Eurosurveillance - Volume 29, Issue 38, 19 September 2024

The wide application of machine learning (ML) holds great potential to improve public health by supporting data analysis informing policy and practice. Its application, however, is often hampered by data fragmentation across organisations and strict regulation by the General Data Protection Regulation (GDPR). Federated learning (FL), as a decentralised approach to ML, has received considerable interest as a means to overcome the fragmentation of data, but it is yet unclear to which extent this approach complies with the GDPR.

Our aim was to understand the potential data protection implications of the use of federated learning for public health purposes.

Building upon semi-structured interviews (n = 14) and a panel discussion (n = 5) with key opinion leaders in Europe, including both FL and GDPR experts, we explored how GDPR principles would apply to the implementation of FL within public health.

Whereas this study found that FL offers substantial benefits such as data minimisation, storage limitation and effective mitigation of many of the privacy risks of sharing personal data, it also identified various challenges. These challenges mostly relate to the increased difficulty of checking data at the source and the limited understanding of potential adverse outcomes of the technology.

Since FL is still in its early phase and under rapid development, it is expected that knowledge on its impracticalities will increase rapidly, potentially addressing remaining challenges. In the meantime, this study reflects on the potential of FL to align with data protection objectives and offers guidance on GDPR compliance.

Brucellosis is a bacterial zoonosis causing severe illness in humans and animals and leading to economic losses in the livestock production in Türkiye and other endemic countries.

We aimed at investigating genomic differences of Brucella isolates from animals and humans in Türkiye.

We used whole genome sequencing (WGS) to assess the genetic diversity of Brucella isolates from 41 provinces in Türkiye and compared with isolates from other countries. We applied allele-based typing and core genome single nucleotide polymorphism (cgSNP) determination.

Of the 106 Turkish Brucella isolates included, 57 were B. abortus and 49 were B. melitensis. One B. melitensis and two B. abortus isolates were identified as vaccine strains. Most (n = 55) B. abortus isolates clustered in three major branches, with no spatial discernible pattern. Of the B. melitensis isolates, 48 were assigned to the Eastern Mediterranean lineage with no discernible patterns between host species, location and sampling date. The Turkish isolates clustered with isolates from neighbouring countries such as Greece and Syria, but some also with isolates from human patients in European countries, like Germany, Norway and Sweden, suggesting that the source may be travel-related.

Several B. melitensis and B. abortus lineages are circulating in Türkiye. To decrease the prevalence and prevent brucellosis in animals and humans, stricter control measures are needed, particularly in areas where humans and animals have close contact. Furthermore, illegal transportation of animals across borders should be more closely controlled and regulated.

In the Netherlands, HIV pre-exposure prophylaxis (PrEP) has been available since 2019. However, the extent of PrEP use prior to HIV diagnosis and development of PrEP-resistance-associated mutations (RAMs) is not known.

We assessed prior PrEP use and potential transmission of PrEP RAMs among men who have sex with men (MSM) and transgender persons (TGP) with a new HIV diagnosis in the Netherlands.

Data on prior PrEP use between 1 January 2018 and 31 December 2022 were available from the Dutch national ATHENA cohort. We assessed proportion of prior PrEP use, detected PrEP associated RAMs and assessed potential onward transmission of RAMs between 2010 and 2022 using a maximum likelihood tree.

Data on prior PrEP use were available for 583/1,552 (36.3%) individuals, with 16% (94/583) reporting prior PrEP use. In 489 individuals reporting no prior PrEP use, 51.5% did not use PrEP due to: low HIV-risk perception (29%), no access (19.1%), personal preference (13.1%), and being unaware of PrEP (19.1%). For PrEP users, 13/94 (13.8%) harboured a M184V/I mutation, of whom two also harboured a K65R mutation. In people with a recent HIV infection, detection of PrEP RAMs increased from 0.23% (2/862) before 2019 to 4.11% (9/219) from 2019. We found no evidence of onward transmission of PrEP RAMs.

The prevalence of PrEP-associated RAMs has increased since PrEP became available in the Netherlands. More widespread access to PrEP and retaining people in PrEP programmes when still at substantial risk is crucial to preventing new HIV infections.