Periodic Reporting for period 1 - I3LUNG (Integrative science, Intelligent data platform for Individualized LUNG cancer care with Immunotherapy)
Reporting period: 2022-06-01 to 2023-11-30
Although immunotherapy is the new standard of care for advanced non-small cell lung cancer (NSCLC), less than 50 % of patients benefit from this treatment in the long term. Programmed death-ligand 1 biomarker is used to predict immunotherapy outcomes with limited efficacy, and other potential biomarkers have not yet been validated in randomised clinical trials. The EU-funded I3LUNG project aims to develop AI-based tools for improving patients’ survival and quality of life.
The project will set up a global platform comprising data from 2 000 patients for AI models validation; moreover, it will collect multi-omics data from 200 NSCLC patients for information integration and application in leading immunotherapy decisions.
The project will set up a global platform comprising data from 2 000 patients for AI models validation; moreover, it will collect multi-omics data from 200 NSCLC patients for information integration and application in leading immunotherapy decisions.
I3LUNG has a three-pronged action plan, all of them ongoing according to the initial schedule as of January 2024.
The first is the development of Artificial Intelligence models for the analysis of data collected from non-small cell lung cancer patients, treated with immunotherapy, enrolled in past clinical trials in the participating clinical centers, or in a prospective study launched in 2022 and still ongoing and open to recruitment. The data collected comes from a wide range of sources and represents a comprehensive overview of the patient’s, and their tumor’s, situation: clinical data of course, as well as a variety of “omics” biological data (microbiota, transcriptome, genomics data, etc.).
This data is accessible via a dedicated, secure Data Sharing and Elaboration Platform specifically built for the project, and as of the end of the first reporting period, already fully operational.
The Platform (named “METALUNG”) also serves as the basis for the next actionable results of the project, in particular the development of three AI-driven tools. They are aimed respectively at: patients, first and foremost - the IPDAS (Informed Patient Decision Aid System) helps the patients enrolled in the study to gain information on their disease and on the treatment options available to them. It also proposes different paths - to be discussed with their carer - based on objective data and their own assessment of their quality of life, and wishes in regards to their care (adverse effects risk, length of the therapy, etc.). The IPDAS is already operational as well in its first iteration, which will be regularly updated until the fourth year of the project when it will enter the last phase for its full, definitive implementation.
The other tools are also implemented, in particular the Physician Decision Support System (PDSS) which provides carers with a user-friendly prediction system from which the I3LUNG AI model will provide them prediction scores of a patient’s different endpoints (overall survival, progression-free survival), each time with the accuracy of the prediction - a vital part of the model, to help guide the treatment option choice they will provide to their patients. Finally, a last part of this toolset will be opened to a wider research audience, providing insights on the model, the data analysis, and the project’s data - this part is under development and currently not in open access, as the model is being built. A first successful test was run in January 2024 and, as of April 2024, the data collection pipelines for each set of data are undergoing the final validation steps.
In addition to the clinical studies, omics analyses and AI toolset developed, the project also has three cross-section objectives.
• Providing enrolled patients with a dedicated mobile app for the assessment of their quality of life (both self-reported questionnaires and objective features analyses - e.g. cough recordings) - already implemented and being distributed to a first set of patients
• A psychological study is currently assessing: the perception of patients regarding this innovative approach, the inclusion of AI in the treatment selection and how much they know about / trust this new actor - and the same approach but on the physicians, to evaluate how AI is changing their daily practice in this specific trial. These studies are ongoing.
• Finally, two partners from the field of social sciences and humanities are active in the project, first of all for an assessment of the financial burden of this new approach and the construction of a budget impact model (a first version was finalized and will be fine tuned in the coming months using the results of the prospective trial). Another fundamental aspect of the project is ensuring its adequacy with the legal and regulatory framework in the EU and abroad (two project partners are from outside Europe - the USA and Israel specifically), both in the clinical trial field but especially considering the regulations around the use of AI - AI Act, new EU guidelines, etc.
Importantly, this project wants to ensure a sharing and dissemination of knowledge on AI in clinical and preclinical oncology, not only to the scientific community but also the general public and the patients in primis - for this, we are very closely collaborating with patient advocates, and focusing on proper and transparent publications, open science principles once actionable results will be ready to be shared, online presence, and interactions with other EU initiatives and research projects in this field.
The first is the development of Artificial Intelligence models for the analysis of data collected from non-small cell lung cancer patients, treated with immunotherapy, enrolled in past clinical trials in the participating clinical centers, or in a prospective study launched in 2022 and still ongoing and open to recruitment. The data collected comes from a wide range of sources and represents a comprehensive overview of the patient’s, and their tumor’s, situation: clinical data of course, as well as a variety of “omics” biological data (microbiota, transcriptome, genomics data, etc.).
This data is accessible via a dedicated, secure Data Sharing and Elaboration Platform specifically built for the project, and as of the end of the first reporting period, already fully operational.
The Platform (named “METALUNG”) also serves as the basis for the next actionable results of the project, in particular the development of three AI-driven tools. They are aimed respectively at: patients, first and foremost - the IPDAS (Informed Patient Decision Aid System) helps the patients enrolled in the study to gain information on their disease and on the treatment options available to them. It also proposes different paths - to be discussed with their carer - based on objective data and their own assessment of their quality of life, and wishes in regards to their care (adverse effects risk, length of the therapy, etc.). The IPDAS is already operational as well in its first iteration, which will be regularly updated until the fourth year of the project when it will enter the last phase for its full, definitive implementation.
The other tools are also implemented, in particular the Physician Decision Support System (PDSS) which provides carers with a user-friendly prediction system from which the I3LUNG AI model will provide them prediction scores of a patient’s different endpoints (overall survival, progression-free survival), each time with the accuracy of the prediction - a vital part of the model, to help guide the treatment option choice they will provide to their patients. Finally, a last part of this toolset will be opened to a wider research audience, providing insights on the model, the data analysis, and the project’s data - this part is under development and currently not in open access, as the model is being built. A first successful test was run in January 2024 and, as of April 2024, the data collection pipelines for each set of data are undergoing the final validation steps.
In addition to the clinical studies, omics analyses and AI toolset developed, the project also has three cross-section objectives.
• Providing enrolled patients with a dedicated mobile app for the assessment of their quality of life (both self-reported questionnaires and objective features analyses - e.g. cough recordings) - already implemented and being distributed to a first set of patients
• A psychological study is currently assessing: the perception of patients regarding this innovative approach, the inclusion of AI in the treatment selection and how much they know about / trust this new actor - and the same approach but on the physicians, to evaluate how AI is changing their daily practice in this specific trial. These studies are ongoing.
• Finally, two partners from the field of social sciences and humanities are active in the project, first of all for an assessment of the financial burden of this new approach and the construction of a budget impact model (a first version was finalized and will be fine tuned in the coming months using the results of the prospective trial). Another fundamental aspect of the project is ensuring its adequacy with the legal and regulatory framework in the EU and abroad (two project partners are from outside Europe - the USA and Israel specifically), both in the clinical trial field but especially considering the regulations around the use of AI - AI Act, new EU guidelines, etc.
Importantly, this project wants to ensure a sharing and dissemination of knowledge on AI in clinical and preclinical oncology, not only to the scientific community but also the general public and the patients in primis - for this, we are very closely collaborating with patient advocates, and focusing on proper and transparent publications, open science principles once actionable results will be ready to be shared, online presence, and interactions with other EU initiatives and research projects in this field.