Periodic Reporting for period 3 - PERSIST (Patients-centered SurvivorShIp care plan after Cancer treatments based on Big Data and Artificial Intelligence technologies)
Période du rapport: 2022-01-01 au 2023-02-28
Cancer concerns all European citizens. 40% of us are likely to be affected and we all know someone who developed the disease. In 2020, 19.3 million of new cases appeared and there were 9.96 million deaths
For a long time, cancer treatment has been the main focus of research, but the moment has come to pay attention and efforts on 'long-term patients'. Cancer survivors have unmet needs, especially when it comes to improving their independence and quality of life (QoL), in addition to extending the duration of life. These are important aspects to tackle, not only from the patient perspective but also for society acceptance and adaptation of the cancer survivors.
PERSIST goals are very ambitious, pursuing a radical breakthrough in the quality of treatment and follow-up of cancer survivors. PERSIST not only develops innovative applications based on artificial intelligence and big data, but also aims to demonstrate their effectiveness in improving the quality of life of these survivors. In this regard, PERSIST approach is fully patient-centric while taking into account all the necessary elements of the healthcare ecosystem, in order to ensure a rapid uptake of the results in the European context. The complete PERSIST system will incorporate, on the one hand, a Clinical Decision Support System based on new models of health data analysis. On the other hand, a mHealth system for the remote and personalized monitoring of each patient. In addition, the partners will develop a Big Data Platform that integrates the two previous systems and will connect to the Electronic Health Records (EHRs) of the hospitals.
For a long time, cancer treatment has been the main focus of research, but the moment has come to pay attention and efforts on 'long-term patients'. Cancer survivors have unmet needs, especially when it comes to improving their independence and quality of life (QoL), in addition to extending the duration of life. These are important aspects to tackle, not only from the patient perspective but also for society acceptance and adaptation of the cancer survivors.
PERSIST goals are very ambitious, pursuing a radical breakthrough in the quality of treatment and follow-up of cancer survivors. PERSIST not only develops innovative applications based on artificial intelligence and big data, but also aims to demonstrate their effectiveness in improving the quality of life of these survivors. In this regard, PERSIST approach is fully patient-centric while taking into account all the necessary elements of the healthcare ecosystem, in order to ensure a rapid uptake of the results in the European context. The complete PERSIST system will incorporate, on the one hand, a Clinical Decision Support System based on new models of health data analysis. On the other hand, a mHealth system for the remote and personalized monitoring of each patient. In addition, the partners will develop a Big Data Platform that integrates the two previous systems and will connect to the Electronic Health Records (EHRs) of the hospitals.
The final version of Big Data Platform was deployed, including new services allowing the feature of executing AI based modules, and new capabilities providing an easier way to query data stored in the platform.
Multilingual Conversation System was completed and a data model to deliver speech-enabled services and collection of patient experiences (mood, PREMs and PROMs) supported by conversational intelligence. MRAST Framework was delivered as a framework of new wellbeing software sensors to recognize, classify and track symptoms and patient risk factors by exploring PGHD, including digital biomarkers, CTCs, and patient-reported experiences (diaries).
CTC chip was successfully manufactured in COP, and priming and blood processing protocols were optimised taking the PDMS prototype as a reference. The chips were used during the clinical trial. A web application for CTC enumeration and classification results was developed. Acquired CTC images were preprocessed and annotated in accordance with COCO standard, to train an AI algorithm able to classify new CTC images using ML.
The trajectories and cohorts were updated and made available through a new designed web-interface. Additionally, the overall improvement of AI algorithms and tools is reflected in the obtained ENRICHED sets of EHR data, which were iteratively improved to support filling gaps in structured data.
The mClinician web app provides a user-friendly interface for clinicians to enter patient data in a structured format, saving these data as FHIR resources. Clinicians can view and modify existing records. It also enables clinicians to manage and monitor patient activity with the data collected through mHealth app.
The mHealth app was developed as an Android mobile application to gather data from the patients and present that data to both the user and the clinician. Data is gathered using a bluetooth smart-band and in-app features such as questionnaires, video diary and emotion dialog.
In the INFERENCE ENGINE of the clinical decision support system (CDSS) a scoring algorithm was developed for assessing the risk of breast and colon cancer recurrence according to patient’s retrospective data, questionnaire responses and patient habits and lifestyle based on the determined risk factors in the knowledge base and their hazard ratios from the researches in the literature.
Also Full clinical validation was done. Second set of patient workshops in each hospital was organised for feedback gathering. Co-creation activities happened in workshops and individual consultations with clinicians.
Legal and ethical aspects correlated to the activities of the Consortium were taken into consideration. The legal qualification of PERSIST Solution under the Medical Device Regulation 2017/745, along with its relevant KERs, was another perspective taken under close observation.
PERSIST consortium presented the project on several events and seminars, as well as to organise project-specific events. Besides, relevant materials were produced like newsletters, press releases, videos etc. Furthermore, several electronic and web dissemination channels kept their dissemination activities, including the project website, its collaboration portal, social media accounts and channels.
Lastly, the “Business Case” activity provided the guidelines for making a reliable business model. It also provided the market test template where applicable.
Multilingual Conversation System was completed and a data model to deliver speech-enabled services and collection of patient experiences (mood, PREMs and PROMs) supported by conversational intelligence. MRAST Framework was delivered as a framework of new wellbeing software sensors to recognize, classify and track symptoms and patient risk factors by exploring PGHD, including digital biomarkers, CTCs, and patient-reported experiences (diaries).
CTC chip was successfully manufactured in COP, and priming and blood processing protocols were optimised taking the PDMS prototype as a reference. The chips were used during the clinical trial. A web application for CTC enumeration and classification results was developed. Acquired CTC images were preprocessed and annotated in accordance with COCO standard, to train an AI algorithm able to classify new CTC images using ML.
The trajectories and cohorts were updated and made available through a new designed web-interface. Additionally, the overall improvement of AI algorithms and tools is reflected in the obtained ENRICHED sets of EHR data, which were iteratively improved to support filling gaps in structured data.
The mClinician web app provides a user-friendly interface for clinicians to enter patient data in a structured format, saving these data as FHIR resources. Clinicians can view and modify existing records. It also enables clinicians to manage and monitor patient activity with the data collected through mHealth app.
The mHealth app was developed as an Android mobile application to gather data from the patients and present that data to both the user and the clinician. Data is gathered using a bluetooth smart-band and in-app features such as questionnaires, video diary and emotion dialog.
In the INFERENCE ENGINE of the clinical decision support system (CDSS) a scoring algorithm was developed for assessing the risk of breast and colon cancer recurrence according to patient’s retrospective data, questionnaire responses and patient habits and lifestyle based on the determined risk factors in the knowledge base and their hazard ratios from the researches in the literature.
Also Full clinical validation was done. Second set of patient workshops in each hospital was organised for feedback gathering. Co-creation activities happened in workshops and individual consultations with clinicians.
Legal and ethical aspects correlated to the activities of the Consortium were taken into consideration. The legal qualification of PERSIST Solution under the Medical Device Regulation 2017/745, along with its relevant KERs, was another perspective taken under close observation.
PERSIST consortium presented the project on several events and seminars, as well as to organise project-specific events. Besides, relevant materials were produced like newsletters, press releases, videos etc. Furthermore, several electronic and web dissemination channels kept their dissemination activities, including the project website, its collaboration portal, social media accounts and channels.
Lastly, the “Business Case” activity provided the guidelines for making a reliable business model. It also provided the market test template where applicable.
At the end of the project PERSIST has got ready a robust, scalable and flexible BIG DATA PLATFORM enabling the flow of complex and fragmented data from disparate applications and heterogeneous sources. This flow issecured and supported by open interfacing industry standards and the use of FHIR for clinical resource definition. Given that part of the data was extracted from hospitals with different storing schema, a protocol for the extraction and preparation of EHR data for secondary use was designed.
PERSIST also offered patients the possibility of remote support and monitoring. The combination of patients’ wearables, a mHealth application and an innovative multimodal sensing network collecting objective data (e.g. physical activity frequency, duration, new biomarkers, and intensity), patients’ perceptions (on their health and experience) and complex phenomena such as mood and depression. Since these are new sources of health and wellbeing data, the corresponding standard mechanism for collecting their data in a secure and reliable way was set.
Additionally, PERSIST has developed a new software for automated CTC counting and new classification models based on image analysis, signal fusion and transfer knowledge concepts for tracking and classification of wellbeing and health parameters of cancer survivors. Artificial Intelligence (AI) was employed to achieve predictive models based on multiple level cohort analysis and identification of common trajectories of cancer patients that can be used to personalize and optimize treatment, and to prevent adverse events and hospital readmissions. This predictive model will be the base of clinical knowledge modules that provide clinicians accurate recommendations for developing, reviewing and updating Survivor Care Plans (SCP) and highlight alarming states and situations where rapid response is required. This will allow associations between lifestyle and wellbeing factors and the outcomes observed during the follow up of cancer survivors, helping in the prevention and timely addressing of long-term side effects.
During the project, four simultaneous clinical studies (in Latvia, Belgium, Slovenia and Spain) were carried out to demonstrate the advantages of monitoring lifestyle factors in cancer survivors and draw conclusions on the usability and performance of the ecosystem from patients and professionals perspectives.
PERSIST also offered patients the possibility of remote support and monitoring. The combination of patients’ wearables, a mHealth application and an innovative multimodal sensing network collecting objective data (e.g. physical activity frequency, duration, new biomarkers, and intensity), patients’ perceptions (on their health and experience) and complex phenomena such as mood and depression. Since these are new sources of health and wellbeing data, the corresponding standard mechanism for collecting their data in a secure and reliable way was set.
Additionally, PERSIST has developed a new software for automated CTC counting and new classification models based on image analysis, signal fusion and transfer knowledge concepts for tracking and classification of wellbeing and health parameters of cancer survivors. Artificial Intelligence (AI) was employed to achieve predictive models based on multiple level cohort analysis and identification of common trajectories of cancer patients that can be used to personalize and optimize treatment, and to prevent adverse events and hospital readmissions. This predictive model will be the base of clinical knowledge modules that provide clinicians accurate recommendations for developing, reviewing and updating Survivor Care Plans (SCP) and highlight alarming states and situations where rapid response is required. This will allow associations between lifestyle and wellbeing factors and the outcomes observed during the follow up of cancer survivors, helping in the prevention and timely addressing of long-term side effects.
During the project, four simultaneous clinical studies (in Latvia, Belgium, Slovenia and Spain) were carried out to demonstrate the advantages of monitoring lifestyle factors in cancer survivors and draw conclusions on the usability and performance of the ecosystem from patients and professionals perspectives.