P trontiers in Physics BRIEF RESEARCH REP OPEN ACCESS Eatted by: Neewegan Universo Slence ans Tectvelogy Norway Reviewed by: ‘Acta Bs, vest of Sete aly Xingu Cre, armouth Cotege, Unted States Correspondence: Mates Pare matiepee8gmalcom Speciaty section: ‘Tn ale wassuttosto Social ys, secon o our Fronts Pn Received 3 rch 2000 ‘evepted: 01 Aon 2020 Published 08 ten 2020 citation: Pere M Corse MeN, Stree Mt ‘29 Sider A 2020) Forecasting ‘COND, Front. Phys 8127 ht 10.3360 7070.00177 Forecasting COVID-19 ‘Matjaz Pore****, Nina Gorigok Miksic“*, Mitja Slavinec' and Andraz Stozer* Facaty of Natura Seerces and Mathomates, Unversity of Mater Marder, Sov, Crna Mase Univers Hosp crea Medes Unversy Tacha, Tawar, "Complety Science Hue Verna, Vienna, Aust, “Marboy Unive) Mea! Center arbor Sevens, ‘Facuty ef Mess, Unvesty cf Marbor Marter, Seveia ‘The World Health Organization declared the coronavirus disease 2019 a pandemic on. March 11th, pointing to the over 178,000 cases in over 110 countries and territories around the world at that time. At the time of writing this manuscript, the number of confimed cases has been surging rapidly past the halt-milion mark, emphasizing the sustained risk of further global spread. Governments around the world are imposing various containment measures while the healthcare system is bracing itself for tsunamis of infected individuals that will seek treatment, tt is therefore important to know what to expect in terms of the growth of the number of cases, and to understand what is needed to arrest the very worrying trends, To that effect, we here show forecasts obtained with a simple iteration method that needs only the dally values of confirmed cases as input. The method takes into account expected recoveries and deaths, and it determines maximally allowed daily growth rates that lead away from exponential increase toward stable and declining numbers, Forecasts show that daily growth rates should be kept at least below 5% it we wish to see plateaus any time soon—unfortunately far trom realty in most countries to date. We provide an executable as well as the source code for a straightforward application of the method on data from other countries. Keywords: COVID-18, pandemic, disease dynamics, exponential grow, vralty 4. INTRODUCTION According to data in real time [1], confirmed coronavirus disease 2019 (COVID-19) cases are growing exponentially in most countries around the world. In Italy and Spain the pandemic is already overburdening the healthcare system (2), and shall the current trends persist, i will not take long before this becomes the grim reality also in many other European countries and the United States Forecasting COVID-19 dissemination thus playsa Key ole [3~7]. In the fist place, to inform governments and healthcare professional what to expect and which measures to impose, and secondly, to motivate the wider public to adhere to the measures that were imposed to decelerate the spreading lest a regrettable scenario will unfold (8,9). ‘Research on epidemic processes hat a long and fruitful history in statistical physics (10, 11] Simple mathematical models that describe the essence of epidemic spreading can be used to fit the data with an overseeable number of parameters, and the oblained values ean then be used to make informed predictions. Tn recent years, che research community has also accumulated overwhelming evidence in favor of complex and heterogeneous connectivity patterns in social networks [12~16] ‘These play a key role in determining the behavior of equilibrium and non-equilibrium systems in general, and the spreading of epidemics and finding optimal containment srategies in particular, Interdisciplinary explorations at the interface of statistical physics, network science, and epidemiology, driven by massive amounts of data recording our health and way of life, have given, 1 ‘pt 20201 veka 8 [aie 127rise to digital epidemiology [17] and to the theory of epidemic processes on complex networks [20], From classical models that assume well-mixed poptlations, to the more recent models that account for behavioral feedback and the structure of ou social networks, we have come a long way in better understanding disease transmission and discase dynamics, We are now able (o use this knowledge to develop ellective prevention strategies {11}, and more broadly, we can use the synergies between these Gilferent fields of research to improve our lives and societies 118,19) Nonetheless, in times of urgency even the simplest model can be too complicated, and the small gaps between dilfereat fields of research can seem like gapping holes. In this paper, wwe therefore present a simple iterative method to forecast the number of COVID-19 cates, under the assumption that governmental data is legitimate and truthful. The goal is not to strive for meticulous accuracy nor to present our method asthe state of the at, but simply to provide first insights and guidelines on elementary principles, We wll be happy four work motivates further research to yield more elaborate and accurate prediction methods 2. METHOD As input, our method requires only the readily available daly values of confirmed cases. We denote these values as xj, where € [0,n) isthe index of days. Assuming we have n values avilable in total, we take the last m values of the x; serie and determine the average growth rate during this time according to get > a ‘We also record the minimal and the maximal growth rate during the last m days as G, and Gy, respectively. The simple iteration sis = (14 Ga) ® already provides a decent forecast beyond = n ~ 1, assuming the original m values are described well by exponential growth. This, however, does not take into account that after k= 14 days the majority of infected will recover, and that after d= 21 days a fraction p ~ 0.04 will die [1, 20-22] (see also ourworldindata.org/coronavirus). By acknowledging these case recovery and fatality rates, we obtain a better forecast Sy = sea pms (IP) ac co) ‘where the asterisk emphasizes that x3, isnot the value that enters back into Equation (2) at the next iteration. If that was the case, the forecasted numbers of cases would drop fast, That might bbe a reasonable assumption if the number of infected would approach the population size, and if recovering from COVID-19 would mean becoming immune to the ditease [25]. The form is not yet the case, while the ater is also questionable given that there are reports of individuals being reinfected and the fact that there are now more diflerent straint of SARS.CoV2 identified and that the viral genome is evolving rapidly [24-26] (see also nextstrain.org/#ncov). Also of note, the values h, d, and p for COVID-19 vary significantly in the existing literature [1, 20~ 22, 27-28], butitis not the scope ofthis paper to determine them accurately, Rathes, we use what seem to be reasonable estimates to illustrate our point. Important ‘do not affect the forecast that significantly. The key factor is the average growth rate Ga, determined as per Equation (2). We have found 7 < m < 14 to yield good results, whereby the lower bound ensures a easonable statistics on Ga, while the ‘upper bound shovld still satisty n — 1 —m > d lest we run out of data (| < 0) in x. in Equation (3). We use m = 14 for the forecasts shown in Figure I. Lastly, if we wish to rely on actual data in Equation (3) beyond i = n — 1, and taking into account 1h < d, we have to impose a forecasting horizon no longer than noth We provide an executable as well as the source code in C for a straightforward application of the above method on any data, The executable searches forthe file data tx in the directory and reads the daly values of confirmed cases, which should be provided one number per line. The executable also asks for the year, month, and day ofthe fist entry in the data tt file, and for the value of m. The frst output file is actual txt, which contains three space separate columns, being the date, the number of cases fon said date (returns what is in datatxt minus those recovered and dead up to then), and the growth rate during the previous day. The second output file is forecast.txt, which also contains three space separate columns, being the date, the forecasted number of eases on said date, and the average daily growth rate used for the prediction, The forecast is made for thirty different average daly growth rates, staring from a 20% increased Gr (as determined whilst calculating Gi, via Equation 1) and decreasing in equal intervals toward growth rate zero, Forecasts obtained ‘with different geovith rates are separated with an empty line sensible variations in h, d, and 3. FORECAST Results of the method are shovin in Figure 1 for the United States, Slovenia, Iran, and Germany for 2 weeks onwards from March 2oth, If the average growth rates during the past 14 days ‘corresponding to ~ 30.6% for the United States, ~ 90% for Slovenia, © 7.5% for Iran, and ~ 18.7% for Germany, persist, ‘we will be looking at ~ 3.9 million cases in the United States, 200 casesin Slovenia, ~ 63,000 casesin Iran, and~ 380, 000 ‘cates in Germany by April 12th, as shown by the solid blue lines in each graph. If the daily growth rates miraculously dropped to zero overnight, we would see what is shown with the solid green lines, That is of course completely unteaistic, but serves to illustrate what would be the best-case scenario. Solid red lines show the forecast oblained if the maximal daily growth rate recorded during the past 11 days, corresponding to ~ 48.9% for the United States, ~ 15.59 for Slovenia, ~ 9.9% for Iran, and ~ 34.2% for Germany, would increase by 20%, This is not the ‘worst-case scenario, but it is arguably bad enough, According to this, Slovenia would have ~ 7,300 cases by April 12th, for example,10 Fortes Stes A Peover a g 3 " g g ; pw t 5 | so Geeta eee PCC Feed sare 35 hae30" age arts “art are iris edd sagt ate Dat " Cue when "PScmany " iF als g° 3 ) A Ee ee a 5] ae ‘ ot Aoi Fe 1B Fab25 May War YO Wari? Mar2e MarSt Aer? Ages Feb 23 Mart Nar® ari War Mar’ for Aor t2 Date Date FIQURE 4 | Fxacats of COD-10 cases for he United Siaes, Sveria. Han, orl Garary-Blok soe dencts th actual da, which wor fis naj st pdted Maren 20 From hs ate aenara we hus nave ne predicted ales, hee e sold Duele denies he contusion of re wend othe past days, nahn wos change. Th uppermost old red ine denes the preicion tan he makmal ly gowth rie recordes during the pas! 4 dass Gy Would reresie by 20%, whe the Larernost cen ne dele ne predten te daly growth ale woul op to er Yom Mh 28h eran, Orange an abe dashed Ines drole pretense ecualy spaced decreasing ly row ates rm op to bolle. Plsteau the ext 1 das nou be raced Unto Sts tage daly growth al = 5.97%; tin tne om Ih bles, Suen arate rout ae = 3.7% 7h fe tom the elle, Fan gl daly gout ale= 3.6% On Irw ‘tom the to), anc Germany target daly rout ato = 5.5% fh ne mie eto Given that the exponential growth sill persists in all four thus far successfully responded to the COVID-19 pandemic examples considered inthis worknote that the vertical scale in have achieved all graphs i logarithmic, and that straight lines thus correspond to exponential growth—the frst goal is to arrest this very 4, OUTLOOK worrying tend. Between the green and the blue line we show forecasts obtained for daily growth rates between zero and the As wehope the presented forecasts clearly shovs, epidemic growth average of the past 14 days with dashed olive lines. By following is ahighly non-linear process, where every day lost to inaction is the lines from bottom upwards, starting with the solid green line, a day too much, ven jasta few days down the road not acing ‘we can identify the one that flattens out by April 12th. For the today can mean the difference between a manageable situation United States, for example it isthe Ath line, which corresponds and a hopelessly overburdened healthcare system. The outlook to the ~ 5.9% daily growth rate from March 29th onwards. This very much depende on whether we take thse facts to heart and ‘would thus be the target if we wished to see a plateau inthe next act accordingly, oF not, Governments ean impose traveling bans, 2 wecks there. For Germany the same target is ~ 55% (Sth line close down shops and restaurants and encourage us to stay al fiom the bottom), for Slovenia itis ~ 3.7% (7th line from the home. Ukimatly, however, tis on each one of us to respect these Dottom), and for Iran it is ~ 3.6% (10th line from the bottom). restrictions and to do all that we can to minimize the chances for ‘These are of course only approximate target values, but further infections. by and large, targeting daily growth rates below at least 5% Keeping the daily growth rates at least below 58 is an seems reasonable and in line with what the countries that have important target for a promising outlook. Data from China, Frontersin Physes[wwastertersh.org 3 penwhere the COVID-19 pandemic seems to be coming to an end, confirm this prognosis, Around mid February the daily growth rates there dropped to around 4% and then to 3% and lower. This marked the beginning of the plateau of confirmed cases, which together with recoveries and deaths led to declining numbers of infected individuals. Singapore, South Korea, and Hong Kong, have also successfully turned their epidemics around by employing strict tactics used in China, Unfortunately, this has not been the ease in many other countries [30] We have two options, The first is to show collective intelligence and restrict our behavior so that new COVID-19. cases will not grow as rapidly as they do now. The second is that we continue to let it slide, until the situation will become so dire that draconian governmental decrees will force us to restrict oar behavior [30]. There is still time to act, but a rosy outlook is moving away from us exponentially fast. REFERENCES 1 Dong Du, Gardner 1 An inercuve webbuved dubboard to tack COVIDI9 in ral time Lancet Infect Du. (2020, 10 1016/5475 509920)90120 +. ped send of ent FRemuns A, Remmi G.COVID.IP and Tay what next? Lancet (2020) oi 10 1016/50140-673620)90627 9. pb aead of print 3.1) Q. Gun X We R Wang XZhow 1, Toog ¥. et al Eady ‘uanassion dynamics la. Wuhan, Chin, of nove coronavuarinfeced peumonin N Eng J Med, 2020) 3821198-207. do: 100S6INEIMow20 4 Thou T, Lin Q, Yang % Lise J, Yang K, Bal W, et al. Podliminay prediction of the base reproduction ‘umber of the Wohan novel Scronairs 2019 nC. Bed Beed ed (030) 133-7. di 101010 5. Zoao $, tin Q. Ran J. Mars 8. Yang G, Wang W. etal Prcliinary ‘etimation ofthe baie reprodcion namiber of novel coronavirus (209 ‘nCaV) im Chins, fom 219 to 2020 data driven analy inthe ety phate ofthe outbreak Int J Infect Dis (2020) 922147. dos. 1010180. 202 ao1aso 6 Lia ¥ Gaye AA, Wilder Smith A, Rodloy | The eepoductve number of ‘COVID-IP is higher commpsed to SARS coronas} Teevel Med. 2020) rasan. ds 101093}/ ant 7. LatA,Bergoa A, AeciaresC Gall Me ZhenderG. Hy phylogenetic etiste ofthe eetve reproduction number af SARS CoV-2] Med Viral (2020, {i 101002) me25723, [pb ahead of print 4 Tppolto , Hs: Ds, Neoumi Macurer Mt, Zama A Toning down the 2019, CoV medi ype ~ and rertorng hope Lancet Respir Med (2020) 230-1 oi 1 to1s/s2218-2600020)300708 9, Mecloskey B Zuma A, Ippolito G, Blumberg L, Arbon F Ciero A ea ‘Mass gathering even abd seducing forer gba spread of COVID. 19 4 pola and pubbe heath lemsaa Lancet (2020) 3950969, do 104016/8014-673620)3068-4 1, PatorSstnrar R, Cavtelane C, Van Mieghem R Verpigntni A pidense proce in complex neborks. Rev Mod Phys, (2015) 47935, os 101103 ReehtodPhye #7925 11, Wang Z% Bauch CT, Bhatacharys S, €Onoio A, ManedsR Pee Me et al Sita phyicr of waeinaton Phyr Rep (2018) 6641-113, oi 101016) phytep 216.0006 12, Bocealti 8, Latort V_ Moteno Y, Chaver M Hwang D. Complex networks: siucure and dynamics Phys Rep. (008) 2475-808 oi: 101016) phytep 205 10.008 18, Holme % sarah | Temporal neworks. Phys Rep. (2012) 519:97-125. do: 101016) phytep 2012.03 001 DATA AVAILABILITY STATEMENT Publicly available datasets were analyzed in this study. The executable, source code and data are available at: http://www. smaljazpere.com/COVID-18. AUTHOR CONTRIBUTIONS MP and AS designed and performed the research, MP, NG, MS, and AS wrote the manuscript. FUNDING ‘This work was supported by the Slovenian Research Agency (Grant Nos. 14-9302, J1-9112, 12-1725, ]3-9289, P1-0403, P3 (0396, N3-0048, N3-0133, and 10-0028), 14 Boceets $, Blancont G, Crado R, del Genio C. Gémer-Gurdeser J Romance M, et al The suture and. dynamics of smllayer networks Phy» Rep. (2014) S44i-122 doi 101026) phyoep 201, 15, Kiel M. Avene A, Banhelemy M, Gleson JR Moreno, Porter MA Malayer networks, Complex New (2014) 203) do 10 2098/omnetensbIs 16 aL Chen D, Ren XL, Zhang QM, Zhang YC, Zhou T. Vit nodes Kentfcalon in compler networks Ps Rep. 2016) 6504-83, ok: 10,1016) pyre 2016 96007 17. Sau M, Benguson L, Bodaar 1], Brewer DD, Brownstein JS, Bude etal. Digital epidemiology Plas Comput Bal (2012) BelO02616 doe 10.371 journal pe 002606 18 Hebing D, Broekmann D, Chadelour 1; Dosnay K,Slanke U, Wooley Mess 0, et al Sting human lives what complexty scence and Information sytems can contrite | Stat Phye (2015) 158735-81 do 10 100710955 01410249 18, Pere M. The socal physics collective, Soi Rep. (018) 96589, dot 10,034/41598019-533004 a0, Huang ¢, Wang Yo Li X Rea L, Zhao J. Ho Y, et al Clnkal features of paens feted wth 2019 novel coronas in. Wabas, Chin Lane (2020) 395497-S05. doe 10L0U680140-673620), 21, Lameini M,Pulolo GCOVIDAS ina: momentous Aecsons and many vncertnie, Lancet Glob Heal, (2020) foe 10 2016/2214 109(2030010 8 (pb ahead af pint. 22, Baud D, QU, Nideen Sunes K, Mano D, Pomir 1, Favre G. Real simate of morality flowing com 19 infection. Lancet Infert Dis (2020) dot 10 016181478 309920)0198 X (Epub ahead of prin) 28, Chen D, Xu W. Le Z, Hoang Z, Lis. Guo Ze al. Recwrence of postive sus-or2 tha tn coma: a ease eeport. Int J Inet Dis (2020, {oi 102016) 2020.03.03. Epub ahead of in. 2A, Zhou F Yang XL, Wang XG, 6, Zhang, Zhang W, «aL. A poeumonia ‘outbred avociated with a new coroners of probable Bat rig, Nate. (2020) 579.2705. dos 101038/3546 020 2012.7 25, Cernlo , Ging FM. Genomic variance of the 2019-nGBV coronas. Med Vio (2020) 92:522-4 do 10 1002} 25700 26, Tang X Wu C.11X, Song ¥ Yo X, Wa, eal On the opin and eotinsing volition of SARS.CoVe2 Nat Sei Rey (2020) nwan0¥6. dot 101098 wus 27, Laver SA, Grane KIL, BL Q, Janse EK, Zheng Q, Meredith HR, ta The incubation pesid of coronasus sete 2019 (cod 19) from publicly reported confemed cases esition and appleaion. Ann Intern Med. (2020), do 107325220-0504 (Epub abead api26 Jung 5 AbSmeuhanov AR, Heyubi X Linton NM, Yang ¥ Yuas Be [Real te eimaton of the kof death fom novel caronsirs (COVID 19) sndeton snference ung exported cee J Chix Med. (2020) 9525, i 103380pems0n0823 29, Thow f Ye T, Dy R, Fan G, Lia ¥, Lv 7%, al Clini course and rik factors for morality of aduk inpatients with COVID19 in ‘Wbin, china retrospective cohort study Lane, (000) 395 1054462 oe 101016 /8040-673620)305563, 30, Cohen Kupfesehmidt K. Countres test tcl in wae” agaist COVID-19, Science. BODO) 3672878. dats 101126icence 37 8 Confit of Intare: The suhos declare tat the rtearch was conducted i the hence of my commerciale ancl seshiondhipe thal cold be contd atx potential conic of interest capri © 2020 Pere, Goiek Mii, vines ad Stter, Thi ian open acces arte ditribted unde the erm af the Creave Commons Atruton Lente (CC BY). The re, itrivton o¢ reproduction in other forumt i permite provided ‘he angina authors) and he copyright owner) ae credited and thatthe original ‘ication in hs journal ted i acodance ith acepted academe rate [No we, debut reproduction i panied which does not comply with these Frentersin Phys | wunctertersb.org ‘Ae 200 | Vokene 8 el 197