The COVID-19 pandemic has prompted many countries to implement a mixture of traditional and novel outbreak control measures. This led to changes in human behaviour and the availability of innovative data sources. In this thesis, I integrated different outbreak surveillance data of COVID-19 cases in Singapore with statistical and mathematical modelling tools to understand the transmission dynamics of SARS- CoV-2 and its implications for outbreak control. Outbreak control measures are often applied in combination but the effectiveness of each measure is seldom independently evaluated. In a retrospective analysis, I used granular epidemiological investigation data and showed that the effectiveness of contact tracing was dependent on the effectiveness of case finding. Furthermore, with the strict quarantine of incoming travellers, the number of imported cases in Singapore in the second half of 2020 was three times higher than that at the start of the outbreak but the effective reproduction number remained below 1. I also found that the outbreak metric on the proportion of cases with no known infectors among all notified cases is not always reflective of the proportion of missed infections among all infections. In 2022, the relaxation of mainland China’s ‘zero-COVID’ strategy led to a surge of cases within China. Using the same dataset as the previous study but focusing on the imported cases arriving from mainland China to Singapore, I analysed the outbreak trajectory in China in real time. I found that the outbreak in China peaked in mid-December 2022 and, together with no apparent risk from novel strains, helped policymakers in Singapore to decide against reactive border control measures. With the emergence of new SARS-CoV-2 variants, there was an increase in observed cases within a short period which could be attributed to a decrease in the generation interval, often proxied by the serial interval. Thus, I also performed a real-time analysis but did not identify a large difference of more than one-day reduction in the Delta variant serial intervals as compared with the wild-type SARS-CoV-2. I further discussed how this finding could be attributed to the small sample size of less than 50 transmission pairs in each study period and could affect the power to detect changes, if any. As a follow-up analysis, I developed a simulation framework to sample transmission pairs and studied the power to detect changes in the generation and serial intervals under varying pathogen biology, outbreak control measures, contact patterns and epidemic dynamics. For a decrease of 0–1.4 days in the incubation period of the Delta variant reported in the literature, I found that a one-day reduction in the serial interval of the Delta variant was unlikely. Overall, a sample size of at least 100 transmission pairs would be required to provide 30–70% power to detect a one-day change in generation and serial intervals. Scenario analysis using outbreak simulation models is also useful when planning for the resumption of large-scale events amidst potential threats of new variants that are more transmissible. Using high-resolution temporal contact networks on cruises, I estimated that mask-wearing interventions, in addition to baseline measures of case isolation and physical distancing, would further reduce the outbreak size by 50% after accounting for the periods of interaction in dining and sports settings when passengers are not wearing masks. Also, the risk of a large outbreak was reduced when regular testing of passengers prior to departure and halfway through the event was implemented without having to wear a mask. Building on the temporal data collected from the cruises and from other studies, I explored the time-varying network properties in cruises, communities, high schools, hospitals and workplaces. The type of contacts that tend to be retained over consecutive timesteps varied across different settings. As the risk of transmission increases with longer contact duration, this implies that outbreak control measures have to be calibrated across each setting. Furthermore, as the terms ‘superspreaders’ and individuals driving ‘superspreading events’ are often used interchangeably in the literature, I classified individuals by ranking their connectivity over time. I found that less than 10% of the population in each network was consistently identified as being highly connected, and are potential ‘superspreaders’ if infectious. Instead, most of the population was highly connected for short periods and could drive ‘superspreading events’ if infectious. Overall, I performed a retrospective analysis of the effectiveness of outbreak control measures, real-time analyses of the epidemiology of SARS-CoV-2, and predictive analyses of transmission dynamics in specific settings. Each study of this thesis helps to identify the strengths and weaknesses of the current surveillance system, and the work will help inform the future pandemic preparedness and response policies in Singapore and across the world.