Manolis Terrovitis
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Manolis Terrovitis has a total of 72 co-authors. Here are the top co-authors by number of shared publications:
- Panagiotis Bouros
Johannes Gutenberg University Mainz, Mainz, Germany
10 shared publications - Nikos Mamoulis
Department of Computer Science and Engineering, University of Ioannina, Ioannina, Greece
7 shared publications - Dimitrios Tsitsigkos
Archimedes, Athena RC, Athens, Greece
6 shared publications - John Liagouris
Boston University, Boston, MA, USA
4 shared publications Nikos Mamoulis
3 shared publications- Nikos Mamoulis
University of Hong Kong
3 shared publications
Latest Publications
- Open Access
- 21 May 2026
- Published by IOS Press
- Vol. 336, 223-227
Feasibility of Causality-Aware Machine Learning for Drug Safety on OMOP-CDM
Regulators increasingly view real-world healthcare data and potential use of Artificial Intelligence (AI) approaches as vital for pharmacovigilance (PV). Large European and global initiatives have invested in the development of scalable pharmacovigilance pipelines, often leveraging common data models such as OMOP-CDM. To this end, we propose a pipeline for PV signal analysis grounded in causal machine learning principles, combining standard machine learning (ML) algorithms with DAG-informed feature selection. We demonstrate the approach using data from Papageorgiou General Hospital in Thessaloniki, Greece.
- 21 May 2026
- Published by IOS Press
- Vol. 336, 223-227
Open AccessFeasibility of Causality-Aware Machine Learning for Drug Safety on OMOP-CDM
Regulators increasingly view real-world healthcare data and potential use of Artificial Intelligence (AI) approaches as vital for pharmacovigilance (PV). Large European and global initiatives have invested in the development of scalable pharmacovigilance pipelines, often leveraging common data models such as OMOP-CDM. To this end, we propose a pipeline for PV signal analysis grounded in causal machine learning principles, combining standard machine learning (ML) algorithms with DAG-informed feature selection. We demonstrate the approach using data from Papageorgiou General Hospital in Thessaloniki, Greece.
- Open Access
- 8 January 2026
- Published by JMIR Publications Inc. in JMIR Preprints
Implementation of Individual Patient Data (IPD) Sharing Strategies in the VACCELERATE Project: A Case Study of COVID-19 Vaccine Trials (Preprint)
BACKGROUND Sharing Individual Patient Data (IPD) from clinical trials is increasingly promoted to improve transparency, reproducibility, and scientific value. However, implementation remains challenging due to ethical, regulatory, and operational constraints, particularly in multinational studies. OBJECTIVE This study evaluates the implementation of IPD sharing strategies within three multinational COVID-19 vaccine trials (EU-COVAT-1 AGED, EU-COVAT-2 BOOSTAVAC, and EU-COVPT-1 CoVacc) conducted under the VACCELERATE platform, focusing on protocol integration, informed consent, anonymisation, and repository-based data governance. METHODS Data sharing strategies were prospectively integrated into all trial protocols. A harmonized informed consent template was developed to secure broad participant consent. Data collection followed Clinical Data Acquisition Standards Harmonization (CDASH) guidelines. A structured, risk-based anonymization methodology was applied to the IPD of the AGED trial, involving data minimization and transformation of quasi-identifiers. The anonymized IPD dataset and relevant trial metadata were prepared for deposition in the ECRIN Clinical Research Data Sharing Repository (crDSR), hosted within a secure trusted research environment, in a GDPR-compliant process. RESULTS All three trials included IPD sharing provisions in their protocols. Full implementation was achieved only in the EU-COVAT-1 AGED trial, which obtained ethics approval for the harmonised consent template and completed anonymisation and repository submission. Operational constraints and national regulatory differences prevented full implementation in the other trials. The EU-COVAT-1 AGED dataset was successfully deposited in the repository, enabling secure long-term preservation and controlled secondary access. CONCLUSIONS The VACCELERATE experience demonstrates that responsible IPD sharing is feasible when integrated early into trial design and supported by harmonised consent, robust anonymisation, and secure repository infrastructure. Persistent regulatory heterogeneity and resource limitations remain key barriers. Early planning and institutional support are essential to making FAIR data sharing routine in multinational clinical research. CLINICALTRIAL EudraCT 2021-004526-29; https://www.clinicaltrialsregister.eu/ctr-search/trial/2021-004526-29/DE/; 2021-004889-35; https://www.clinicaltrialsregister.eu/ctr-search/search?query=eudract_number:2021-004889-35; and 2021-004526-29; https://www.clinicaltrialsregister.eu/ctr-search/search?query=eudract_number:2021-004526-29 INTERNATIONAL REGISTERED REPORT RR2-https://doi.org/10.1016/j.ijid.2024.107161
- 8 January 2026
- Published by JMIR Publications Inc. in JMIR Preprints
Open AccessImplementation of Individual Patient Data (IPD) Sharing Strategies in the VACCELERATE Project: A Case Study of COVID-19 Vaccine Trials (Preprint)
BACKGROUND Sharing Individual Patient Data (IPD) from clinical trials is increasingly promoted to improve transparency, reproducibility, and scientific value. However, implementation remains challenging due to ethical, regulatory, and operational constraints, particularly in multinational studies. OBJECTIVE This study evaluates the implementation of IPD sharing strategies within three multinational COVID-19 vaccine trials (EU-COVAT-1 AGED, EU-COVAT-2 BOOSTAVAC, and EU-COVPT-1 CoVacc) conducted under the VACCELERATE platform, focusing on protocol integration, informed consent, anonymisation, and repository-based data governance. METHODS Data sharing strategies were prospectively integrated into all trial protocols. A harmonized informed consent template was developed to secure broad participant consent. Data collection followed Clinical Data Acquisition Standards Harmonization (CDASH) guidelines. A structured, risk-based anonymization methodology was applied to the IPD of the AGED trial, involving data minimization and transformation of quasi-identifiers. The anonymized IPD dataset and relevant trial metadata were prepared for deposition in the ECRIN Clinical Research Data Sharing Repository (crDSR), hosted within a secure trusted research environment, in a GDPR-compliant process. RESULTS All three trials included IPD sharing provisions in their protocols. Full implementation was achieved only in the EU-COVAT-1 AGED trial, which obtained ethics approval for the harmonised consent template and completed anonymisation and repository submission. Operational constraints and national regulatory differences prevented full implementation in the other trials. The EU-COVAT-1 AGED dataset was successfully deposited in the repository, enabling secure long-term preservation and controlled secondary access. CONCLUSIONS The VACCELERATE experience demonstrates that responsible IPD sharing is feasible when integrated early into trial design and supported by harmonised consent, robust anonymisation, and secure repository infrastructure. Persistent regulatory heterogeneity and resource limitations remain key barriers. Early planning and institutional support are essential to making FAIR data sharing routine in multinational clinical research. CLINICALTRIAL EudraCT 2021-004526-29; https://www.clinicaltrialsregister.eu/ctr-search/trial/2021-004526-29/DE/; 2021-004889-35; https://www.clinicaltrialsregister.eu/ctr-search/search?query=eudract_number:2021-004889-35; and 2021-004526-29; https://www.clinicaltrialsregister.eu/ctr-search/search?query=eudract_number:2021-004526-29 INTERNATIONAL REGISTERED REPORT RR2-https://doi.org/10.1016/j.ijid.2024.107161
- 10 May 2025
- Published by Springer Nature
- p. 1378-1380
km-Anonymity
- Open Access
- 12 November 2024
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Spatial Algorithms and Systems
- Vol. 11 (1) , 1-37
Efficient Distance Queries on Non-point Data
Distance queries, including distance-range queries, k -nearest neighbors search, and distance joins, are very popular in spatial databases. However, they have been studied mainly for point data. Inspired by a recent approach on indexing non-point objects for rectangular range queries, we propose a secondary partitioning approach for space-partitioning indices, which is appropriate for distance queries. Our approach classifies the contents of each primary partition into 16 secondary partitions, taking into consideration the begin and end values of objects with respect to the spatial extent of the primary partition. Based on this, we define algorithms for three popular spatial query types, that avoid duplicate results and skip unnecessary computations. We compare our approach to the previous secondary partitioning method and to state-of-the-art data-partitioning indexing and find that it has a significant performance advantage.
- 12 November 2024
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Spatial Algorithms and Systems
- Vol. 11 (1) , 1-37
Open AccessEfficient Distance Queries on Non-point Data
Distance queries, including distance-range queries, k -nearest neighbors search, and distance joins, are very popular in spatial databases. However, they have been studied mainly for point data. Inspired by a recent approach on indexing non-point objects for rectangular range queries, we propose a secondary partitioning approach for space-partitioning indices, which is appropriate for distance queries. Our approach classifies the contents of each primary partition into 16 secondary partitions, taking into consideration the begin and end values of objects with respect to the spatial extent of the primary partition. Based on this, we define algorithms for three popular spatial query types, that avoid duplicate results and skip unnecessary computations. We compare our approach to the previous secondary partitioning method and to state-of-the-art data-partitioning indexing and find that it has a significant performance advantage.
- Open Access
- 12 April 2024
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Knowledge Discovery From Data
- Vol. 18 (6) , 1-35
Adaptive Content-Aware Influence Maximization via Online Learning to Rank
How can we adapt the composition of a post over a series of rounds to make it more appealing in a social network? Techniques that progressively learn how to make a fixed post more influential over rounds have been studied in the context of the Influence Maximization (IM) problem, which seeks a set of seed users that maximize a post’s influence. However, there is no work on progressively learning how a post’s features affect its influence. In this article, we propose and study the problem of Adaptive Content-Aware Influence Maximization (ACAIM), which calls to find k features to form a post in each round so as to maximize the cumulative influence of those posts over all rounds. We solve ACAIM by applying, for the first time, an Online Learning to Rank (OLR) framework for IM purposes. We introduce the CATRID propagation model , which expresses how posts disseminate in a social network using click probabilities and post visibility criteria and develop a simulator that runs CATRID via a training-testing scheme based on real posts of the VK social network, so as to realistically represent the learning environment. We deploy three learners that solve ACAIM in an online (real-time) manner. We experimentally prove the practical suitability of our solutions via exhaustive experiments on multiple brands (operating as different case studies ) and several VK datasets; the best learner is evaluated on 45 separate case studies yielding convincing results.
- 12 April 2024
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Knowledge Discovery From Data
- Vol. 18 (6) , 1-35
Open AccessAdaptive Content-Aware Influence Maximization via Online Learning to Rank
How can we adapt the composition of a post over a series of rounds to make it more appealing in a social network? Techniques that progressively learn how to make a fixed post more influential over rounds have been studied in the context of the Influence Maximization (IM) problem, which seeks a set of seed users that maximize a post’s influence. However, there is no work on progressively learning how a post’s features affect its influence. In this article, we propose and study the problem of Adaptive Content-Aware Influence Maximization (ACAIM), which calls to find k features to form a post in each round so as to maximize the cumulative influence of those posts over all rounds. We solve ACAIM by applying, for the first time, an Online Learning to Rank (OLR) framework for IM purposes. We introduce the CATRID propagation model , which expresses how posts disseminate in a social network using click probabilities and post visibility criteria and develop a simulator that runs CATRID via a training-testing scheme based on real posts of the VK social network, so as to realistically represent the learning environment. We deploy three learners that solve ACAIM in an online (real-time) manner. We experimentally prove the practical suitability of our solutions via exhaustive experiments on multiple brands (operating as different case studies ) and several VK datasets; the best learner is evaluated on 45 separate case studies yielding convincing results.
- Open Access
- 13 November 2023
- Published by Association for Computing Machinery (ACM)
- p. 1-4
Efficient Nearest Neighbor Queries on Non-point Data
Nearest neighbor (NN) queries are ubiquitous in spatial databases, but have been studied mainly for point data. Inspired by recent work on indexing non-point objects for range queries, we propose a secondary partitioning scheme for space-partitioning indices, tailored to NN search. Our scheme classifies the contents of each primary partition into 16 secondary partitions, considering the begin and end of objects with respect to the spatial extent of the primary partition. Based on this, we design algorithms for both incremental NN and k-NN search that avoid duplicate results and skip unnecessary computations. We compare our scheme to the state-of-the-art indexing and find that it has a significant performance advantage.
- 13 November 2023
- Published by Association for Computing Machinery (ACM)
- p. 1-4
Open AccessEfficient Nearest Neighbor Queries on Non-point Data
Nearest neighbor (NN) queries are ubiquitous in spatial databases, but have been studied mainly for point data. Inspired by recent work on indexing non-point objects for range queries, we propose a secondary partitioning scheme for space-partitioning indices, tailored to NN search. Our scheme classifies the contents of each primary partition into 16 secondary partitions, considering the begin and end of objects with respect to the spatial extent of the primary partition. Based on this, we design algorithms for both incremental NN and k-NN search that avoid duplicate results and skip unnecessary computations. We compare our scheme to the state-of-the-art indexing and find that it has a significant performance advantage.
- 31 August 2023
- Published by Springer Nature
- p. 293-311
Databases and Information Systems: Contributions from ADBIS 2023 Workshops and Doctoral Consortium
- 31 August 2023
- Published by Springer Nature
- p. 293-311
Databases and Information Systems: Contributions from ADBIS 2023 Workshops and Doctoral Consortium
- 24 July 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Knowledge and Data Engineering
- Vol. 36 (3) , 1341-1355
Two-Layer Space-Oriented Partitioning for Non-Point Data
Non-point spatial objects (e.g., polygons, linestrings, etc.) are ubiquitous. We study the problem of indexing non-point objects in memory for range queries and spatial intersection joins. We propose a secondary partitioning technique for space-oriented partitioning indices (e.g., grids), which improves their performance significantly, by avoiding the generation and elimination of duplicate results. Our approach is easy to implement and can be used by any space-partitioning index to significantly reduce the cost of range queries and intersection joins. In addition, the secondary partitions can be processed independently, which makes our method appropriate for distributed and parallel indexing. Experiments on real datasets confirm the advantage of our approach against alternative duplicate elimination techniques and data-oriented state-of-the-art spatial indices. We also show that our partitioning technique, paired with optimized partition-to-partition join algorithms, typically reduces the cost of spatial joins by around 50%.
- 24 July 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Knowledge and Data Engineering
- Vol. 36 (3) , 1341-1355
Two-Layer Space-Oriented Partitioning for Non-Point Data
Non-point spatial objects (e.g., polygons, linestrings, etc.) are ubiquitous. We study the problem of indexing non-point objects in memory for range queries and spatial intersection joins. We propose a secondary partitioning technique for space-oriented partitioning indices (e.g., grids), which improves their performance significantly, by avoiding the generation and elimination of duplicate results. Our approach is easy to implement and can be used by any space-partitioning index to significantly reduce the cost of range queries and intersection joins. In addition, the secondary partitions can be processed independently, which makes our method appropriate for distributed and parallel indexing. Experiments on real datasets confirm the advantage of our approach against alternative duplicate elimination techniques and data-oriented state-of-the-art spatial indices. We also show that our partitioning technique, paired with optimized partition-to-partition join algorithms, typically reduces the cost of spatial joins by around 50%.
- Open Access
Two-layer Space-oriented Partitioning for Non-point Data
Non-point spatial objects (e.g., polygons, linestrings, etc.) are ubiquitous. We study the problem of indexing non-point objects in memory for range queries and spatial intersection joins. We propose a secondary partitioning technique for space-oriented partitioning indices (e.g., grids), which improves their performance significantly, by avoiding the generation and elimination of duplicate results. Our approach is easy to implement and can be used by any space-partitioning index to significantly reduce the cost of range queries and intersection joins. In addition, the secondary partitions can be processed independently, which makes our method appropriate for distributed and parallel indexing. Experiments on real datasets confirm the advantage of our approach against alternative duplicate elimination techniques and data-oriented state-of-the-art spatial indices. We also show that our partitioning technique, paired with optimized partition-to-partition join algorithms, typically reduces the cost of spatial joins by around 50%.
Open AccessTwo-layer Space-oriented Partitioning for Non-point Data
Non-point spatial objects (e.g., polygons, linestrings, etc.) are ubiquitous. We study the problem of indexing non-point objects in memory for range queries and spatial intersection joins. We propose a secondary partitioning technique for space-oriented partitioning indices (e.g., grids), which improves their performance significantly, by avoiding the generation and elimination of duplicate results. Our approach is easy to implement and can be used by any space-partitioning index to significantly reduce the cost of range queries and intersection joins. In addition, the secondary partitions can be processed independently, which makes our method appropriate for distributed and parallel indexing. Experiments on real datasets confirm the advantage of our approach against alternative duplicate elimination techniques and data-oriented state-of-the-art spatial indices. We also show that our partitioning technique, paired with optimized partition-to-partition join algorithms, typically reduces the cost of spatial joins by around 50%.
- 22 February 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Photovoltaics
- Vol. 13 (3) , 461-466
Data-Driven Soiling Detection in PV Modules
Soiling is the accumulation of dirt in solar panels that leads to a decreasing trend in solar energy yield and may be the cause of vast revenue losses. The effect of soiling can be reduced by washing the panels, which is, however, a procedure of non-negligible cost. Moreover, soiling monitoring systems are often unreliable or very costly. We study the problem of estimating the soiling ratio in photovoltaic (PV) modules, i.e., the ratio of the real power output to the power output that would be produced if solar panels were clean. A key advantage of our algorithms is that they estimate soiling, without needing to train on labeled data, i.e., periods of explicitly monitoring the soiling in each park, and without relying on generic analytical formulas that do not take into account the peculiarities of each installation. We consider as input a time series comprising a minimum set of measurements that are available to most PV park operators. Our experimental evaluation shows that we significantly outperform current state-of-the-art methods for estimating soiling ratio.
- 22 February 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Photovoltaics
- Vol. 13 (3) , 461-466
Data-Driven Soiling Detection in PV Modules
Soiling is the accumulation of dirt in solar panels that leads to a decreasing trend in solar energy yield and may be the cause of vast revenue losses. The effect of soiling can be reduced by washing the panels, which is, however, a procedure of non-negligible cost. Moreover, soiling monitoring systems are often unreliable or very costly. We study the problem of estimating the soiling ratio in photovoltaic (PV) modules, i.e., the ratio of the real power output to the power output that would be produced if solar panels were clean. A key advantage of our algorithms is that they estimate soiling, without needing to train on labeled data, i.e., periods of explicitly monitoring the soiling in each park, and without relying on generic analytical formulas that do not take into account the peculiarities of each installation. We consider as input a time series comprising a minimum set of measurements that are available to most PV park operators. Our experimental evaluation shows that we significantly outperform current state-of-the-art methods for estimating soiling ratio.