Periodic Reporting for period 2 - REBECCA (REsearch on BrEast Cancer induced chronic conditions supported by Causal Analysis of multi-source data)
Reporting period: 2022-10-01 to 2024-03-31
The REBECCA project aspires to tap into the potential of Real-World Data (RWD) for supporting groundbreaking clinical research on complex chronic conditions (CCC) as a complement to Randomised Controlled Trials (RCT). REBECCA moves beyond the analysis of Electronic health records (EHR), by combining it with detailed monitoring data from multiple wearable, online behaviour and registry data to monitor patients’ functional, emotional and Quality of Life trajectories, with high temporal granularity. REBECCA also targets explainable causal modelling combined with deep learning to account for observed and latent confounders in RWD analysis. The project focuses on the complex array of chronic comorbidities developed during breast cancer recovery, in particular studying the impact of primary and adjuvant cancer treatment on patients’ quality of life and assessing the value of detailed patient monitoring as a means for improved patient care, but it also intends to demonstrate the extensibility of REBECCA to other forms of cancer. To this end, a total of seven studies in Sweden, Norway and Spain will produce new knowledge on clinical management of cancer patients that will shape future guidelines and practices for post-cancer treatments. Best practices resulting from the REBECCA studies will be disseminated to researchers, public health and regulatory bodies throughout Europe to facilitate wider adoption of RWD in clinical research. In addition, the REBECCA platform, capable of detailed monitoring and privacy preserving federated cross-country data analysis, aims to provide an infrastructure for continued progress on use of RWD beyond the end of the project. Through these activities, REBECCA aims at the mass adoption of RWD for understanding CCCs and ultimately at establishing RWD as a valuable clinical research and patient management tool.
Since the beginning of the project, research work focuses on five interrelated domains:
1. Design and development of the necessary Real World Data collection infrastructure:
It consists of (i) the REBECCA Patient mobile app (Android and iOS), which includes integration with Garmin wearables, (ii) the REBECCA Companion app, (iii) a web browser plugin for monitoring online activity, (iv) REBECCA Electronic Case Report Forms for capturing EHR data, (v) backend servers for storage, processing and analytics, (vi) the Clinical Dashboard. This complex infrastructure has already been launched and recently updated to its latest version, REBECCA V3. It supports 6 ongoing REBECCA studies, collecting data from breast cancer patients and survivors, as well as prostate cancer patients. Feedback and co-creation with patients have resulted in improvements to the patient app, better fitting their needs and priorities.
2. Enriching Real-World Data with behavioural and online behaviour indicators:
This includes (i) indicators for patient mobility within living space, mainly derived from location recordings, (ii) physical activity and physiological indicators from wearable recordings, such as steps and heart rate, (iii) sleep patterns and quality, (iv) patient’s emotional status inferred from online behaviours, including search patterns and social media interactions, and (v) dietary habit indicators, including meal timing, schedule, duration, and preferences.
Signal processing and deep learning algorithms were developed to extract markers of patients’ functional and emotional status. These include, (i) geographic trajectories imputation to tackle data gaps, (ii) mobility indicators like “time staying at home” and “entropy of stays and moves”, (iii) transportation mode detection from GPS and acceleration signals, (iv) deep learning models for assessing sentiment, stress and anxiety. Ongoing work is focused on discovering associations between these novel indicators and the three complex chronic conditions that are studied within REBECCA: Chemotherapy Induced Peripheral Neuropathy, (ii) Cancer-Related Fatigue, and (iii) Osteoporosis and Osteopenia resulting from treatment with aromatase inhibitors.
3. Causal modelling of the complex relations between behaviours, emotional state, Complex Chronic Conditions and Treatment:
This includes the design of a Directed Acyclic Graph for each targeted cancer-related comorbidity, development of novel tools for treatment effect estimation and causal analysis, and analysis of retrospective registry data.
4. Use of REBECCA to support clinical research and improve patient management:
The REBECCA infrastructure has been deployed to six studies. Two observational studies, REBECCA-CRF and REBECCA-OST, are in progress, investigating the use of RWD obtained through REBECCA system. The former targets understanding of Breast Cancer-Related Fatigue, while the latter targets the understanding of Adjuvant-treatment induced Osteopenia/Osteoporosis in Breast Cancer patients. Two intervention studies, REBECCA-SUH-Inter-QoL and REBECCA-INCLIVA-Inter-QoL, have started after obtaining all necessary permissions. The feasibility study REBECCA-KI-Feas-QoL is also in the stage of finalizing recruitment. In addition, the preliminary stages, including protocol definition and ethical approvals, for two more studies, REBECCA-CIPN and REBECCA-PROST, have been completed and recruitment is set to begin.
5. Sustainability of REBECCA:
The project’s dissemination is based on its web site that is regularly updated, the social media of the project, as well as on active participation in scientific conferences and submission of novel research results for publication. Regarding future exploitation, REBECCA has identified an extensive list of stakeholders in Spain, Norway, Sweden, and Europe in general. REBECCA conducted three “pre-engagement” workshops with local stakeholders in Sweden, Spain, Norway, as well as the main stakeholders’ workshop. The development of REBECCA's business plan is progressing and a Cost-Benefit Analysis has started.
1. Design and development of the necessary Real World Data collection infrastructure:
It consists of (i) the REBECCA Patient mobile app (Android and iOS), which includes integration with Garmin wearables, (ii) the REBECCA Companion app, (iii) a web browser plugin for monitoring online activity, (iv) REBECCA Electronic Case Report Forms for capturing EHR data, (v) backend servers for storage, processing and analytics, (vi) the Clinical Dashboard. This complex infrastructure has already been launched and recently updated to its latest version, REBECCA V3. It supports 6 ongoing REBECCA studies, collecting data from breast cancer patients and survivors, as well as prostate cancer patients. Feedback and co-creation with patients have resulted in improvements to the patient app, better fitting their needs and priorities.
2. Enriching Real-World Data with behavioural and online behaviour indicators:
This includes (i) indicators for patient mobility within living space, mainly derived from location recordings, (ii) physical activity and physiological indicators from wearable recordings, such as steps and heart rate, (iii) sleep patterns and quality, (iv) patient’s emotional status inferred from online behaviours, including search patterns and social media interactions, and (v) dietary habit indicators, including meal timing, schedule, duration, and preferences.
Signal processing and deep learning algorithms were developed to extract markers of patients’ functional and emotional status. These include, (i) geographic trajectories imputation to tackle data gaps, (ii) mobility indicators like “time staying at home” and “entropy of stays and moves”, (iii) transportation mode detection from GPS and acceleration signals, (iv) deep learning models for assessing sentiment, stress and anxiety. Ongoing work is focused on discovering associations between these novel indicators and the three complex chronic conditions that are studied within REBECCA: Chemotherapy Induced Peripheral Neuropathy, (ii) Cancer-Related Fatigue, and (iii) Osteoporosis and Osteopenia resulting from treatment with aromatase inhibitors.
3. Causal modelling of the complex relations between behaviours, emotional state, Complex Chronic Conditions and Treatment:
This includes the design of a Directed Acyclic Graph for each targeted cancer-related comorbidity, development of novel tools for treatment effect estimation and causal analysis, and analysis of retrospective registry data.
4. Use of REBECCA to support clinical research and improve patient management:
The REBECCA infrastructure has been deployed to six studies. Two observational studies, REBECCA-CRF and REBECCA-OST, are in progress, investigating the use of RWD obtained through REBECCA system. The former targets understanding of Breast Cancer-Related Fatigue, while the latter targets the understanding of Adjuvant-treatment induced Osteopenia/Osteoporosis in Breast Cancer patients. Two intervention studies, REBECCA-SUH-Inter-QoL and REBECCA-INCLIVA-Inter-QoL, have started after obtaining all necessary permissions. The feasibility study REBECCA-KI-Feas-QoL is also in the stage of finalizing recruitment. In addition, the preliminary stages, including protocol definition and ethical approvals, for two more studies, REBECCA-CIPN and REBECCA-PROST, have been completed and recruitment is set to begin.
5. Sustainability of REBECCA:
The project’s dissemination is based on its web site that is regularly updated, the social media of the project, as well as on active participation in scientific conferences and submission of novel research results for publication. Regarding future exploitation, REBECCA has identified an extensive list of stakeholders in Spain, Norway, Sweden, and Europe in general. REBECCA conducted three “pre-engagement” workshops with local stakeholders in Sweden, Spain, Norway, as well as the main stakeholders’ workshop. The development of REBECCA's business plan is progressing and a Cost-Benefit Analysis has started.
(1) Launched the REBECCA 360° monitoring platform, featuring tools for capturing real-world data and algorithms for extracting functional and emotional status indicators. The updated platform is integrated into the latest system release, REBECCA V3, and supports the ongoing data collection in six clinical studies. REBECCA’s Early Warning System complements the 360° platform by monitoring patient compliance and tracking behaviour to detect abrupt changes. The results are available to their attending physicians via REBECCA’s Clinical Dashboard.
(2) Directed Acyclic Graphs (DAGs) have been chosen to represent causal relations between behaviours, emotional state, Complex Chronic Conditions (CCCs), and treatment. Three specific DAGs have been created for each of REBECCA's CCCs, which include Chemotherapy Induced Peripheral Neuropathy, Cancer-Related Fatigue, and Osteoporosis and Osteopenia resulting from treatment with aromatase inhibitors.
(3) Proposed a new data-driven AI model, the NN-Dragonnet, for treatment effect estimation, which outperformed models in the state of the art.
(4) REBECCA infrastructure is deployed to six ongoing clinical studies and is gathering real-world data from multiple sources.
(5) Developed the “Advisory Guidelines for a healthy lifestyle for Breast Cancer Survivors”, a guidance manual that was compiled through comprehensive literature research, aiming to connect existing guidelines with real-world measurements, such as the parameters of the REBECCA 360°.
(2) Directed Acyclic Graphs (DAGs) have been chosen to represent causal relations between behaviours, emotional state, Complex Chronic Conditions (CCCs), and treatment. Three specific DAGs have been created for each of REBECCA's CCCs, which include Chemotherapy Induced Peripheral Neuropathy, Cancer-Related Fatigue, and Osteoporosis and Osteopenia resulting from treatment with aromatase inhibitors.
(3) Proposed a new data-driven AI model, the NN-Dragonnet, for treatment effect estimation, which outperformed models in the state of the art.
(4) REBECCA infrastructure is deployed to six ongoing clinical studies and is gathering real-world data from multiple sources.
(5) Developed the “Advisory Guidelines for a healthy lifestyle for Breast Cancer Survivors”, a guidance manual that was compiled through comprehensive literature research, aiming to connect existing guidelines with real-world measurements, such as the parameters of the REBECCA 360°.