debilitating lung diseases. As such, THOR is not merely a technological advancement; it is a necessary evolution in the continuing story of healthcare modernization, with far-reaching implications for patients, providers, and the broader medical community.
Strategic Imperative:
Respiratory diseases not only degrade the quality of life, but also impose significant economic burdens on healthcare systems worldwide. Current diagnostic processes, particularly in imaging like chest CT scans, are hampered by time-consuming manual analyses prone to variability. This often leads to non-personalized and sub-optimal treatment plans. The introduction of THOR aims to disrupt this status quo by automating the analysis of chest CT scans, delivering rapid, accurate, and personalized medical reports to lung doctors, thereby facilitating tailored treatment plans.
Objectives of the Project:
THOR is designed to integrate seamlessly into existing hospital infrastructure. The primary objectives of THOR are to:
1. Provide lung doctors with structured, quantified reports that include disease-specific biomarkers and visual data, enabling personalized patient care.
2. Enhance the accuracy of these analyses by ensuring robustness against factors such as variation induced by disease manifestation, demographics, and scanners variability.
3. Reduce the time required for radiologists to analyse chest CT scans from an average of 15-40 minutes to under 5 minutes using THOR.
These objectives address the critical needs within the lung care value chain by significantly improving the efficiency and effectiveness of medical diagnostics and treatment planning.
Scale and Significance of Impact:
The implementation of THOR is expected to have substantial impacts across multiple dimensions:
• Clinical Efficiency: By reducing analysis time and improving report accuracy, THOR will allow for more patients to be seen and accurately diagnosed, thus potentially improving patients' quality of life through timely and appropriate treatment interventions.
• Economic Benefits: More efficient diagnostic processes will reduce the operational costs for healthcare providers and potentially decrease the financial burden on the healthcare system through more effective management of respiratory diseases.
• Strategic Healthcare Innovation: By setting a new standard for the integration of AI in clinical workflows, THOR aims to lead a shift towards more data-driven, precise, and patient-centered healthcare.
The deployment of THOR represents a strategic innovation in the management of respiratory diseases, particularly at a time when global healthcare systems face unprecedented pressures. Its ability to integrate advanced AI technology within the existing hospital workflows promises not only to enhance the efficiency of medical personnel but also to elevate the standard of care provided to patients suffering from debilitating lung diseases. As such, THOR is not merely a technological advancement; it is a necessary evolution in the continuing story of healthcare modernization, with far-reaching implications for patients, providers, and the broader medical community.
Market validation through WP2 shows that THOR's stakeholder recognise benefits of adopting AI-CDSS such as improving lung CT scan workflow, automation of manual processes, addressing variability in interpretation, offering visual representations alongside quantitative data. Challenges to be overcome are financial in terms of cost and reimbursement and adoption and buy-in of AI-based solutions.
An AI-model suite for the assessment of major lung diseases has been developed. The model evaluation carried out through WP3 shows that the output of the prototype already provides a comprehensive picture for COPD patients and a largely comprehensive picture for BE and CF patients. Adding more quantification outputs could further increase comprehensiveness.
Runtime evaluations were performed against expectations in the market for the envisioned use case. Run times may already be sufficient for use in a clinical setting, despite some further optimisations have been identified and will be carried out further. In addition, to ensure robustness of the output of the algorithms, an in-depth analysis of artefacts that can induce poor image quality, impacting the reliability of the algorithm have been carried out. A few solutions have been implemented to address and mitigate such confounding factors.
To ensure integration and deployment in the clinical setting, throughout the project several activities are been carried out to containerise the AI-model suite (ensuring portability, scalability, isolation, consistency, rapid deployment, simplified updates). Furthermore, to ensure seamless integration, a cloud infrastructure has been built, to seamlessly transmit DICOMs, orchestrating analysis, automating tasks from download, AI-model run, report generation. Finally, integration endpoints for PACS have been developed for cloud-based and on-premise deployment.
An AI-model suite for the assessment of major lung diseases has been developed. The model evaluation carried out through WP3 shows that the output of the prototype already provides a comprehensive picture for COPD patients and a largely comprehensive picture for BE and CF patients. Adding more quantification outputs could further increase comprehensiveness.
Runtime evaluations were performed against expectations in the market for the envisioned use case. Run times may already be sufficient for use in a clinical setting, despite some further optimisations have been identified and will be carried out further. In addition, to ensure robustness of the output of the algorithms, an in-depth analysis of artefacts that can induce poor image quality, impacting the reliability of the algorithm have been carried out. A few solutions have been implemented to address and mitigate such confounding factors.
To ensure integration and deployment in the clinical setting, throughout the project several activities are been carried out to containerise the AI-model suite (ensuring portability, scalability, isolation, consistency, rapid deployment, simplified updates). Furthermore, to ensure seamless integration, a cloud infrastructure has been built, to seamlessly transmit DICOMs, orchestrating analysis, automating tasks from download, AI-model run, report generation. Finally, integration endpoints for PACS have been developed for cloud-based and on-premise deployment.
The development of the THOR platform will allow to transition from a fragmented AI imaging tool to end-to-end clinical decision support system with an extensive suite for the evaluation of pulmonary diseases combined with significant efficiency improvements. The project will create impact by integrating structured and automated reporting capabilities that address the specific needs of radiologists and pulmonologists. This will be available through a scalable platform that can support diverse healthcare settings and integration preferences.