202 Chapter 9 have advantages and disadvantages, and may be more or less suitable for specific surveillance settings. Aggregated surveillance may be more appropriate if the number of cases is very large, or if trends are more important than individual cases. In contrast, case based surveillance may be better if more detailed information from cases is needed. Sophisticated (syndromic) disease surveillance systems can set thresholds for potential outbreak alarms. Finding the right threshold for outbreak alerts is of great importance, to prevent as many false positive and false negative alerts as possible11,16. Finding an appropriate threshold typically involves a combination of statistical analyses and consideration of the specific context, and can differ between different pathogens17. For example, the threshold for alarm for infectious diseases that are uncommon but severe, such as cholera or SARS, will be lower than for common, seasonal infectious diseases, like influenza17. Either way, human interpretation will be needed. Setting up a (syndromic) surveillance system that works sufficiently well (timely, large amount of data, reasonable alarm threshold) using traditional, automated and or big data sources may still prove to be both costly and highly complex11. In the first section of this thesis, two approaches were used to improve the surveillance of infectious diseases. Progress was made in the development of a sentinel SARI surveillance system in the Netherlands, using automated collection of proxy data, streamlining surveillance processes for efficiency and speed. Moreover, in this proof-of-concept study, it was demonstrated that these surveillance proxies could easily be collected from routine healthcare data. Additionally, the study in Chapter 3 found no evidence for the presence of a broader outbreak of mild hepatitis cases among children during the outbreak of 2022, although there were local differences. Finally, this study demonstrated the using summary-level laboratory data could be used for federated analysis in a safe and efficient manner. The advancements in data science and technology have created many new possibilities for understanding and managing viral infectious diseases, but there are also many remaining challenges. The first potential problem is disease surveillance in lower income countries, in which there are typically fewer automated data sources and digital opportunities4. To capture global and regional disease trends, global disease surveillance is needed. Second, (near) real-time data analysis, which is ideal for disease surveillance, requires robust methods for data sharing, analyses and communication with the public. Currently, there is no such system in place for hospital data in the Netherlands. The problem of data sharing is related to the third challenge: privacy. Very granular data can in some specific cases still be used to identify individuals, especially when linking the data to other data sources13. For example, if there is highly specific geographical information, with small age groups, there may only be one or two individuals in certain groups for analysis, which increases the risk of identification. There is a trade-off between improved surveillance and privacy of individuals4. Finally, it is essential to continuously validate data sources and ensure that data quality is sufficient for disease surveillance4,16. Google Flu Trends is a well-
RkJQdWJsaXNoZXIy MTk4NDMw