Joana Félix China
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Joana Félix China has a total of 16 co-authors. Here are the top co-authors by number of shared publications:
- G. Niklas Norén
Uppsala Monitoring Centre, Uppsala, Sweden
5 shared publications - Daniele Sartori
Uppsala Monitoring Centre, Uppsala, Sweden
2 shared publications - Florence van Hunsel
Netherlands Pharmacovigilance Centre Lareb, MH’s-Hertogenbosch, The Netherlands
2 shared publications - Jim W. Barrett
Uppsala Monitoring Centre, Uppsala, Sweden
2 shared publications - Joep Scholl
Netherlands Pharmacovigilance Centre Lareb, MH’s-Hertogenbosch, The Netherlands
2 shared publications - Johan Ellenius
Uppsala Monitoring Centre, Uppsala, Sweden
2 shared publications
Latest Publications
- Open Access
- 23 October 2026
- Published by Springer Nature in Drug Safety
- p. 1-13
Toward a Framework for Case Identification in Pharmacovigilance: Not All Reports are Created Equal
- 23 October 2026
- Published by Springer Nature in Drug Safety
- p. 1-13
Open AccessToward a Framework for Case Identification in Pharmacovigilance: Not All Reports are Created Equal
- Open Access
- 9 October 2026
- Published by JMIR Publications Inc. in JMIR AI
- Vol. 5, e95282
A Scalable Predictive Modeling Approach to Duplicate Detection in Pharmacovigilance: Algorithm Development and Validation Study
Background: Adverse event reports referring to the same case but treated as independent can negatively impact statistical analysis and may mislead clinical assessment. Pharmacovigilance relies on large databases of adverse event reports to discover potential new causal associations. Their size necessitates computational methods to identify duplicates at scale. The current state-of-the-art is statistical record linkage which outperforms rule-based approaches. In particular, vigiMatch is in routine use for VigiBase, the World Health Organization global database of adverse event reports, and represents the first statistical duplicate detection approach in pharmacovigilance deployed at scale. Originally developed for both drugs and vaccines, its application to vaccines has been limited due to inconsistent performance across countries. Objective: This study aims to advance the state-of-the-art for duplicate detection in large-scale pharmacovigilance databases and achieve more consistent performance across adverse event reports from different countries. Methods: This paper extends vigiMatch from probabilistic record linkage to predictive modeling, refining features for drugs, vaccines, and adverse events using country-specific reporting rates, extracting dates from free text, and training separate support vector machine classifiers for drugs and vaccines. Recall was evaluated using 5 independent reference sets. Precision was assessed by annotating random selections of report pairs classified as duplicates. Results: Precision for the new method was 92% for vaccines and 54% for drugs, compared with 41% for a previous generation method. Recall ranged from 80% to 85% across test sets for vaccines and from 40% to 86% for drugs, compared with 24%-53% for the comparator method. Conclusions: Predictive modeling, use of free text, and country-specific features advance state-of-the-art for duplicate detection in pharmacovigilance.
- 9 October 2026
- Published by JMIR Publications Inc. in JMIR AI
- Vol. 5, e95282
Open AccessA Scalable Predictive Modeling Approach to Duplicate Detection in Pharmacovigilance: Algorithm Development and Validation Study
Background: Adverse event reports referring to the same case but treated as independent can negatively impact statistical analysis and may mislead clinical assessment. Pharmacovigilance relies on large databases of adverse event reports to discover potential new causal associations. Their size necessitates computational methods to identify duplicates at scale. The current state-of-the-art is statistical record linkage which outperforms rule-based approaches. In particular, vigiMatch is in routine use for VigiBase, the World Health Organization global database of adverse event reports, and represents the first statistical duplicate detection approach in pharmacovigilance deployed at scale. Originally developed for both drugs and vaccines, its application to vaccines has been limited due to inconsistent performance across countries. Objective: This study aims to advance the state-of-the-art for duplicate detection in large-scale pharmacovigilance databases and achieve more consistent performance across adverse event reports from different countries. Methods: This paper extends vigiMatch from probabilistic record linkage to predictive modeling, refining features for drugs, vaccines, and adverse events using country-specific reporting rates, extracting dates from free text, and training separate support vector machine classifiers for drugs and vaccines. Recall was evaluated using 5 independent reference sets. Precision was assessed by annotating random selections of report pairs classified as duplicates. Results: Precision for the new method was 92% for vaccines and 54% for drugs, compared with 41% for a previous generation method. Recall ranged from 80% to 85% across test sets for vaccines and from 40% to 86% for drugs, compared with 24%-53% for the comparator method. Conclusions: Predictive modeling, use of free text, and country-specific features advance state-of-the-art for duplicate detection in pharmacovigilance.
- Open Access
Towards a Framework for Case Identification in Pharmacovigilance: Not All Reports are Created Equal
Background: Retrieval of adverse event reports based on coded drug-event co-occurrence enables large-scale pharmacovigilance analyses, but yields candidate reports rather than validated cases, risking misinterpretation if used alone. Aim: To develop and apply a framework for identification and characterization of clinically meaningful case series in pharmacovigilance. Methods: We conducted two case studies. The first developed and refined the framework in an information-rich setting, focusing on drug-induced impulsivity across selected drugs; the second tested its applicability in a more routine, information-poor setting, focusing on drug-induced suicidality. Results: In Case 1, non-relevant reports were frequent for drugs with uncertain evidence and negative controls (≈20-40%) compared to drugs with established causal roles (4%). The emerging framework assessed relevance based on exposure, event, drug-event relationship, and population. For suspected adverse drug reactions, relevant reports were further characterized by reporter suspicion and evidentiary qualifiers supporting or refuting causality; higher suspicion was associated with more supportive qualifiers. Applied to Case 2, the framework ruled out 69% of reports as non-relevant but highlighted substantial non-assessability (17%). Conclusions: In pharmacovigilance, retrieval is not equivalent to case identification. Relevance is question-specific and shaped by how reports are captured, processed, and retrieved. This can be especially critical for emerging or bias-prone safety questions. Transparent and reproducible case definition and adjudication are essential for interpretable analyses. Graphical Code-based retrieval of drug–event co-occurrence yields a heterogeneous set of candidate reports that does not directly correspond to a clinically meaningful case series. The proposed framework separates case identification from retrieval by adjudicating relevance across exposure, event, relationship, and context, ruling out non-relevant or non-assessable reports. Relevant reports are further qualified by reporter suspicion and evidentiary features, enabling more interpretable pharmacovigilance analyses.
Open AccessTowards a Framework for Case Identification in Pharmacovigilance: Not All Reports are Created Equal
Background: Retrieval of adverse event reports based on coded drug-event co-occurrence enables large-scale pharmacovigilance analyses, but yields candidate reports rather than validated cases, risking misinterpretation if used alone. Aim: To develop and apply a framework for identification and characterization of clinically meaningful case series in pharmacovigilance. Methods: We conducted two case studies. The first developed and refined the framework in an information-rich setting, focusing on drug-induced impulsivity across selected drugs; the second tested its applicability in a more routine, information-poor setting, focusing on drug-induced suicidality. Results: In Case 1, non-relevant reports were frequent for drugs with uncertain evidence and negative controls (≈20-40%) compared to drugs with established causal roles (4%). The emerging framework assessed relevance based on exposure, event, drug-event relationship, and population. For suspected adverse drug reactions, relevant reports were further characterized by reporter suspicion and evidentiary qualifiers supporting or refuting causality; higher suspicion was associated with more supportive qualifiers. Applied to Case 2, the framework ruled out 69% of reports as non-relevant but highlighted substantial non-assessability (17%). Conclusions: In pharmacovigilance, retrieval is not equivalent to case identification. Relevance is question-specific and shaped by how reports are captured, processed, and retrieved. This can be especially critical for emerging or bias-prone safety questions. Transparent and reproducible case definition and adjudication are essential for interpretable analyses. Graphical Code-based retrieval of drug–event co-occurrence yields a heterogeneous set of candidate reports that does not directly correspond to a clinically meaningful case series. The proposed framework separates case identification from retrieval by adjudicating relevance across exposure, event, relationship, and context, ruling out non-relevant or non-assessable reports. Relevant reports are further qualified by reporter suspicion and evidentiary features, enabling more interpretable pharmacovigilance analyses.
- Open Access
- 13 March 2026
- Published by JMIR Publications Inc. in JMIR Preprints
Duplicate detection in pharmacovigilance: a scalable predictive modelling approach (Preprint)
BACKGROUND Unlinked adverse event reports referring to the same case impede statistical analysis and may mislead clinical assessment. Pharmacovigilance relies on large databases of adverse event reports to discover potential new causal associations, and computational methods are required to identify duplicates at scale. Current state-of-the-art is statistical record linkage which outperforms rule-based approaches. In particular, vigiMatch is in routine use for VigiBase, the WHO global database of adverse event reports, and represents the first statistical duplicate detection approach in pharmacovigilance deployed at scale. Originally developed for both medicines and vaccines, its application to vaccines has been limited due to inconsistent performance across countries. OBJECTIVE To advance state-of-the-art for duplicate detection in large-scale pharmacovigilance databases and achieve more consistent performance across adverse event reports from different countries. METHODS This paper extends vigiMatch from probabilistic record linkage to predictive modelling, refining features for medicines, vaccines, and adverse events using country-specific reporting rates, extracting dates from free text, and training separate support vector machine classifiers for medicines and vaccines. Recall was evaluated using 5 independent labelled test sets. Precision was assessed by annotating random selections of report pairs classified as duplicates. RESULTS Precision for the new method was 92% for vaccines and 54% for medicines, compared with 41% for the comparator method. Recall ranged from 80-85% across test sets for vaccines and from 40–86% for medicines, compared with 24–53% for the comparator method. CONCLUSIONS Predictive modeling, use of free text, and country-specific features advance state-of-the-art for duplicate detection in pharmacovigilance.
- 13 March 2026
- Published by JMIR Publications Inc. in JMIR Preprints
Open AccessDuplicate detection in pharmacovigilance: a scalable predictive modelling approach (Preprint)
BACKGROUND Unlinked adverse event reports referring to the same case impede statistical analysis and may mislead clinical assessment. Pharmacovigilance relies on large databases of adverse event reports to discover potential new causal associations, and computational methods are required to identify duplicates at scale. Current state-of-the-art is statistical record linkage which outperforms rule-based approaches. In particular, vigiMatch is in routine use for VigiBase, the WHO global database of adverse event reports, and represents the first statistical duplicate detection approach in pharmacovigilance deployed at scale. Originally developed for both medicines and vaccines, its application to vaccines has been limited due to inconsistent performance across countries. OBJECTIVE To advance state-of-the-art for duplicate detection in large-scale pharmacovigilance databases and achieve more consistent performance across adverse event reports from different countries. METHODS This paper extends vigiMatch from probabilistic record linkage to predictive modelling, refining features for medicines, vaccines, and adverse events using country-specific reporting rates, extracting dates from free text, and training separate support vector machine classifiers for medicines and vaccines. Recall was evaluated using 5 independent labelled test sets. Precision was assessed by annotating random selections of report pairs classified as duplicates. RESULTS Precision for the new method was 92% for vaccines and 54% for medicines, compared with 41% for the comparator method. Recall ranged from 80-85% across test sets for vaccines and from 40–86% for medicines, compared with 24–53% for the comparator method. CONCLUSIONS Predictive modeling, use of free text, and country-specific features advance state-of-the-art for duplicate detection in pharmacovigilance.
- Open Access
- 15 March 2025
- Published by Springer Nature in Drug Safety
- Vol. 48 (7) , 767-779
Narrative Search Engine for Case Series Assessment Supported by Artificial Intelligence Query Suggestions
Introduction Manual identification of case narratives with specific relevant information can be challenging when working with large numbers of adverse event reports (case series). The process can be supported with a search engine, but building search queries often remains a manual task. Suggesting terms to add to the search query could support assessors in the identification of case narratives within a case series. Objective The aim of this study is to explore the feasibility of identifying case narratives containing specific characteristics with a narrative search engine supported by artificial intelligence (AI) query suggestions. Methods The narrative search engine uses Best Match 25 (BM25) and suggests additional query terms from two word embedding models providing English and biomedical words to a human in the loop. We calculated the percentage of relevant narratives retrieved by the system (recall) and the percentage of retrieved narratives relevant to the search (precision) on an evaluation dataset including narratives from VigiBase, the World Health Organization global database of adverse event reports for medicines and vaccines. Exact-match search and BM25 search with the Relevance Model (RM3), an alternative way to expand queries, were used as comparators. Results The gold standard included 55/750 narratives labelled as relevant. Our narrative search engine retrieved on average 56.4% of the relevant narratives (recall), which is higher when compared with exact-match search (21.8%), without a significant drop in precision (54.5% to 43.1%). The recall is also higher as compared with RM3 (34.4%). Conclusions Our study demonstrates that a narrative search engine supported by AI query suggestions can be a viable alternative to an exact-match search and BM25 search with RM3, since it can facilitate the retrieval of additional relevant narratives during signal assessments.
- 15 March 2025
- Published by Springer Nature in Drug Safety
- Vol. 48 (7) , 767-779
Open AccessNarrative Search Engine for Case Series Assessment Supported by Artificial Intelligence Query Suggestions
Introduction Manual identification of case narratives with specific relevant information can be challenging when working with large numbers of adverse event reports (case series). The process can be supported with a search engine, but building search queries often remains a manual task. Suggesting terms to add to the search query could support assessors in the identification of case narratives within a case series. Objective The aim of this study is to explore the feasibility of identifying case narratives containing specific characteristics with a narrative search engine supported by artificial intelligence (AI) query suggestions. Methods The narrative search engine uses Best Match 25 (BM25) and suggests additional query terms from two word embedding models providing English and biomedical words to a human in the loop. We calculated the percentage of relevant narratives retrieved by the system (recall) and the percentage of retrieved narratives relevant to the search (precision) on an evaluation dataset including narratives from VigiBase, the World Health Organization global database of adverse event reports for medicines and vaccines. Exact-match search and BM25 search with the Relevance Model (RM3), an alternative way to expand queries, were used as comparators. Results The gold standard included 55/750 narratives labelled as relevant. Our narrative search engine retrieved on average 56.4% of the relevant narratives (recall), which is higher when compared with exact-match search (21.8%), without a significant drop in precision (54.5% to 43.1%). The recall is also higher as compared with RM3 (34.4%). Conclusions Our study demonstrates that a narrative search engine supported by AI query suggestions can be a viable alternative to an exact-match search and BM25 search with RM3, since it can facilitate the retrieval of additional relevant narratives during signal assessments.
- Open Access
- 20 January 2025
- Published by Springer Nature in Drug Safety
- Vol. 48 (4) , 401-413
Clinical Relatedness and Stability of vigiVec Semantic Vector Representations of Adverse Events and Drugs in Pharmacovigilance
Introduction Individual case reports are essential to identify and assess previously unknown adverse effects of medicines. On these reports, information on adverse events (AEs) and drugs are encoded in hierarchical terminologies. Encoding differences may hinder the retrieval and analysis of clinically related reports relevant to a topic of interest. Recent studies have explored the use of data-driven semantic vector representations to support analysis of pharmacovigilance data. Objective This study aims to evaluate the stability and clinical relatedness of vigiVec, a semantic vector representation for codes of AEs and drugs. Methods vigiVec is a published adaptation to pharmacovigilance of the publicly available Word2Vec model, applied to structured data instead of free text. It provides vector representations for MedDRA® Preferred Terms and WHODrug Global active ingredients, learned from reporting patterns in VigiBase, the WHO global database of adverse event reports for medicines and vaccines. For this study, a 20-dimensional Skip-gram architecture with window size 250 was used. Our evaluation focused on nearest neighbors identified by the cosine similarity of vigiVec vector representations. Clinical relatedness was measured through term intruder detection, whereby a medical doctor was tasked to identify a randomly selected term—the intruder—included among the four nearest neighbors to a specific AE or drug. Stability was measured as the average overlap in the ten nearest neighbors for each AE or drug, in repeated fittings of vigiVec. Results Among the ten nearest neighbors, 1.8 AEs on average belonged to the same MedDRA High Level Term (HLT; e.g., coagulopathies), and 1.3 drugs belonged to the same Anatomical Therapeutic Chemical level 3 (ATC-3; e.g., opioids). In the intruder detection task, when neighbors and intruders were both chosen from the same HLT, the intruder detection rate was 46%. When selected from different HLTs, it was 79%. By random chance, we should expect 20% (1 in 5). Corresponding rates for drugs were 42% in same ATC-3 and 65% in different ATC-3. The stability of nearest neighbors was 80% for AEs and 64% for drugs. Conclusion Nearest neighbors identified with vigiVec are stable and show high level of clinical relatedness. They are often from different parts of the existing hierarchies and complement these.
- 20 January 2025
- Published by Springer Nature in Drug Safety
- Vol. 48 (4) , 401-413
Open AccessClinical Relatedness and Stability of vigiVec Semantic Vector Representations of Adverse Events and Drugs in Pharmacovigilance
Introduction Individual case reports are essential to identify and assess previously unknown adverse effects of medicines. On these reports, information on adverse events (AEs) and drugs are encoded in hierarchical terminologies. Encoding differences may hinder the retrieval and analysis of clinically related reports relevant to a topic of interest. Recent studies have explored the use of data-driven semantic vector representations to support analysis of pharmacovigilance data. Objective This study aims to evaluate the stability and clinical relatedness of vigiVec, a semantic vector representation for codes of AEs and drugs. Methods vigiVec is a published adaptation to pharmacovigilance of the publicly available Word2Vec model, applied to structured data instead of free text. It provides vector representations for MedDRA® Preferred Terms and WHODrug Global active ingredients, learned from reporting patterns in VigiBase, the WHO global database of adverse event reports for medicines and vaccines. For this study, a 20-dimensional Skip-gram architecture with window size 250 was used. Our evaluation focused on nearest neighbors identified by the cosine similarity of vigiVec vector representations. Clinical relatedness was measured through term intruder detection, whereby a medical doctor was tasked to identify a randomly selected term—the intruder—included among the four nearest neighbors to a specific AE or drug. Stability was measured as the average overlap in the ten nearest neighbors for each AE or drug, in repeated fittings of vigiVec. Results Among the ten nearest neighbors, 1.8 AEs on average belonged to the same MedDRA High Level Term (HLT; e.g., coagulopathies), and 1.3 drugs belonged to the same Anatomical Therapeutic Chemical level 3 (ATC-3; e.g., opioids). In the intruder detection task, when neighbors and intruders were both chosen from the same HLT, the intruder detection rate was 46%. When selected from different HLTs, it was 79%. By random chance, we should expect 20% (1 in 5). Corresponding rates for drugs were 42% in same ATC-3 and 65% in different ATC-3. The stability of nearest neighbors was 80% for AEs and 64% for drugs. Conclusion Nearest neighbors identified with vigiVec are stable and show high level of clinical relatedness. They are often from different parts of the existing hierarchies and complement these.