Empirical Quantitative Analysis of COVID-19 Forecasting Models

Zhao, Yun; Wang, Yuqing; Liu, Junfeng; Xia, Haotian; Xu, Zhenni; Hong, Qinghang; Zhou, Zhiyang; Petzold, Linda R.

doi:10.1109/icdmw53433.2021.00069

Cited by 4 publications

(5 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The third stage (16)(17)(18)(19)(20)(21)(22)(23)(24)(25)(26)(27)(28)(29)(30)(31) On 16 December, the first batch of 5,025 sets of centralized isolation houses in Shaoxing was built, followed by the largest temporary quarantine site with 600 beds in Shangyu put into use on 17 December. During this stage, Shangyu conducted another four consecutive mass NAATs.…”

Section: Interventions In Shaoxingmentioning

confidence: 99%

“…The SEIAR model was fitted on dynamically changing data, including daily numbers of immediately confirmed cases, asymptomatic carriers and cases removed from the model. The limited-memory Broyden-Fletcher-Goldfarb-Shanno (L-BFGS) algorithm [21] was used to find the best estimation of unknown parameters by fitting them to find the zero solution of the equations. Epidemic curve fitting was then performed based on the actual number of cumulative infections.…”

Section: Covid-19 Transmission Modellingmentioning

confidence: 99%

See 1 more Smart Citation

Evaluation of phase-adjusted interventions for COVID-19 using an improved SEIR model

Jiang,

Gu,

Liu

et al. 2023

Epidemiol. Infect.

View full text Add to dashboard Cite

A local COVID-19 outbreak with two community clusters occurred in a large industrial city, Shaoxing, China, in December 2021 after serial interventions were imposed. We aimed to understand the reason by analysing the characteristics of the outbreak and evaluating the effects of phase-adjusted interventions. Publicly available data from 7 December 2021 to 25 January 2022 were collected to analyse the epidemiological characteristics of this outbreak. The incubation period was estimated using Hamiltonian Monte Carlo method. A well-fitted extended susceptible-exposed-infectious-recovered model was used to simulate the impact of different interventions under various combination of scenarios. There were 387 SARS-CoV-2-infected cases identified, and 8.3% of them were initially diagnosed as asymptomatic cases. The estimated incubation period was 5.4 (95% CI 5.2–5.7) days for all patients. Strengthened measures of comprehensive quarantine based on tracing led to less infections and a shorter duration of epidemic. With a same period of incubation, comprehensive quarantine was more effective in containing the transmission than other interventions. Our findings reveal an important role of tracing and comprehensive quarantine in blocking community spread when a cluster occurred. Regions with tense resources can adopt home quarantine as a relatively affordable and low-impact intervention measure compared with centralized quarantine.

show abstract

Section: Interventions In Shaoxingmentioning

confidence: 99%

Section: Covid-19 Transmission Modellingmentioning

confidence: 99%

Evaluation of phase-adjusted interventions for COVID-19 using an improved SEIR model

Jiang,

Gu,

Liu

et al. 2023

Epidemiol. Infect.

View full text Add to dashboard Cite

show abstract

“…BERT-based models like BERT (Kenton & Toutanova, 2019) and RoBERTa (Liu et al, 2019) are representative examples. These models have been applied to a diverse set of tasks, including disease prediction (Zhao et al, 2021), text classification (Wang et al, 2022b), time series analysis (Wang et al, 2022c), and more. However, the introduction of models like GPT-3 (Brown et al, 2020) marked a significant shift away from heavy reliance on extensive task-specific fine-tuning.…”

Section: Related Workmentioning

confidence: 99%

Towards Benchmarking and Improving the Temporal Reasoning Capability of Large Language Models

Tan¹,

Ng²

2023

Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)

View full text Add to dashboard Cite

Reasoning about time is essential for understanding the nuances of events described in natural language. Previous research on this topic has been limited in scope, characterized by a lack of standardized benchmarks that would allow for consistent evaluations across different studies. In this paper, we introduce TRAM, a temporal reasoning benchmark composed of ten datasets, encompassing various temporal aspects of events such as order, arithmetic, frequency, and duration, designed to facilitate a comprehensive evaluation of the temporal reasoning capabilities of large language models (LLMs). We conduct an extensive evaluation using popular LLMs, such as GPT-4 and Llama2, in both zero-shot and few-shot learning scenarios. Additionally, we employ BERT-based models to establish the baseline evaluations. Our findings indicate that these models still trail human performance in temporal reasoning tasks. It is our aspiration that TRAM will spur further progress in enhancing the temporal reasoning abilities of LLMs. Our data is available at https://github.com/EternityYW/TRAM-Benchmark.

show abstract

“…With the growing demand of long-term time series forecasting accuracy in various domains [8], [58], [60], [61], [62], [63], [64], [65], [66], traditional forecasting models, e.g. ARIMA [20], [21], SES [67], are no longer able to deal with more and more complicated forecasting situations.…”

Section: Appendix B Related Workmentioning

confidence: 99%

Respecting Time Series Properties Makes Deep Time Series Forecasting Perfect

Shen¹,

Wei²,

Wang³

2022

Preprint

View full text Add to dashboard Cite

How to handle time features shall be the core question of any time series forecasting model. Ironically, it is often ignored or misunderstood by deep-learning based models, even those baselines which are state-of-the-art. This behavior makes their inefficient, untenable and unstable. In this paper, we rigorously analyze three prevalent but deficient/unfounded deep time series forecasting mechanisms or methods from the view of time series properties, including normalization methods, multivariate forecasting and input sequence length. Corresponding corollaries and solutions are given on both empirical and theoretical basis. We thereby propose a novel time series forecasting network, i.e. RTNet, on the basis of aforementioned analysis. It is general enough to be combined with both supervised and self-supervised forecasting format. Thanks to the core idea of respecting time series properties, no matter in which forecasting format, RTNet shows obviously superior forecasting performances compared with dozens of other SOTA time series forecasting baselines in three real-world benchmark datasets. By and large, it even occupies less time complexity and memory usage while acquiring better forecasting accuracy. The source code is available at https://github.com/OrigamiSL/RTNet.

show abstract

Empirical Quantitative Analysis of COVID-19 Forecasting Models

Cited by 4 publications

References 20 publications

Evaluation of phase-adjusted interventions for COVID-19 using an improved SEIR model

Evaluation of phase-adjusted interventions for COVID-19 using an improved SEIR model

Towards Benchmarking and Improving the Temporal Reasoning Capability of Large Language Models

Respecting Time Series Properties Makes Deep Time Series Forecasting Perfect

Contact Info

Product

Resources

About