SCRI Research: Predicts Omicron onset risk in Hong Kong and finds Omicron spread can be better controlled by strict and targeted measures and booster vaccination
Professor John Shi, Director of SCRI, PolyU has proposed an extended Weight Kernel Density Estimation (WKDE) model to predict the spatiotemporal risk of COVID-19 symptom onset. Now this model has been adapted to understand how to track and control the spatiotemporal spread of Omicron variant.
Prediction of Onset Risk in Hong Kong
The prediction results show that from 11 to 17 January 2022, the overall risk of Omicron in Hong Kong continued to increase, and the growth rate was slower than that in South Africa. High-risk areas include: North Point, Kowloon Tong, Tuen Mun, Tai Po, Causeway Bay, Yau Tsim Mong, Tsuen Wan, Tseung Kwan O, etc. The deployment of vaccines and testing resources should be strengthened for these areas, and high-risk areas are concentrated in densely populated and high-traffic areas. The team calls on citizens to temporarily reduce non-essential travel in these areas. In addition, the forecast confirms that the recent tightening of social distancing measures by the government and the reduction in travel of Hong Kong citizens (according to Apple Mobility Trends Reports) have played a positive role in slowing the spatiotemporal spread of Omicron.
Fig. 1. (a) The symptom onset risk prediction in Hong Kong from 11 to 17 January 2022. (b) The predicted onset risk with and without the tighting of social distancing measures from 7 January 2022.
Omicron spread can be better controlled by strict and targeted measures and booster vaccination
Summary. A study led by Prof. Wenzhong Shi from the Otto Poon Charitable Foundation Smart Cities Research Institute (SCRI) at The Hong Kong Polytechnic University (PolyU) provides the first information on how to track and control the spatiotemporal spread of SARS-CoV-2 Omicron variant in South Africa. It is found that compared with current Alert Levels 1-4 in all provinces, the imposition of lockdown in the high-onset-risk Gauteng together with the Alert Level 1 in other provinces has a higher potential to effectively control the spread of Omicron and could reduce the overall symptom onset risk by up to 15.34% from 26 Nov 2021 to 10 Dec 2021. Meanwhile, if the current daily vaccination speed in each province increased by 10 times, the daily overall onset risk was estimated to reduce by only up to 7.86% from 26 Nov 2021 to 10 Dec 2021. This highlights the necessity to vaccinate the 1.6 million people in South Africa who have been fully vaccinated over 6 months ago with the booster dose of vaccine as soon as possible. This research is currently under peer review for publication. [Research article about this study]
Method and findings. The province-level WKDE model made the onset risk prediction based on the following data: i) Daily human mobility, ii) time-varying vaccination rates and vaccination efficiency, iii) daily COVID-19 effective reproductive number R, iv) weekly respiratory pathogens surveillance reports, v) weekly levels of SARS-CoV-2 in wastewater treatment plants, vi) weekly COVID-19 cases admitted to sentinel hospital surveillance sites, vii) the weekly percentage testing positive, and viii) daily social distancing level at the province scale. The model achieved over 80% accuracy in the onset risk prediction for the following 7 days.
Based on the onset risk prediction results, the actual spatiotemporal spread of Omicron in South Africa was analyzed, and the Omicron spread under different scenarios with different epidemic alert levels and vaccination rate levels was further simulated. It was found that:
i) The spatiotemporal spread was relatively slow during the first stage and following the emergence of Omicron in Gauteng. The spatial spread of Omicron accelerated after it had become the dominant variant, and continued to spread from Gauteng to the neighboring provinces and main transport nodes.
ii) Compared with current Alert Levels 1-4 in all provinces, the imposition of Alert Level 5 (lockdown) in the high-onset-risk Gauteng together with Alert Level 1 in other 8 provinces had a higher potential to effectively control the spread of Omicron. This recommended strict and targeted measure was estimated to be able to reduce the overall symptom onset risk by up to 15.34% from 26 Nov 2021 to 10 Dec 2021, and, moreover, it can reduce the spread of the Omicron epidemic in the provinces where main international airports are located to other parts of the world.
iii) Due to declining vaccine efficiency over time, even if the daily vaccination rates in each province increased by 10 times since 26 Nov 2021, the daily overall onset risk was only reduced by 0.34%-7.86% by 10 Dec 2021.
‘It is important to effectively control the human mobility from high-onset-risk area to other areas by lockdown in high-risk areas,’ said Prof. Shi. ‘It is also noted that, by implementing general gathering control measures in other areas, the impact on social and economic activities in other regions can be reduced. Boosters need to be vaccinated to strengthen the control of the Omicron epidemic. In sum, our study shows that the Omicron outbreaks could be better controlled through targeted strict measures and the booster vaccination.’
(a)
(b)
Fig. 2. The risk of COVID-19 symptom onset in the five epidemic alert scenarios (i.e., the current Alert Level 1 in all 9 provinces, Alert Level 2 in all 9 provinces, Alert Level 3 in all 9 provinces, Alert Level 4 in all 9 provinces, and Alert Level 5 in Gauteng together with Alert Level 1 in other 8 provinces) from November 25th, 2021 to December 10th, 2021.
(a)
(b)
Fig. 3. The risk of COVID-19 symptom onset with the current daily vaccination rates, 5 times the vaccination rates, 10 times the vaccination rate at the Alert Level 5 for Gauteng together with Alert Level 1 for the remaining 8 provinces from November 26th, 2021 to December 10th, 2021.
Related works: spatial prediction of COVID-19 onset risk
Predicting the risk of COVID-19 is the key to combating the pandemic worldwide. Most existing COVID-19 risk prediction methods focus on confirmed cases. However, people are more infectious in the days around and following the symptom onset (e.g., fever or cough), and there is a spatially variant delay of 4~5 days on average from the onset to diagnosis. Therefore, predicting the spatiotemporal risk of COVID-19 onset, which is different from the risk in terms of confirm cases, is essential for timely anti-epidemic measures.
The team of Prof. Shi developed an extended WKDE model to predict the spatiotemporal risk of COVID-19 onset within 14 days. The model was used for evaluating the effect of Wuhan lockdown in reducing the COVID-19 risk in other cities in China. The lockdown was found to delay the arrival of the COVID-19 onset risk peak for 1–2 days and lower risk peak values in other Chinese cities. The decrease of onset risk was more than 8% in over 40% of Chinese cities, and was up to 21.3%. Lockdown was the most effective in areas with medium onset risk before the lockdown. [Research article about this study]
The extended WKDE model was further developed to predicting the onset risk within Hong Kong. Based on the prediction result of the model, a spatial and dynamic solution based on the community-scale COVID-19 onset risk prediction result was also developed for precisely allocating COVID-19 vaccines to different areas and population groups in Hong Kong. [Research article about this study]
A further improved model was also used to evaluate the anti-epidemic measure in Taiwan. When the COVID-19 spread in Taiwan rebounded in May 2021, the epidemic alert in entire Taiwan rose to Level 3 (closing business places and public venues). However, the study found that compared with Level 3 Alert in entire Taiwan, Level 4 Alert (lockdown) in Taipei and New Taipei with the highest onset risk and Level 2 Alert in the rest of Taiwan (re-open venues, gather control) can better control the epidemic and reduce onset risk of up to 91.36%. Also, Increasing the daily vaccination rate in each district by up to 5 to 10 times would further reduce the onset risk by 6.07% to 62.22%. [Research article about this study]
Acknowledgements
This study was supported by Otto Poon Charitable Foundation Smart Cities Research Institute, The Hong Kong Polytechnic University (Work Program: CD03), and National Key R&D Program of China (2019YFB2103102).
About the Smart Cities Research Institute (SCRI), PolyU
Otto Poon Charitable Foundation Smart Cities Research Institute (SCRI) provides an interdisciplinary platform for PolyU’s experts to develop an international leading area in Smart Cities by capitalizing on existing interdisciplinary research strengths, including but not limited to departments of Faculty of Construction and Environment, Faculty of Engineering, Faculty of Applied Science and Textiles, and Faculty of Business with including other research institutes. To respond to the current gap of the unique traffic characteristics of Hong Kong, the SCRI first initiates a pilot research on smart mobility. As an internationally leading center of excellence in smart mobility, the proposed research framework aims to develop a three-year strategic plan, including four research initiatives.
Media enquiries
Please contact Prof. John Shi via lswzshi@polyu.edu.hk .
.png?bc=ffffff&h=202&w=385&hash=2E56BE36C3A9DF2111EB9AE94246ECB8)
