Periodic Reporting for period 2 - PANCAIM (Pancreatic cancer AI for genomics and personalized Medicine)
Berichtszeitraum: 2022-07-01 bis 2023-12-31
Pancreatic ductal adenocarcinoma (PDAC) is one of the deadliest cancers worldwide, associated with a 98% loss of expected healthy life. Artificial Intelligence may help improve those outcomes. The central PANCAIM concept is to successfully exploit available genomic, imaging and clinical data to improve personalized medicine of pancreatic cancer.
PANCAIM is unique as it integrates the whole spectrum of genomics with radiomics and pathomics, the three future pillars of personalized medicine. Since the integration of these three modalities is very challenging, PANCAIM uses a simple, data-efficient, two-staged AI approach: Firstly, AI biomarkers transform the unimodal data domains into interpretable likelihoods of intermediate disease features. A second AI layer merges the biomarkers and responds with an integrated assessment of prognosis, prediction and monitoring of therapy response, to assist physicians in clinical decision-making.
PANCAIM builds on four key concepts of AI in Healthcare: data providers, clinical/domain expertise, AI developers, and MedTech companies to connect to data and bring AI to healthcare. PANCAIM partners provide eleven Pan European repositories of almost 6000 patients that are open to ongoing accrual, ensuring both data quantity and quality which are the main factors for successful AI.
SME Collective Minds builds the platform that hosts the data and provides a trustable connection to healthcare providers for data transfer. SME The Hyve ensures that cross-center clinical summary data is harmonized into a common clinical data model. Together with exposure of the data summary and access mechanisms, this renders PANCAIM fully compliant with the FAIR principles. Six Pan European academic centers provide clinical expertise across all modalities and help realize a curated, high-quality annotated data set. Finally, Siemens Healthineers provides their AI expertise and tooling to bring AI into healthcare for clinical validation and swift clinical integration in 3000 health care institutes.
PANCAIM is on track to develop a disruptive innovative solution for the use of AI in clinical decision-making for PDAC, through the following elements / specific objectives:
1. Develop the PANCAIM digital platform integrating genomics, imaging and clinical PDAC data
2. Develop and use unimodal AI biomarkers for integrative research
3. Develop and select the most promising AI-assisted clinical products integrating omics and medical imaging data
4. Implement and validate clinical products
5. Sustain the platform for further research and clinical applications
PANCAIM is unique as it integrates the whole spectrum of genomics with radiomics and pathomics, the three future pillars of personalized medicine. Since the integration of these three modalities is very challenging, PANCAIM uses a simple, data-efficient, two-staged AI approach: Firstly, AI biomarkers transform the unimodal data domains into interpretable likelihoods of intermediate disease features. A second AI layer merges the biomarkers and responds with an integrated assessment of prognosis, prediction and monitoring of therapy response, to assist physicians in clinical decision-making.
PANCAIM builds on four key concepts of AI in Healthcare: data providers, clinical/domain expertise, AI developers, and MedTech companies to connect to data and bring AI to healthcare. PANCAIM partners provide eleven Pan European repositories of almost 6000 patients that are open to ongoing accrual, ensuring both data quantity and quality which are the main factors for successful AI.
SME Collective Minds builds the platform that hosts the data and provides a trustable connection to healthcare providers for data transfer. SME The Hyve ensures that cross-center clinical summary data is harmonized into a common clinical data model. Together with exposure of the data summary and access mechanisms, this renders PANCAIM fully compliant with the FAIR principles. Six Pan European academic centers provide clinical expertise across all modalities and help realize a curated, high-quality annotated data set. Finally, Siemens Healthineers provides their AI expertise and tooling to bring AI into healthcare for clinical validation and swift clinical integration in 3000 health care institutes.
PANCAIM is on track to develop a disruptive innovative solution for the use of AI in clinical decision-making for PDAC, through the following elements / specific objectives:
1. Develop the PANCAIM digital platform integrating genomics, imaging and clinical PDAC data
2. Develop and use unimodal AI biomarkers for integrative research
3. Develop and select the most promising AI-assisted clinical products integrating omics and medical imaging data
4. Implement and validate clinical products
5. Sustain the platform for further research and clinical applications
The realized PANCAIM cloud repository features an innovative multi-modal data model, which allows us to harmonize, annotate and curate genomics, pathomics, radiomics and clinical data from 6 clinical centers. The repository is hosted compliant with GDPR and relevant other regulations. Uploads of anonymised patient data into the digital platform have progressed and a dashboard and registry are automatically updated It now contains 3000 patient data and is rapidly expanding. A framework of legal documents with digital signature possibilities has been developed that allows to rapdily onboard new partners and perform new research projects. Three external partners are connecting and over 12 research projects are running.
PANCAIM has delivered the first common multi-modal set of variables for clinical, pathology, radiology and genomic data. The AI-assisted annotation of medical imaging files is possible, cases that need more attention can be manually edited. Several AI models investigating unimodal AI biomarkers and identifying the most predictive features have been developed with promising results. Experiments regarding the integration of multimodal data are underway. Project results have been presented at several conferences and events as well as in peer-reviewed publications.
An AI algorithm for helping detect pancreas cancer on radiology CT imaging was developed, validated and implemented to show high accuracy in especially detecting small tumors which could help spot PDAC in clinical practice. Several AI algorithms predicting overall survival on unimodal genomics and pathology features are being developed and tested. We formulated and employed multimodal prognostic scores to categorize patients into high- or low-risk groups based on their projected OS time using different scores according to the information used. The utilization of the multimodal prognostic scores demonstrated superior discriminatory ability, surpassing the performance achieved solely with the stage of the tumour. The developed CT AI model has been integrated in the Teamplay platform by Siemens Healthineers and being made available to our clinical partners for validation on unseen clinical data. Discussions on how to sustain the platform for further research and clinical applications are ongoing, and several third parties have expressed interest in collaborating with PANCAIM or contributing or using the repository.
PANCAIM has delivered the first common multi-modal set of variables for clinical, pathology, radiology and genomic data. The AI-assisted annotation of medical imaging files is possible, cases that need more attention can be manually edited. Several AI models investigating unimodal AI biomarkers and identifying the most predictive features have been developed with promising results. Experiments regarding the integration of multimodal data are underway. Project results have been presented at several conferences and events as well as in peer-reviewed publications.
An AI algorithm for helping detect pancreas cancer on radiology CT imaging was developed, validated and implemented to show high accuracy in especially detecting small tumors which could help spot PDAC in clinical practice. Several AI algorithms predicting overall survival on unimodal genomics and pathology features are being developed and tested. We formulated and employed multimodal prognostic scores to categorize patients into high- or low-risk groups based on their projected OS time using different scores according to the information used. The utilization of the multimodal prognostic scores demonstrated superior discriminatory ability, surpassing the performance achieved solely with the stage of the tumour. The developed CT AI model has been integrated in the Teamplay platform by Siemens Healthineers and being made available to our clinical partners for validation on unseen clinical data. Discussions on how to sustain the platform for further research and clinical applications are ongoing, and several third parties have expressed interest in collaborating with PANCAIM or contributing or using the repository.
PANCAIM is the first to enable AI-assisted multimodal diagnosis of PDAC by pioneering integration of genomics and medical imaging from across the EU. The PANCAIM project is pooling existing -omics and medical imaging data and will surpass the expected 6k patients data from across the EU. This data is made available in a cloud repository, making PANCAIM the first to offer a high-quality data set of pancreas data combining genomics, pathology and radiology data with clinical variables. The integration of such multimodal data is starting to show novel types of analyses to provide diagnostic aid in the whole care path of pancreas cancer, making the PANCAIM repository a valuable resource for the European research community as well as clinicians.
PANCAIM partners are working on using these data to develop easily interpretable and highly transparent multimodal AI models for clinical decision support in pancreatic cancer. For instance, PANCAIM AI models pioneering the integration of genomics and medical imaging will enable AI-assisted detection of pancreatic cancer that may help reach patients at a stage where treatment is still feasible, and provide a better selection of patients that will benefit from surgery and those that will benefit from chemo(radiation)therapy.
The PANCAIM project is on schedule to deliver a better understanding of the currently under-researched pancreatic cancer and therefore improve personalized treatment of PDAC patients which will improve patients’ quality of life. In addition, PANCAIM’s effective AI solutions will be available soon to support drug development (pharmacogenetics) using novel AI models and has the potential to significantly reduce direct and indirect healthcare costs, e.g. by offering workflow improvements in the hospital through providing automated recommendations to clinicians, and better personalised care in choosing the best possible treatment for a specific patient.
PANCAIM partners are working on using these data to develop easily interpretable and highly transparent multimodal AI models for clinical decision support in pancreatic cancer. For instance, PANCAIM AI models pioneering the integration of genomics and medical imaging will enable AI-assisted detection of pancreatic cancer that may help reach patients at a stage where treatment is still feasible, and provide a better selection of patients that will benefit from surgery and those that will benefit from chemo(radiation)therapy.
The PANCAIM project is on schedule to deliver a better understanding of the currently under-researched pancreatic cancer and therefore improve personalized treatment of PDAC patients which will improve patients’ quality of life. In addition, PANCAIM’s effective AI solutions will be available soon to support drug development (pharmacogenetics) using novel AI models and has the potential to significantly reduce direct and indirect healthcare costs, e.g. by offering workflow improvements in the hospital through providing automated recommendations to clinicians, and better personalised care in choosing the best possible treatment for a specific patient.