Contact tracing of COVID-19 and healthcare operations in Hong Kong

Abstract

We explore the structure of COVID-19 infection relation in a group-based approach. In each study, instead of analysing all patients who fall within a period as other epidemic models suggest, we only consider patients who present in the same group. We begin by collecting Hong Kong’s historical COVID-19 data and construct the infection relationship between patients using directed graphs. Since we restrict the in-degree to be 1, i.e. each patient is infected by a previous patient, the relation has a random recursive tree shape. Scale-free power-law has long been used to demonstrate the connectivities between vertices in a social contact network. We can intuitively find that a lot of patients do not further contaminate others and only a few patients spread the disease widely. Therefore, we use the number of infections of each patient to fit the power-law distribution and prove the data set follows the power-law distribution. Referring to the depth of a random graph, we formulate the equations in order to compute the probability that outbreaks terminate in different layers. Besides, we derive other equations that determine the probability of different cluster sizes. We show the theoretical result computed by the above equations gives a similar output with the simulation result. It allows us to gain more insights to understand the degree and trend of the outbreak within a group. We add a chapter which introduces another research topic about the effect on the service time by the workload in a Chinese Medicine clinic. Following the similar study of Kc and Terwiesch [1], we show that the rise in instant workload can motivate the workers which boosts the working efficiency and shortens the consultation time. But a consistent and intense workload leads to fatigue of labours and reduces the service rate.

Date of Award	2021
Original language	English
Awarding Institution	The Hong Kong University of Science and Technology
Supervisor	Jin QI (Supervisor)