Abstract
This study aimed to predict seasonal influenza activity and detection of influenza outbreaks. Data of all registered cases (n = 53,526) of influenza-like illnesses (ILIs) from sentinel sites of healthcare centers in Iran were obtained from the FluNet web-based tool, World Health Organization (WHO), from 2010 to 2015. The status of the ILI activity was obtained from the FluNet and considered as the gold standard of the seasonal activity of influenza during the study period. The cumulative sum (CUSUM) as an outbreak detection method was used to predict the seasonal activity of influenza. Also, time series similarity between the ILI trend and CUSUM was assessed using the cross-correlogram. Of 7684 (14%) positive cases of influenza, about 71% were type A virus and 28% were type B virus. The majority of the outbreaks occurred in winter and autumn. Results of the cross-correlogram showed that there was a considerable similarity between time series graphs of the ILI cases and CUSUM values. However, the CUSUM algorithm did not have a good performance in the timely detection of influenza activity. Despite a considerable similarity between time series of the ILI cases and CUSUM algorithm in weekly lag, the seasonal activity of influenza in Iran could not be predicted by the CUSUM algorithm.
HIGHLIGHTS
-
•
Activity of influenza in Iran could not be predicted by CUSUM algorithm.
-
•
The majority of the outbreaks occurred in winter and autumn.
-
•
CUSUM algorithm did not have a good performance in detection of influenza outbreaks.
Keywords: CUSUM, influenza, Iran, outbreak, public health surveillance
1. INTRODUCTION
Influenza is one of the main causes of mortality around the world. Influenza as an epidemic-prone disease causes major economic effects due to hospitalization and absenteeism from work and school. Influenza causes 3–5 million cases annually. Influenza is a major concern of public health, with more than 250–500 thousand deaths in the world and more than 200 thousand hospitalizations in the United States [1–6].
Influenza is an acute viral infection which affects the nose, throat, bronchi, and sometimes the lungs. The infection usually lasts for a week and it has a sudden onset with symptoms such as high fever, muscle pain, headache, cough, sore throat, and weakness. The influenza virus is transmitted through respiratory droplets from person to person. There are three types of seasonal influenza viruses: A, B, and C virus. Influenza type A (H3N2) and A (H1N1) are now circulating among humans. Influenza outbreaks occur in the winter. The average duration of seasonal influenza is 15–16 weeks [4,7,8].
Accurate prediction of the seasonal activity of influenza is important for public health authorities because it is an epidemic-prone disease. The role of timely detection of influenza outbreaks is evident in effective implementation of interventions to control outbreaks. The implementation of syndromic surveillance systems, by using outbreak detection methods, enabled the identification of possible outbreaks of influenza. Influenza surveillance systems attempt to monitor aberrations of influenza-like illness (ILI) cases as a proxy of influenza activity. Seasonal activity of influenza could be detected early using outbreak detection methods/algorithms such as the cumulative sum (CUSUM) algorithm as an appropriate method [9–11]. However, there are many different tools for early detection of influenza-related outbreaks, which utilize both clinical and nonclinical data [11–20].
This study aimed to predict the seasonal influenza activity and detection of influenza outbreaks from 2010 to 2015 using the CUSUM algorithm.
2. MATERIALS AND METHODS
The data of all registered cases of ILI from sentinel sites of healthcare centers in Iran from January 2010 to December 2015, which were reported to the World Health Organization (WHO) on a weekly basis, were obtained from the FluNet web-based tool, WHO (http://www.who.int/influenza/gisrs_laboratory/flunet/en). An ILI case is defined as “a person with fever, cough, and acute respiratory syndrome.” During the study period, 53,526 cases of ILI from Iran were registered at FluNet [21].
The status of the ILI activity, including sporadic, endemic, and local outbreaks, was obtained from the FluNet and considered as the gold standard of the seasonal activity of influenza during the study period. Epidemiological profiles of the ILI cases and influenza cases (week of occurrence, season, year, and type of positive cases) were presented using descriptive statistics and a time series graph.
The CUSUM, as an outbreak detection method, was used to analyze the ILI data for early detection of the seasonal activity of influenza. The main reason for using CUSUM to apply ILI data in this work is its potential for better sensitivity to timely detection of outbreaks. The CUSUM statistics was estimated using Equation (1) [22–24]:
| (1) |
where Yt is the number of the ILI cases at weekt, σ is a standard deviation of the observed ILI cases during the study period, and CUSUMt − 1 is the estimated number 1 week ago, at week t − 1. The upper control limit for the CUSUM algorithm is considered as a criterion for an outbreak detection alarm while the CUSUM statistic is greater than the upper control limit and estimated by Equation (2) [25]:
| (2) |
where μ is the mean of the observed ILI cases during the study period, σ is the standard deviation, and h is a fixed coefficient, which ranges from 0.25 to 2.5 in this study. The rationale behind this range is to achieve the optimum parameters. Details on the CUSUM calculation using the Excel spreadsheet are described elsewhere [26].
Time series similarity was also assessed between the ILI trend and CUSUM using the cross-correlogram. In practice, the ability of CUSUM to predict influenza-related outbreaks was assessed by the degree of similarity between the ILI trend and CUSUM using the cross-correlogram. Staff of surveillance systems can monitor the trend of CUSUM using the cross-correlogram to detect any aberration from the usual trend of ILI data. Accordingly, CUSUSM, as an outbreak detection tool, predicts potential outbreaks that might be missed by observation of raw data of ILI as a proxy of influenza activity. It is evident that the main aim of this work is prediction and early detection of outbreaks, not forecasting of time series data on ILI or confirmed cases of influenza.
3. RESULTS
There were 53,526 ILI cases from 2010 to 2015. Of 7684 (14%) positive cases of influenza, 5497 (71%) cases were type A and 2187 (28%) cases were type B virus. Some 32% of ILI cases from sentinel sites in Iran were registered in 2015. Influenza type A was the most prevalent detected type in comparison with other types of confirmed cases in 2015. However, type B was the most prevalent type in 2011. Table 1 shows the frequency distribution of ILI and influenza cases by the influenza type and year of occurrence. The highest number of ILI cases occurred in 2015 and the lowest number, 4353 cases, was seen in 2012. Among the positive cases of ILI as influenza cases, type A virus was more prevalent in comparison with type B influenza virus (Table 1).
Table 1.
Frequency distribution of influenza-like illness and influenza cases by the influenza type and year of occurrence in Iran
| Year | ILI cases (N = 53,526) n (%) | Influenza cases by type | n (%)a |
|---|---|---|---|
| 2010 | 7062 (13.0) | Type A | 257 (45) |
| Type B | 310 (55) | ||
| Negative ILI cases | 6495 | ||
| 2011 | 10,055 (19.0) | Type A | 1102 (55) |
| Type B | 900 (45) | ||
| Negative ILI cases | 8053 | ||
| 2012 | 4353 (8.0) | Type A | 473 (80) |
| Type B | 120 (20) | ||
| Negative ILI cases | 3760 | ||
| 2013 | 8321 (15.5) | Type A | 289 (70) |
| Type B | 126 (30) | ||
| Negative ILI cases | 7906 | ||
| 2014 | 6719 (12.5) | Type A | 419 (61) |
| Type B | 263 (39) | ||
| Negative ILI cases | 6037 | ||
| 2015 | 17,016 (32.0) | Type A | 2957 (86) |
| Type B | 468 (14) | ||
| Negative ILI cases | 13,591 |
Percentage was calculated for the positive cases of ILI; ILI, influenza-like illness.
In the case of seasonal occurrence, more than 52% of the ILI cases occurred during winter, followed by 31% during autumn. Results revealed that influenza led to local outbreaks during 80 weeks in Iran and more than half of these outbreaks occurred in winter. About 52% of ILI cases occurred in winter according to registered data at FluNet. Influenza had a sporadic activity in Iran during 201 weeks from 2010 to 2015. Details on the distribution of ILI and influenza cases by the influenza type and ILI activity are shown in Table 2.
Table 2.
Frequency distribution of influenza-like illness (ILI) and influenza cases by the influenza type and ILI activity from 2010 to 2015 in Iran
| Seasons | ILI | Type of positive ILI cases | ILI activity | |||||
|---|---|---|---|---|---|---|---|---|
| A type | B type | A and B | Sporadic | Local outbreak | No report | No activity | ||
| Spring | 6456 (12.06) | 225 (4.09) | 452 (20.67) | 677 (8.81) | 68 (33.83) | 9 (11.25) | 4 (14.29) | 0 |
| Summer | 2374 (4.44) | 40 (0.73) | 54 (2.47) | 94 (1.22) | 69 (34.32) | 0 (0.00) | 13 (46.42) | 3 (75) |
| Autumn | 16,763 (31.32) | 2480 (45.12) | 727 (33.24) | 3207 (41.74) | 43 (21.39) | 25 (31.25) | 6 (21.42) | 1 (25) |
| Winter | 27,933 (52.19) | 2752 (50.06) | 954 (43.62) | 3706 (48.23) | 21 (10.44) | 46 (57.50) | 5 (17.75) | 0 |
| Total | 53,526 (100) | 5497 (100) | 2187 (100) | 7684 (100) | 201 (100) | 80 (100) | 28 (100) | 4 (100) |
Data are presented as n (%).
The results of the present study showed that there was a considerable similarity between time series graphs of ILI cases and CUSUM values at different values of fixed parameters. This finding reveals that CUSUM was able to detect potential outbreaks of influenza in a timely manner in comparison with routine monitoring of raw data on ILI. The corresponding graphs for fixed parameters with h = 2.5, 1.5, and 0.25 are shown in Figs. 1–3, respectively. However, the CUSUM algorithm did not have a good performance in timely detection of the seasonal activity of influenza.
Figure 1.
Time series graphs of the cumulative sum (CUSUM) algorithm using a fixed parameter (h = 2.5) and influenza-like illness (ILI) cases during 2010 and 2015
Figure 3.
Time series graphs of the cumulative sum (CUSUM) algorithm using a fixed parameter (h = 0.25) and influenza-like illness (ILI) cases during 2010 and 2015
Figure 2.
Time series graphs of the cumulative sum (CUSUM) algorithm using a fixed parameter (h = 1.5) and influenza-like illness (ILI) cases during 2010 and 2015
There was a considerable similarity between time series of the ILI cases and the CUSUM algorithm in weekly lag (Figs. 4 and 5). The results of the cross-correlogram showed that the highest coefficient of correlation between time series was seen at lag + 1 with 0.94, and the lowest magnitude was 0.83. This finding indicates that CUSUM was able to detect influenza-related outbreaks early in comparison with routine monitoring of raw data on ILI. However, there was a weekly lag between alarms of outbreak onset by CUSUM and the status of the ILI activity as the gold standard of the seasonal activity of influenza.
Figure 4.
Cross-correlation between the influenza-like illness (ILI) cases and cumulative sum (CUSUM) algorithm in weekly lag with fixed coefficient (h) = 0.25
Figure 5.
Cross-correlation between the influenza-like illness (ILI) cases and cumulative sum (CUSUM) algorithm in weekly lag with fixed coefficient (h) = 2
4. DISCUSSION
Sentinel surveillance of common respiratory infections is a useful approach because of the high prevalence of such diseases. In the case of influenza, WHO utilizes sentinel surveillance systems worldwide and registers both ILI cases and positive cases of influenza. It is evident that such surveillance systems have some weaknesses, especially regarding timely detection of influenza outbreaks. However, there are sentinel ILI reporting sites in Iran which report both ILI and influenza cases to WHO; this is in accordance with similar studies [27,28]. These studies have reported that the proportions of positive influenza among the ILI cases were 14.1% and 16.2%. Nevertheless, there are some studies that reported different proportions of positive influenza among the ILI cases [29,30].
Of 7684 (14%) positive cases of influenza, about 71% were type A and 28% were type B virus. Results of a similar study in Australia revealed that type A with 71% was more prevalent in comparison with other types of the positive cases of influenza [31].
This study demonstrated that the seasonal activity of influenza and occurrence of outbreaks are limited to winter and autumn. This epidemiological profile of influenza in Iran is similar to that in the published literature. For example, Durand et al. [32] reported that half of human influenza cases occurred from January to March. This finding was reported during 2004–2013.
In addition to addressing the epidemiological profile of ILI and influenza, another key finding of our study was applying the CUSUM algorithm to the potential prediction of the seasonal activity of influenza in Iran. Capabilities of the CUSUM algorithm in timely detection of aberrations have been investigated on health data [24,26,33–35]. Our study focused on the potential application of CUSUM for early detection of any change/aberration on the time series of the ILI cases. Published literature has reported adequacy of the CUSUM performance [36–39]. However, such similar studies did not report the appropriate timeliness of CUSUM. Accordingly, the results of our study are in accordance with those of other studies, especially regarding a considerable similarity between time series graphs of the ILI cases and CUSUM algorithm values.
Further studies should be conducted to investigate the potential performance of CUSUM and other outbreak detection methods on the ILI cases, especially at daily lag instead of weekly lag, as one of the limitations of this study.
5. CONCLUSION
Despite a considerable similarity between the time series of the ILI cases and CUSUM algorithm in weekly lag, the seasonal activity of influenza in Iran could not be predicted by the CUSUM algorithm. Further studies are required to assess the capability of CUSUM using the ILI data at daily lag instead of a weekly report.
AUTHORS’ CONTRIBUTIONS
MK and SH conceived the idea for the manuscript, conducted the statistical analyses, interpreted the data and drafted the manuscript. MK, SH, MF, and YM conducted the data preparation and analysis. All authors read, provided feedback, and approved the final submitted version of the manuscript.
CONFLICTS OF INTEREST
The authors declare that they have no competing interests.
REFERENCES
- [1].Comach G, Teneza-Mora N, Kochel TJ, Espino C, Sierra G, Camacho DE, et al. Sentinel surveillance of influenza-like illness in two hospitals in Maracay, Venezuela: 2006–2010. PLoS One. 2012;7:e44511. doi: 10.1371/journal.pone.0044511. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [2].Costello BE, Simon HK, Massey R, Hirsh DA. Pandemic H1N1 influenza in the pediatric emergency department: a comparison with previous seasonal influenza outbreaks. Ann Emerg Med. 2010;56:643–48. doi: 10.1016/j.annemergmed.2010.03.001. [DOI] [PubMed] [Google Scholar]
- [3].Gasparini R, Bonanni P, Amicizia D, Bella A, Donatelli I, Cristina ML, et al. Influenza epidemiology in Italy two years after the 2009–2010 pandemic: need to improve vaccination coverage. Hum Vaccin Immunother. 2013;9:561–67. doi: 10.4161/hv.23235. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [4].Hickmann KS, Fairchild G, Priedhorsky R, Generous N, Hyman JM, Deshpande A, et al. Forecasting the 2013–2014 influenza season using Wikipedia. PLoS Comput Biol. 2015;11:e1004239. doi: 10.1371/journal.pcbi.1004239. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [5].Lai PL, Panatto D, Ansaldi F, Canepa P, Amicizia D, Patria AG, et al. Burden of the 1999–2008 seasonal influenza epidemics in Italy: comparison with the H1N1v (A/California/07/09) pandemic. Hum Vaccin. 2011;7(Suppl):217–25. doi: 10.4161/hv.7.0.14607. https://www.ncbi.nlm.nih.gov/pubmed/21922688. [DOI] [PubMed] [Google Scholar]
- [6].Paget J, Marquet R, Meijer A, van der Velden K. Influenza activity in Europe during eight seasons (1999–2007): an evaluation of the indicators used to measure activity and an assessment of the timing, length and course of peak activity (spread) across Europe. BMC Infect Dis. 2007;7:141. doi: 10.1186/1471-2334-7-141. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [7].Davidson MW, Haim DA, Radin JM. Using networks to combine “big data” and traditional surveillance to improve influenza predictions. Sci Rep. 2015;5:8154. doi: 10.1038/srep08154. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [8].Ginsberg J, Mohebbi MH, Patel RS, Brammer L, Smolinski MS, Brilliant L. Detecting influenza epidemics using search engine query data. Nature. 2009;457:1012–14. doi: 10.1038/nature07634. [DOI] [PubMed] [Google Scholar]
- [9].Chan TC, King CC, Yen MY, Chiang PH, Huang CS, Hsiao CK. Probabilistic daily ILI syndromic surveillance with a spatio-temporal Bayesian hierarchical model. PLoS One. 2010;5:e11626. doi: 10.1371/journal.pone.0011626. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [10].Chan TC, Teng YC, Hwang JS. Detection of influenza-like illness aberrations by directly monitoring Pearson residuals of fitted negative binomial regression models. BMC Public Health. 2015;15:168. doi: 10.1186/s12889-015-1500-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [11].Yang X, Liu X, Meng L, Liu D, Yu D, Li H, et al. [Study on early warning method for influenza epidemic in Gansu province] Zhonghua Liu Xing Bing Xue Za Zhi. 2016;37:430–33. doi: 10.3760/cma.j.issn.0254-6450.2016.03.029. (in Chinese). [DOI] [PubMed] [Google Scholar]
- [12].Dai XQ, Liu M, Zhang TH, Yang XS, Li SL, Li XG, et al. Clinical predictors for diagnosing pandemic (H1N1) 2009 and seasonal influenza (H3N2) in fever clinics in Beijing, China. Biomed Environ Sci. 2012;25:61–8. doi: 10.3967/0895-3988.2012.01.009. [DOI] [PubMed] [Google Scholar]
- [13].Hall G, Krahn T, Van Dijk A, Evans G, Moore K, Maier A, et al. Emergency department surveillance as a proxy for the prediction of circulating respiratory viral disease in Eastern Ontario. Can J Infect Dis Med Microbiol. 2013;24:150–54. doi: 10.1155/2013/386018. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3852452/ [DOI] [PMC free article] [PubMed] [Google Scholar]
- [14].Nsoesie EO, Leman SC, Marathe MV. A Dirichlet process model for classifying and forecasting epidemic curves. BMC Infect Dis. 2014;14:12. doi: 10.1186/1471-2334-14-12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [15].Pawelek KA, Oeldorf-Hirsch A, Rong L. Modeling the impact of twitter on influenza epidemics. Math Biosci Eng. 2014;11:1337–56. doi: 10.3934/mbe.2014.11.1337. [DOI] [PubMed] [Google Scholar]
- [16].Preis T, Moat HS. Adaptive nowcasting of influenza outbreaks using Google searches. R Soc Open Sci. 2014;1:140095. doi: 10.1098/rsos.140095. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [17].Saat Z, Abdul Rashid TR, Yusof MA, Kassim FM, Thayan R, Kuen LS, et al. Seasonal influenza virus strains circulating in Malaysia from 2005 to 2009. Southeast Asian J Trop Med Public Health. 2010;41:1368–73. https://www.ncbi.nlm.nih.gov/pubmed/21329312. [PubMed] [Google Scholar]
- [18].Socan M, Erculj V, Lajovic J. Early detection of influenza like illness through medication sales. Cent Eur J Public Health. 2012;20:156–62. doi: 10.21101/cejph.a3735. https://www.ncbi.nlm.nih.gov/pubmed/22966744. [DOI] [PubMed] [Google Scholar]
- [19].Truyers C, Lesaffre E, Bartholomeeusen S, Aertgeerts B, Snacken R, Brochier B, et al. Computerized general practice based networks yield comparable performance with sentinel data in monitoring epidemiological time-course of influenza-like illness and acute respiratory illness. BMC Fam Pract. 2010;11:24. doi: 10.1186/1471-2296-11-24. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [20].Zhang HL, Lai SJ, Li ZJ, Lan YJ, Yang WZ. [Application of cumulative sum control chart algorithm in the detection of infectious disease outbreaks] Zhonghua Liu Xing Bing Xue Za Zhi. 2010;31:1406–409. (in Chinese). https://www.ncbi.nlm.nih.gov/pubmed/21223675. [PubMed] [Google Scholar]
- [21].Influenza Laboratory Surveillance Information Available from: http://www.who.int/influenza/gisrs_laboratory/flunet/en/ (accessed December 4, 2018).
- [22].Chen H, Zeng D, Yang P. Infectious disease informatics and biosurveillance. New York: Springer Science; 2010. [Google Scholar]
- [23].Griffin BA, Jain AK, Davies-Cole J, Glymph C, Lum G, Washington SC, Stoto MA. Early detection of influenza outbreaks using the DC Department of Health’s syndromic surveillance system. BMC Public Health. 2009;9:483. doi: 10.1186/1471-2458-9-483. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [24].Karami M, Soori H, Mehrabi Y, Haghdoost A, Gouya M, Esmailnasab N. Evaluating the performance of an outbreak-detection algorithms using semi-synthetic approach: cumulative sum algorithm. Iran J Epidemiol. 2013;9:29–38. https://www.researchgate.net/publication/287700375_Evaluating_the_performance_of_an_outbreak-detection_algorithms_using_semi-synthetic_approach_Cumulative_sum_algorithm. [Google Scholar]
- [25].Wagner MM, Moore AW, Aryel RM. Handbook of biosurveillance. 1st ed. London: Elsevier Academic; 2006. [Google Scholar]
- [26].Fariadras M, Karami M, Esmailnasab N, Moghimbeigi A, Pazhouhi H. Alarm threshold levels of meningitis outbreak in Hamadan province (2010–2012) J Kermanshah Univ Med Sci. 2015;19:278–85. https://www.ingentaconnect.com/content/doaj/22519068/2015/00000019/00000005/art00006. [Google Scholar]
- [27].Lin J, Kang M, Zhong H, Zhang X, Yang F, Ni H, et al. Influenza seasonality and predominant subtypes of influenza virus in Guangdong, China, 2004–2012. J Thorac Dis. 2013;5(Suppl 2):S109–S17. doi: 10.3978/j.issn.2072-1439.2013.08.09. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [28].McDonald SA, van Boven M, Wallinga J. An evidence synthesis approach to estimating the proportion of influenza among influenza-like illness patients. Epidemiology. 2017;28:484–91. doi: 10.1097/EDE.0000000000000646. [DOI] [PubMed] [Google Scholar]
- [29].Haghshenas M, Jafarian E, Babamahmoodi F, Tabrizi A, Nandoost S, Alizadeh-Navaei R. Prevalence of influenza A/H3N2 virus in northern Iran from 2011 to 2013. Caspian J Intern Med. 2015;6:116–19. https://www.ncbi.nlm.nih.gov/pubmed/26221512. [PMC free article] [PubMed] [Google Scholar]
- [30].Tallo VL, Kamigaki T, Tan AG, Pamaran RR, Alday PP, Mercado ES, et al. Estimating influenza outpatients’ and inpatients’ incidences from 2009 to 2011 in a tropical urban setting in the Philippines. Influenza Other Respir Viruses. 2014;8:159–168. doi: 10.1111/irv.12223. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [31].O’Brien K, Barr IG. Annual report of the National Influenza Surveillance Scheme, 2006. Commun Dis Intell Q Rep. 2007;31:167–79. doi: 10.33321/cdi.2007.31.12. https://www.ncbi.nlm.nih.gov/pubmed/17724991. [DOI] [PubMed] [Google Scholar]
- [32].Durand LO, Glew P, Gross D, Kasper M, Trock S, Kim IK, et al. Timing of influenza A(H5N1) in poultry and humans and seasonal influenza activity worldwide, 2004–2013. Emerg Infect Dis. 2015;21:202–8. doi: 10.3201/eid2102.140877. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [33].Karami M. Validity of evaluation approaches for outbreak detection methods in syndromic surveillance systems. Iran J Public Health. 2012;41:102–3. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3521881/ [PMC free article] [PubMed] [Google Scholar]
- [34].Li J, Yang P, Wu SS, Wang XL, Liu S, Wang QY. [Early detection on the onset of scarlet fever epidemics in Beijing, using the cumulative sum] Zhonghua Liu Xing Bing Xue Za Zhi. 2013;34:526–30. (in Chinese). https://www.ncbi.nlm.nih.gov/pubmed/24016449. [PubMed] [Google Scholar]
- [35].Polanco C, Castañón-González JA, Macias AE, Samaniego JL, Buhse T, Villanueva-Martinez S. Detection of severe respiratory disease epidemic outbreaks by CUSUM-based overcrowd-severe-respiratory-disease-index model. Comput Math Methods Med. 2013;2013:213206. doi: 10.1155/2013/213206. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [36].Boyle JR, Sparks RS, Keijzers GB, Crilly JL, Lind JF, Ryan LM. Prediction and surveillance of influenza epidemics. Med J Aust. 2011;194:S28–S33. doi: 10.1111/irv.12442. [DOI] [PubMed] [Google Scholar]
- [37].Hendriks W, Boshuizen H, Dekkers A, Knol M, Donker GA, van der Ende A, et al. Temporal cross-correlation between influenza-like illnesses and invasive pneumococcal disease in The Netherlands. Influenza Other Respir Viruses. 2017;11:130–37. doi: 10.1111/irv.12442. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [38].Johnson HA, Wagner MM, Hogan WR, Chapman W, Olszewski RT, Dowling J, et al. Analysis of Web access logs for surveillance of influenza. Stud Health Technol Inform. 2004;107(Pt 2):1202–06. https://europepmc.org/abstract/med/15361003. [PubMed] [Google Scholar]
- [39].Zheng W, Aitken R, Muscatello DJ, Churches T. Potential for early warning of viral influenza activity in the community by monitoring clinical diagnoses of influenza in hospital emergency departments. BMC Public Health. 2007;7:250. doi: 10.1186/1471-2458-7-250. [DOI] [PMC free article] [PubMed] [Google Scholar]