Periodic Reporting for period 2 - FEMaLe (Finding Endometriosis using Machine Learning)
Reporting period: 2022-07-01 to 2023-12-31
The framework 'P4 Medicine' (predictive, preventative, personalized, participatory) was developed to detect and prevent disease through close monitoring, deep statistical analysis, biomarker testing, and patient health coaching to best use the limited healthcare resources and produce maximum benefit for all patients. However, we have seen only few feasible examples over the past 10 years.
The Finding Endometriosis using Machine Learning (FEMaLe) project will revitalise the concept to develop and demonstrate the Scalable Multi-Omics Platform (SMOP) that converts multi-omic person population datasets into a personalised predictive model to improve intervention along the continuum of care for people with endometriosis.
We will design, validate and implement a comprehensive model for the detection and management of people with endometriosis to facilitate shared decision making between the patient and the healthcare provider, enable the delivery of precision medicine, and drive new discoveries in endometriosis treatment to deliver novel therapies and improve quality of life for patients.
We will rely on participatory processes, advanced computer sciences, state-of-the-art technologies, and patient-shared data to deliver:
• Mobile health app for people with endometriosis.
• Clinical decision support (CDS) tools for targeted healthcare providers (risk stratification tool for general practitioners, multi-marker signature tool for gynaecologists, and non-invasive diagnostic tool for radiologist).
• Computer vision-based software tool for real time augmented reality guided surgery of endometriosis.
Health maintenance organisations (HMO) expect to be able to reduce overall cost of treatment by at least 20%, while improving patient outcomes, using CDS tools. The SMOP will be based on open protocol, embedded in all ethical and legal frameworks, to enable tailored and personalised usage to improve the lives of patients across Europe beyond the project period.
The Finding Endometriosis using Machine Learning (FEMaLe) project will revitalise the concept to develop and demonstrate the Scalable Multi-Omics Platform (SMOP) that converts multi-omic person population datasets into a personalised predictive model to improve intervention along the continuum of care for people with endometriosis.
We will design, validate and implement a comprehensive model for the detection and management of people with endometriosis to facilitate shared decision making between the patient and the healthcare provider, enable the delivery of precision medicine, and drive new discoveries in endometriosis treatment to deliver novel therapies and improve quality of life for patients.
We will rely on participatory processes, advanced computer sciences, state-of-the-art technologies, and patient-shared data to deliver:
• Mobile health app for people with endometriosis.
• Clinical decision support (CDS) tools for targeted healthcare providers (risk stratification tool for general practitioners, multi-marker signature tool for gynaecologists, and non-invasive diagnostic tool for radiologist).
• Computer vision-based software tool for real time augmented reality guided surgery of endometriosis.
Health maintenance organisations (HMO) expect to be able to reduce overall cost of treatment by at least 20%, while improving patient outcomes, using CDS tools. The SMOP will be based on open protocol, embedded in all ethical and legal frameworks, to enable tailored and personalised usage to improve the lives of patients across Europe beyond the project period.
To support impact monitoring and assessment throughout the FEMaLe project in WP1, we have co-produced impact cases and conducted constructive evaluation interviews per work package. A total of 18 impact cases and 9 follow-up interviews across 10 WPs have been completed to track and closely monitor KPIs, planned events, and progress.
We have continuously updated the FEMaLe Website with latest news about activities and achievements as well as utilised our four social media channels effectively to put endometriosis on the (media) agenda in WP9.
We have co-produced three white papers in WP2 in order to: a) discuss ethics in healthcare systems in general, explaining the existing solutions and describing current challenges as well as providing recommendations for next steps; b) gain a deeper understanding of Gender and Inclusion from an international perspective and provide an updated overview of how these factors may influence women's health and illness, with a particular focus on endometriosis; and c) to examine how to work with Open Science and Responsible Research and Innovation (RRI) in a European project and provide an overview of how this impacts the project's results.
We have published a second Danish register-based study looking into potential consequences related to diagnostic delay of endometriosis (10.1093/humrep/dead164) in WP3.
We have utilised genetic and phenotypic data from the UK Biobank to identify 295 high-risk genotype (HRC) combinations consisting of 633 unique single nucleotide polymorphisms (SNPs) in WP4. Replication was done using the Copenhagen Hospital Biobank (CHB), where 31 and 33 out of the 295 HRC (62 and 82 unique SNPs, respectively) could be replicated, depending on replication strategy. From the 33 genotype combinations, 15 genetic subtypes of endometriosis were identified with odds ratios ranging from 1.12 to 1.90. In total, 3,204 of the 4,165 endometriosis cases in CHB were assigned to at least one genetic subtype, while 961 cases did not carry any replicated HRC. The largest subtype contained 1,775 endometriosis cases. Four of the 15 genetic clusters associated with surgical endometriosis phenotypes, while nine clusters associated with at least one comorbidity. We are investigating HRGCs combinations further, aiming to leverage these for refining endometriosis subtypes.
Lucy App content has been successfully translated and validated from Hungarian into English and then Danish and Swedish in WP5. Moreover, recognising the potential to engage an even larger audience, we have taken proactive steps to translate the app into Polish, Italian, and Romanian, languages not initially planned for inclusion. We have translated Lucy App data into an Endometriosis Risk Score to facilitate shared decision making between patient and healthcare professional. At this stage, the overarching profile structure is in place, but we need to reach a consensus on the validation method and outline a clear strategy for data exchange with the Lucy App. To ensure that the validation process is aligned with our goals, it is critical to integrate accurate information into the algorithm efficiently. Preliminary data analysis has been performed.
In total during the past 18 months, we have published 4 peer-reviewed articles and performed 38 international conference presentations in the shape of 25 oral and 13 poster presentations.
We have successfully trained our models, using a dataset annotated by expert surgeons, which have accurately predicted information on new, previously unseen images in WP6.
We have proposed a pipeline for data collection and annotation in WP7, based on Delphi method, which is put in practice and has shown to reduce the inter-surgeon variability in annotation of the incision zones.
We have developed a new GDPR-compliant research platform for the MY-ENDO program in WP8, which has been tested and evaluated by researchers and mindfulness experts, and all technical issues have been solved. The content of the MY-ENDO program has been translated into English and Hungarian, and both translations have been linguistically validated.
We have refined FEMaLe's Correlate platform as our dedicated Knowledge Management platform to support communication and collaboration in WP10, integrating seamlessly other FEMale tools to optimize the flow of knowledge.
We have continuously updated the FEMaLe Website with latest news about activities and achievements as well as utilised our four social media channels effectively to put endometriosis on the (media) agenda in WP9.
We have co-produced three white papers in WP2 in order to: a) discuss ethics in healthcare systems in general, explaining the existing solutions and describing current challenges as well as providing recommendations for next steps; b) gain a deeper understanding of Gender and Inclusion from an international perspective and provide an updated overview of how these factors may influence women's health and illness, with a particular focus on endometriosis; and c) to examine how to work with Open Science and Responsible Research and Innovation (RRI) in a European project and provide an overview of how this impacts the project's results.
We have published a second Danish register-based study looking into potential consequences related to diagnostic delay of endometriosis (10.1093/humrep/dead164) in WP3.
We have utilised genetic and phenotypic data from the UK Biobank to identify 295 high-risk genotype (HRC) combinations consisting of 633 unique single nucleotide polymorphisms (SNPs) in WP4. Replication was done using the Copenhagen Hospital Biobank (CHB), where 31 and 33 out of the 295 HRC (62 and 82 unique SNPs, respectively) could be replicated, depending on replication strategy. From the 33 genotype combinations, 15 genetic subtypes of endometriosis were identified with odds ratios ranging from 1.12 to 1.90. In total, 3,204 of the 4,165 endometriosis cases in CHB were assigned to at least one genetic subtype, while 961 cases did not carry any replicated HRC. The largest subtype contained 1,775 endometriosis cases. Four of the 15 genetic clusters associated with surgical endometriosis phenotypes, while nine clusters associated with at least one comorbidity. We are investigating HRGCs combinations further, aiming to leverage these for refining endometriosis subtypes.
Lucy App content has been successfully translated and validated from Hungarian into English and then Danish and Swedish in WP5. Moreover, recognising the potential to engage an even larger audience, we have taken proactive steps to translate the app into Polish, Italian, and Romanian, languages not initially planned for inclusion. We have translated Lucy App data into an Endometriosis Risk Score to facilitate shared decision making between patient and healthcare professional. At this stage, the overarching profile structure is in place, but we need to reach a consensus on the validation method and outline a clear strategy for data exchange with the Lucy App. To ensure that the validation process is aligned with our goals, it is critical to integrate accurate information into the algorithm efficiently. Preliminary data analysis has been performed.
In total during the past 18 months, we have published 4 peer-reviewed articles and performed 38 international conference presentations in the shape of 25 oral and 13 poster presentations.
We have successfully trained our models, using a dataset annotated by expert surgeons, which have accurately predicted information on new, previously unseen images in WP6.
We have proposed a pipeline for data collection and annotation in WP7, based on Delphi method, which is put in practice and has shown to reduce the inter-surgeon variability in annotation of the incision zones.
We have developed a new GDPR-compliant research platform for the MY-ENDO program in WP8, which has been tested and evaluated by researchers and mindfulness experts, and all technical issues have been solved. The content of the MY-ENDO program has been translated into English and Hungarian, and both translations have been linguistically validated.
We have refined FEMaLe's Correlate platform as our dedicated Knowledge Management platform to support communication and collaboration in WP10, integrating seamlessly other FEMale tools to optimize the flow of knowledge.
We have achieved an impressive total organic reach of 8,703,046 views (350% increase), mainly due to popularised publications (48%), radio/TV/webinars/podcasts (29.1%), social media activities (15.3%), communication campaigns (2.8%), (public) events online and offline (2.2%), the FEMaLe website and newsletters (0.6%), and synergies (0.5%).
In Denmark, FEMaLe was granted representation for the Danish Ministry of the Interior and Health's Health Committee in March 2023, leading to the closed Expert Hearing on Endometriosis in May with solid FEMaLe participation, and culminating in the Public Hearing on Endometriosis in November 2023 with heavy FEMaLe engagement. The immediate result of FEMaLe's efforts in translating research knowledge and insights into policy is that Denmark's national support organisation for Endometriosis (Endometriose Fællesskabet) is now, partially, funded from the state budget.
Moreover, five FEMaLers were presenting at an Endometriosis Policy Event in the European Parliament later in November 2023, in close collaboration with the European Health Data Space.
In Denmark, FEMaLe was granted representation for the Danish Ministry of the Interior and Health's Health Committee in March 2023, leading to the closed Expert Hearing on Endometriosis in May with solid FEMaLe participation, and culminating in the Public Hearing on Endometriosis in November 2023 with heavy FEMaLe engagement. The immediate result of FEMaLe's efforts in translating research knowledge and insights into policy is that Denmark's national support organisation for Endometriosis (Endometriose Fællesskabet) is now, partially, funded from the state budget.
Moreover, five FEMaLers were presenting at an Endometriosis Policy Event in the European Parliament later in November 2023, in close collaboration with the European Health Data Space.