key: cord-1020898-dj5jsoiv
authors: Yokogawa, S.; Ishigaki, Y.; Kitamura, H.; Saito, A.; Kawauchi, Y.
title: Prevention of SARS-CoV-2 airborne transmission in a workplace based on CO2 sensor network
date: 2022-03-12
journal: nan
DOI: 10.1101/2022.03.04.22271934
sha: deb514c74dfa0207f036e6d80726cae116104e7f
doc_id: 1020898
cord_uid: dj5jsoiv

We measured the compartmental air change per hour (ACH) using a CO2 sensor network in an office space where a cluster of COVID-19 infections attributed to aerosol transmission occurred. Generalized linear mixed models and dynamic time warping were used for a time series data analysis, and the results indicated that the ventilation conditions were poor at the time of the cluster outbreak, and that the low ACH in the room likely contributed to the outbreak. In addition, the adverse effects of inappropriate partitions and the effectiveness of ventilation improvements were investigated in detail. ACH of less than 2 /h was considered a main contributor for the formation of the COVID-19 cluster in the studied facility.

This work was supported by JSPS KAKENHI Grant No. 21K19820. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted March 12, 2022. ; https://doi.org/10.1101/2022.03.04.22271934 doi: medRxiv preprint Introduction 45 Controlling the spread of COVID-19 has become a priority worldwide. After case 46 reports in December 2019, social distancing has been widely adopted as a 47 containment strategy. The adoption of this social lifestyle has reaffirmed the importance 48 of direct human connections and face-to-face interaction. Therefore, it is essential to 49 control the risks and ensure safety in educational, public, and workplaces, which is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted March 12, 2022. ; https://doi.org/10.1101/2022.03.04.22271934 doi: medRxiv preprint requires efforts of managers, supervisors, administrators, and all other stakeholders. 51 To reduce the risk of COVID-19 transmission, measures against all three routes 52 of infection (contact, droplet, and aerosol) and to reduce the probability of infection 53 through multiple defenses, social distance, mask, vaccine, etc., are required. However, 54 compared to contact and droplet transmission, which can be prevented by social 55 distancing and the use of masks, aerosol transmission is difficult to visually recognize, 56 and the effectiveness of respective countermeasures has not been confirmed. 57 Accordingly, mass transmissions of COVID-19 have been reported in poorly ventilated 58 areas. 1 In addition, the inappropriate use of plastic sheeting for preventing droplet 59 infection has caused clusters of infectious diseases and threatened workplace safety. 2

To avoid such risks, the use of CO2 sensors to control indoor air quality has 61 attracted significant attention. [3] [4] [5] [6] [7] The measurement of indoor CO2 concentration 62 (referring to exhaled air) is considered an effective method for indirect risk 63 management to ensure that exhaled aerosol particles containing SARS-CoV-2 do not 64 remain indoors. Therefore, these devices have been widely installed as a safety 65 measure in places where people gather, such as restaurants, stores, classrooms, and 66 offices. The guideline for its operation considers a provisional control value of 800- is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint measures so as to improve the transmission risk and the safety of workplaces. In the 75 manufacturing industry, quality assurance focuses on processes to ensure quality, and 76 the proper control of a process is an indicator of product quality. 10 Similarly, the risk 77 management of COVID-19 requires the management and control of the environment 78 (i.e., ventilation), rather than of the CO2 concentration itself.

The ventilation in a building room can generally be calculated by dividing the 80 building volume by the ventilation volume of the installed ventilation measures. 81 However, years after the construction of a building, the layout of rooms might change 82 or the performance of the ventilation system can be degraded, so the capacity of the Therefore, a systematic evaluation should be conducted by analyzing the time series is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted March 12, 2022. ; https://doi.org/10.1101/2022.03.04.22271934 doi: medRxiv preprint of data from sensors arranged in a network, and these results can be used to improve 89 ventilation and reduce transmission risks.

In this study, we used a CO2 sensor network and conducted a tracer gas is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The objectives of this research were to 1) demonstrate a method for evaluating 109 and determining the state of air quality management in an office with a complex 110 geometry using a CO2 sensor network, and 2) verify the effectiveness of ventilation 111 improvement measures. The workplace is located on the second floor of a three-story building that was 126 . CC-BY 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted March 12, 2022. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted March 12, 2022. ; https://doi.org/10.1101/2022.03.04.22271934 doi: medRxiv preprint yielded a p-value of 0.02 for the null hypothesis of no difference in infection rates. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted March 12, 2022. ; https://doi.org/10.1101/2022.03.04.22271934 doi: medRxiv preprint numbers 1 to 8 indicate the locations of each sensor. The TR-76Ui sensor can detect 159 CO2 concentrations from 0 to 9,999 ppm, with an accuracy of ±50 ppm (±5%). 160 The experiment was conducted from 10:30 am to 12:30 pm on November 28, 2021. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint the sensors. The stable mass balance of well-mixed air can be described as:

where is the concentration of indoor pollutants at time , is the number of i.e. = 0, Equation (2) can be transformed into:

This equation suggests that a decrease in the normalized concentration of CO2 with is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint in ACH and its dependence on the position relative to ventilation points.

The observed CO2 behavior reflected well the characteristics of the workplace 194 compartmental ventilation. Figure 2 is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted March 12, 2022. ; https://doi.org/10.1101/2022.03.04.22271934 doi: medRxiv preprint 14 intercept. The ACHs estimated for each condition and sensor are shown in Table 2 . 230 The statistical software JMP Pro Ver. 16 was used for the regression analysis. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted March 12, 2022. ; https://doi.org/10.1101/2022.03.04.22271934 doi: medRxiv preprint listed in Table 3 . The estimated covariance parameters of the variational effects are 272 shown in Table 4 . The results shown in Table 3 indicate that the interactions between 273 ventilation time and sensor location were highly significant. Moreover, the decrease in 274 CO2 concentration with ventilation time for Conditions 1 and 2 was also highly 275 significant. The results shown in Table 4 also indicate that the random effects were not 276 significant on their own for both Conditions 1 and 2. These results suggest that is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint Figure 5 shows the fitting results of the estimated GLMM to the observed values. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted March 12, 2022. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted March 12, 2022. ; https://doi.org/10.1101/2022.03.04.22271934 doi: medRxiv preprint 21 measured over time. DTW is an algorithm for measuring the similarity between two 313 time-series data, which may vary in speed. Similarities in CO2 variation can be detected 314 using DTW, even if there are accelerations and decelerations during an observation.

To calculate the DTW distance, we used the statistical language R package "dtw" Ver. The matrix of the DTW distance between each sensor data is shown in Figure 6 . is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted March 12, 2022. ; https://doi.org/10.1101/2022.03.04.22271934 doi: medRxiv preprint with measures against contact and droplet transmission, the maximum height of 341 partitions should be strictly controlled, and they should be installed at a height and 342 orientation that do not interfere with ventilation. Measurement using sensor networks 343 is effective in detecting such a ventilation bias. In addition, the observed bias of CO2 is 344 more complex in rooms with larger sizes, complex geometries, and various uses, 2,11 345 and this study agrees with previously reported results. This index for ACH is consistent with the results of previous studies on 353 tuberculosis and is considered highly valid. 14-16 In large workplaces with complex 354 layouts and partitions, ventilation conditions become complex, so local monitoring and 355 quantitative evaluation using sensor networks, as shown in this study, are effective.

In future studies, real-time data from the CO2 sensor network should be analyzed 357 to identify compartments with increased risk to issue alerts at appropriate times. This is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted March 12, 2022. ; https://doi.org/10.1101/2022.03.04.22271934 doi: medRxiv preprint 24 temperature, humidity, illumination, barometric pressure, and human detection. The 360 authors have already reported a method of applying topological data analysis to 361 multidimensional time-series data from many sensors. 11 In the future, we hope to 362 develop a method for the diagnosing of anomalies by combining data integration with 363 machine learning and deep learning. 

Probable airborne transmission of SARS-CoV-2 in a 375 poorly ventilated restaurant

Experimental 378 investigation to verify if excessive plastic sheeting shielding produce micro clusters 379 of SARS-CoV-2. medRxiv. 2021;21257321

Exhaled CO2 as a COVID-19 infection risk proxy for different 382

It is made available under a perpetuity.is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprintThe copyright holder for this this version posted March 12, 2022. ; https://doi.org/10.1101/2022.03.04.22271934 doi: medRxiv preprint