Postdoc: Better real-world health-data distributed analytics research

DEIB, Milano (IT) Apr. 2024 - now Pietro Pinoli, Anna Bernasconi

https://www.better-health-project.eu/
Main achievements:
- Propose a general FAIR data model for handling different health data
- Define a set of metadata for each dataset and align it with existing healthcare ontologies
- Propose and implement an ETL pipeline to integrate and FAIRify such data
- Expose the integrated data in a decentralized structure following the PHT (Personal Health Train) for researchers to analyze data and find possible causes of 3 rare diseases (intellectual disability, eye distrophies, and suicidal behaviors for autistic patients)

PhD: User-oriented exploration of semi-structured datasets

Institut Polytechnique de Paris & Inria Saclay (FR) Jan. 2021 - March 2024 Pr. Ioana Manolescu, SME WeDoData

My PhD work received the accessit (2nd place) for the prize given to PhDs with significant contributions to the data management community.
Main achievements:
- Entity-Relationship-like summaries out of semi-structured datasets (system Abstra)
- Entity path enumeration between entities of interest in heterogeneous data (system PathWays)
Summer schools: HiParis 2021, MDD 2022, HiParis 2023
Transversal training: "Teaching at university", "Public speaking" and English courses
See also Talks and Teaching sections

PhD manuscript
Defense slides

LIRIS (FR) Feb. 2020 - July 2020 Fabien Duchateau, Franck Favetta, start-up Home in Love

Propose an algorithm to select top-k features about neighborhoods to help the prediction
Predict a set of environment variables, e.g., the landscape, or the wealth of a neighborhood
Develop a cartographic visualization interface for neighborhoods
Develop an interface for generic tuning of prediction algorithms

LIRIS (FR) Jan. 2019 - June 2019 Fabien Duchateau, Franck Favetta

Integrate heterogeneous cartographic data from Geonames, Bing, Here and Open Street Maps
Create a global schema based on individual data providers schemas
Propose a tunable formula for detecting correspondences, based on attribute similarity, and automatically estimate their quality
Merge correspondences (different strategies)
Develop an interface for matching and merging correspondences between points of interest (geographic entities) with an estimation of the quality

Poster (in French)

LIRIS (FR) May 2018 - July 2018 Fabien Duchateau, Franck Favetta, start-up Home in Love

Integrate data from heterogeneous sources (Excel, JSON, GeoJSON...)
Use prediction algorithms to recommend neighborhoods based on few criteria
Use clustering algorithms to classify neighborhoods based on their characteristics
Develop an interface for facilitating comparison and recommendation of neighborhoods in France

Internship report (in French)