MERGE: A model for multi-input biomedical federated learning
Summary: Driven by the deep learning (DL) revolution, artificial intelligence (AI) has become a fundamental tool for many biomedical tasks, including analyzing and classifying diagnostic images. Imaging, however, is not the only source of information. Tabular data, such as personal and genomic data...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2023-11-01
|
| Series: | Patterns |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2666389923002404 |
| _version_ | 1797630611819593728 |
|---|---|
| author | Bruno Casella Walter Riviera Marco Aldinucci Gloria Menegaz |
| author_facet | Bruno Casella Walter Riviera Marco Aldinucci Gloria Menegaz |
| author_sort | Bruno Casella |
| collection | DOAJ |
| description | Summary: Driven by the deep learning (DL) revolution, artificial intelligence (AI) has become a fundamental tool for many biomedical tasks, including analyzing and classifying diagnostic images. Imaging, however, is not the only source of information. Tabular data, such as personal and genomic data and blood test results, are routinely collected but rarely considered in DL pipelines. Nevertheless, DL requires large datasets that often must be pooled from different institutions, raising non-trivial privacy concerns. Federated learning (FL) is a cooperative learning paradigm that aims to address these issues by moving models instead of data across different institutions. Here, we present a federated multi-input architecture using images and tabular data as a methodology to enhance model performance while preserving data privacy. We evaluated it on two showcases: the prognosis of COVID-19 and patients’ stratification in Alzheimer’s disease, providing evidence of enhanced accuracy and F1 scores against single-input models and improved generalizability against non-federated models. The bigger picture: Deep learning models must be trained with large datasets, which often requires pooling data from different sites and sources. In research fields dealing with sensitive information subject to data regulations, such as biomedical research, data pooling can generate concerns about data access and sharing across institutions, which can affect performance, energy consumption, privacy, and security. Federated learning is a cooperative learning paradigm that addresses such concerns by sharing models instead of data across different institutions. |
| first_indexed | 2024-03-11T11:09:31Z |
| format | Article |
| id | doaj.art-785fe518d3564ca2aedf7a4b157242b2 |
| institution | Directory Open Access Journal |
| issn | 2666-3899 |
| language | English |
| last_indexed | 2024-03-11T11:09:31Z |
| publishDate | 2023-11-01 |
| publisher | Elsevier |
| record_format | Article |
| series | Patterns |
| spelling | doaj.art-785fe518d3564ca2aedf7a4b157242b22023-11-12T04:41:04ZengElsevierPatterns2666-38992023-11-01411100856MERGE: A model for multi-input biomedical federated learningBruno Casella0Walter Riviera1Marco Aldinucci2Gloria Menegaz3Department of Computer Science, University of Turin, 10149 Turin, ItalyDepartment of Computer Science, University of Verona, 37134 Verona, Italy; Corresponding authorDepartment of Computer Science, University of Turin, 10149 Turin, ItalyDepartment of Engineering for Innovation Medicine, University of Verona, 37134 Verona, ItalySummary: Driven by the deep learning (DL) revolution, artificial intelligence (AI) has become a fundamental tool for many biomedical tasks, including analyzing and classifying diagnostic images. Imaging, however, is not the only source of information. Tabular data, such as personal and genomic data and blood test results, are routinely collected but rarely considered in DL pipelines. Nevertheless, DL requires large datasets that often must be pooled from different institutions, raising non-trivial privacy concerns. Federated learning (FL) is a cooperative learning paradigm that aims to address these issues by moving models instead of data across different institutions. Here, we present a federated multi-input architecture using images and tabular data as a methodology to enhance model performance while preserving data privacy. We evaluated it on two showcases: the prognosis of COVID-19 and patients’ stratification in Alzheimer’s disease, providing evidence of enhanced accuracy and F1 scores against single-input models and improved generalizability against non-federated models. The bigger picture: Deep learning models must be trained with large datasets, which often requires pooling data from different sites and sources. In research fields dealing with sensitive information subject to data regulations, such as biomedical research, data pooling can generate concerns about data access and sharing across institutions, which can affect performance, energy consumption, privacy, and security. Federated learning is a cooperative learning paradigm that addresses such concerns by sharing models instead of data across different institutions.http://www.sciencedirect.com/science/article/pii/S2666389923002404DSML 2: Proof-of-Concept: Data science output has been formulated, implemented, and tested for one domain/problem |
| spellingShingle | Bruno Casella Walter Riviera Marco Aldinucci Gloria Menegaz MERGE: A model for multi-input biomedical federated learning Patterns DSML 2: Proof-of-Concept: Data science output has been formulated, implemented, and tested for one domain/problem |
| title | MERGE: A model for multi-input biomedical federated learning |
| title_full | MERGE: A model for multi-input biomedical federated learning |
| title_fullStr | MERGE: A model for multi-input biomedical federated learning |
| title_full_unstemmed | MERGE: A model for multi-input biomedical federated learning |
| title_short | MERGE: A model for multi-input biomedical federated learning |
| title_sort | merge a model for multi input biomedical federated learning |
| topic | DSML 2: Proof-of-Concept: Data science output has been formulated, implemented, and tested for one domain/problem |
| url | http://www.sciencedirect.com/science/article/pii/S2666389923002404 |
| work_keys_str_mv | AT brunocasella mergeamodelformultiinputbiomedicalfederatedlearning AT walterriviera mergeamodelformultiinputbiomedicalfederatedlearning AT marcoaldinucci mergeamodelformultiinputbiomedicalfederatedlearning AT gloriamenegaz mergeamodelformultiinputbiomedicalfederatedlearning |