Periodic Reporting for period 1 - AIDA (An Artificially Intelligent Diagnostic Assistant for gastric inflammation)
Reporting period: 2023-01-01 to 2024-06-30
Most cases of gastric cancer (GC) are detected at a late stage, when patients have a median life expectancy of about a year. Diagnosing people at risk of developing GC at the pre-symptomatic stage, typically chronic gastric inflammation could significantly improve the outlook.
Artificial intelligence (AI) can help clinicians make sense of their own data by automating much of the treatment and analysis, which require manual work and years of experience. But it can do more: it can bring together available data from various sources into a vast data lake and cross-correlate the data to derive a ‘risk score’ for gastric cancer and shed light on the mechanisms of its evolution.
This project aims to develop and validate a multidisciplinary AI-powered assistant, Aida, that helps clinicians diagnose precancerous inflammation, suggests personalised therapeutic strategies for medical treatment and follow-up, and makes personalised recommendations for monitoring patient health status, thus contributing to GC prevention. Aida is primarily a toolbox for clinicians with apps and dashboards but there is also a website for patients and the general public. It should be stressed that Aida will not replace clinicians or researchers, but ease the most tedious and error-prone of their tasks while helping establish a more accurate diagnosis using multidisciplinary data that lies beyond the clinician’s normal professional horizon. A related advantage of artificial intelligence that is often overlooked, especially in computer-aided diagnoses, is its ability to restrict itself to facts and remove the emotional element from important decision-making. This makes Aida a true assistant as opposed to an autonomous agent. The final responsibility of any diagnosis or intervention remains with the clinician at all times. Aida will be ethical and trusted, and aligned to approaches such as the European Data Spaces and GaiaX.
At the end of the project, Aida will start preparation for MDR certification under Class 2a in order to be used as a medically certified tool in clinical practice.
After the project, the results will live on in an association that acts as a transnational focal point for chronic gastric inflammation — and GC in general. We hope that the solid, inclusive design principles of Aida, its societal relevance and its durability will spawn a vigorous ecosystem around chronic gastric inflammation, its understanding and its treatment. And we hope that it will inspire other data collaboratives in health — for other chronic inflammations, other forms of cancer or other ailments altogether.
Artificial intelligence (AI) can help clinicians make sense of their own data by automating much of the treatment and analysis, which require manual work and years of experience. But it can do more: it can bring together available data from various sources into a vast data lake and cross-correlate the data to derive a ‘risk score’ for gastric cancer and shed light on the mechanisms of its evolution.
This project aims to develop and validate a multidisciplinary AI-powered assistant, Aida, that helps clinicians diagnose precancerous inflammation, suggests personalised therapeutic strategies for medical treatment and follow-up, and makes personalised recommendations for monitoring patient health status, thus contributing to GC prevention. Aida is primarily a toolbox for clinicians with apps and dashboards but there is also a website for patients and the general public. It should be stressed that Aida will not replace clinicians or researchers, but ease the most tedious and error-prone of their tasks while helping establish a more accurate diagnosis using multidisciplinary data that lies beyond the clinician’s normal professional horizon. A related advantage of artificial intelligence that is often overlooked, especially in computer-aided diagnoses, is its ability to restrict itself to facts and remove the emotional element from important decision-making. This makes Aida a true assistant as opposed to an autonomous agent. The final responsibility of any diagnosis or intervention remains with the clinician at all times. Aida will be ethical and trusted, and aligned to approaches such as the European Data Spaces and GaiaX.
At the end of the project, Aida will start preparation for MDR certification under Class 2a in order to be used as a medically certified tool in clinical practice.
After the project, the results will live on in an association that acts as a transnational focal point for chronic gastric inflammation — and GC in general. We hope that the solid, inclusive design principles of Aida, its societal relevance and its durability will spawn a vigorous ecosystem around chronic gastric inflammation, its understanding and its treatment. And we hope that it will inspire other data collaboratives in health — for other chronic inflammations, other forms of cancer or other ailments altogether.
Main activities and achievements relative to this first reporting period include:
- Establishment of all the boards, documents (DMP, Q&RMP, E&DR, DTA, MTA) and approvals (ethical committees, NDAs) for the proper retrospective and prospective data collection, activating all the clinical sites
- Establishment of the project's dedicated website and all its social media channels
- Establishment of all the protocols (lab handbook, sample collection, data harmonisation)
- Definition of data requirements and harmonisation for pathology and endoscopy data collection, with its later definition of cohorts necessary to train all of the planned AI models
- Development and implementation of the electronic case report form (REDCap)
- Development of the AI Clinician system workflows utilising Reinforcement Learning to recommend patient-specific first-line treatments for H. pylori.
- Development of an integrative AI development framework for detecting gastric inflammation and providing personalised risk stratification using endoscopy and pathology data
- Design and testing of the technical architecture of the data collaborative
- Identification of the convergencies and divergencies between the AIDA data collaborative and EU-wide standards and principles on health data, data spaces, FAIR, and AI
- Development and deployment of the H. Pylori dashboard
- Definition and agreement of the AIDA Association vision and mission, with its establishment already in progress.
- Establishment of all the boards, documents (DMP, Q&RMP, E&DR, DTA, MTA) and approvals (ethical committees, NDAs) for the proper retrospective and prospective data collection, activating all the clinical sites
- Establishment of the project's dedicated website and all its social media channels
- Establishment of all the protocols (lab handbook, sample collection, data harmonisation)
- Definition of data requirements and harmonisation for pathology and endoscopy data collection, with its later definition of cohorts necessary to train all of the planned AI models
- Development and implementation of the electronic case report form (REDCap)
- Development of the AI Clinician system workflows utilising Reinforcement Learning to recommend patient-specific first-line treatments for H. pylori.
- Development of an integrative AI development framework for detecting gastric inflammation and providing personalised risk stratification using endoscopy and pathology data
- Design and testing of the technical architecture of the data collaborative
- Identification of the convergencies and divergencies between the AIDA data collaborative and EU-wide standards and principles on health data, data spaces, FAIR, and AI
- Development and deployment of the H. Pylori dashboard
- Definition and agreement of the AIDA Association vision and mission, with its establishment already in progress.