COVID-19 Road to Recovery

I’ve been tracking the Los Angeles County Department of Public Health COVID-19 Dashboard, and using the data from the Cumulative and Daily Cases and Deaths by Date tables in the CSV download, I’ve updated the 7-day running averages of new cases/day. Here’s what it looks like:



As you can see, there has been what looks like a steady decline in new cases/day for the past 2 weeks. This satisfies the Cases criterion in the Opening Up America Again guidelines which propose gating criteria prior to entering a phased approach to reopening. In reality, most cities had already begun to implement their reopening plans prior to satisfying all the gating criteria. Los Angeles entered Stage II of its Safer L.A. plan on May 8.



Stage II included minor adjustments to the Safer at Home order which allowed a limited number of businesses to reopen if they could provide deliveries or curbside/doorside pickup. Given that Los Angeles entered Stage II on May 8 and new COVID-19 cases have declined since May 15, it appears that the reopening plan has not negatively impacted the trajectory of recovery.

New COVID-19 Cases in Los Angeles - Fun with Numbers

Earlier today, I wrote about the Los Angeles County Department of Public Health COVID-19 Dashboard which shows the number of daily COVID-19 cases, and I provided a figure in my blog to illustrate the trend. However, I felt unsatisfied with the dramatic day-to-day variation in the number of cases, so I decided to calculate rolling averages over the prior 2, 3, 4, 5, 6, and 7 days. Here’s what it looks like:



Based on a calculated 7-day rolling average, we reached a peak of 897 cases on May 11, 2020 and have seen a decline since then. The last day for which we have data are for May 21, 2020 where the rolling 7-day average was 547 cases.

Not surprisingly, I noticed that the dramatic peaks and valleys of the original data follow a 7-day cycle. For each 7-day period, the nadir usually falls on a Sunday (e.g, April 26, May 3, May 10, and May 17), and the peaks occur the following Monday and/or Tuesday which are likely to be the cases that were not reported over the weekend. I can think of a couple of possible reasons for this. First, there may be less staffing on weekends which may result in less case reporting. However, given that the nadir usually occurs on Sundays, I also wonder if patients seek healthcare less frequently on Sundays, perhaps because they are attending church or other activities. Regardless, a 7-day rolling average may be a good way to look at statistical trends because it smooths out the daily variation in case reporting.

I hope that we continue seeing a downward trend of daily cases as we continue reopening businesses.

Reopening Los Angeles

The Los Angeles County Department of Public Health provides a web page for Metrics to Guide Reopening LA County. Criteria include:

• Deaths: 7-day average number of deaths has not increased over the past 14 days
• Hospitalizations and Hospital Capacity: the 3-day average number of hospitalized patients has not increased over the past 14 days
• Supply of Personal Protective Equipment in Hospitals: at least 60% of hospitals have 15+ days of available PPE
• Testing: an average of at least 15,000 tests (equal to 1.5 tests per 1,000 residents) have been done each day over the past 7 days
• Contact Tracing: at least 90% of COVID-19 cases have had follow-up investigation initiated within 1 day of assignment

Currently all criteria are met, with the exception of the PPE criterion where we’ve fallen short on the metric for having 15+ days of available gowns for at least 60% of hospitals (currently at 54%). But generally it looks like we’re in good shape. These criteria differ in some aspects from the White House Opening up America Again guidelines which include 3 gating criteria:

• Symptoms: Downward trajectory of influenza-like illnesses (ILI) reported within a 14-day period AND Downward trajectory of covid-like syndromic cases reported within a 14-day period
• Cases: Downward trajectory of documented cases within a 14-day period OR Downward trajectory of positive tests as a percent of total tests within a 14-day period (flat or increasing volume of tests)
• Hospitals: Treat all patients without crisis care AND Robust testing program in place for at-risk healthcare workers, including emerging antibody testing

I think there are pros and cons for each set of criteria. The White House criteria are overall very sensible from a medical perspective but can be difficult to measure given our gaps in understanding of actual symptoms and prevalence of COVID-19. The Los Angeles Department of Public Health criteria are measurable (again, with some limitations), and it would be nice if they added a criterion for “declining daily cases” similar to the White House criteria. The data for cumulative and daily cases are actually available at the Los Angeles County Department of Public Health COVID-19 Dashboard, and here is the figure provided today:



Data for cumulative and daily cases and deaths are also available for download in the “Counts by Date Table” section. The following caveats are also provided: Recent dates are incomplete due to due to lags in reporting. Cases reported by Episode Date which is the earliest existing value of: Date of Onset, Date of Diagnosis, Date of Death, Date Received, Specimen Collection Date. Deaths reported by Date of Death or Date Received if Date of Death is missing. Number of daily cases will not match the number of newly reported Los Angeles County cases as episode date reflects date of underlying illness rather than date of report.

In any case, by most metrics it is appears that we are on the path to safely re-opening Los Angeles, and I hope that is true for all other cities in the United States.

Google COVID-19 Community Mobility Reports

To help measure the impact of social distancing, Google has provided COVID-19 Community Mobility Reports that chart movement trends over time at the following places:

• Grocery & pharmacy: Mobility trends for places like grocery markets, food warehouses, farmers markets, specialty food shops, drug stores, and pharmacies
• Parks: Mobility trends for places like local parks, national parks, public beaches, marinas, dog parks, plazas, and public gardens
• Transit stations: Mobility trends for places like public transport hubs such as subway, bus, and train stations
• Retail & recreation: Mobility trends for places like restaurants, cafes, shopping centers, theme parks, museums, libraries, and movie theaters
• Residential:Mobility trends for places of residence
• Workplaces: Mobility trends for places of work

Reports in PDF format are available for a variety of countries and regions. For the United States, individual reports are available further subdivided into state- and county-level data.



Changes for each day are compared to a baseline value for that day of the week, where the baseline is defined as the 5-week period from January 3 to February 6, 2020. The data come from users who have opted in to Location History for their Google Account. Google recognizes that anonymity may be compromised in sparsely populated areas. In such instances, they state, “Not enough data for this date: Currently, there is not enough data to provide a complete analysis of this place. Google needs a significant volume of data to generate an aggregated and anonymous view of trends.”

In addition to providing reports in PDF format, Google also allows visitors to download data in CSV format. Data geeks rejoice! More information is available in this blog. What does community mobility look like in your area?

Goodbye Traditional Router, Hello Mesh Router!

I live in a condo complex where all units have 3 floors plus an underground parking area. Since July 2014, I’ve used a traditional modem-router (NETGEAR C3000-100NAS) that broadcasts wifi to our entire home from a single access point on the 2nd floor. Wireless access has been very reliable on all 3 floors. However, the wireless signal has been weak to nonexistent from the underground garage, and that would sometimes make it difficult for me to look up navigation options while sitting in my car before I begin my drive. My workaround has been to look up driving routes while still on the 1st floor or to turn off wifi while in the garage to force my phone to switch to cellular data. No biggie.

In the last 2 months, we have increased our internet usage due to working and schooling from home as well as adopting video conferencing as a more common method of communication. I’ve noticed that our Apple TV and Xbox on the first floor have had intermittent connectivity problems which I initially chalked up to our internet service provider (Spectrum, formerly Time Warner Cable) facing a surge in demand due to shelter at home orders. However, I began to notice dead spots in certain areas of the 1st floor which made me wonder if my router was starting to malfunction.

It turns out that routers can slow down with age, so I decided to test my hypothesis by upgrading to a mesh router. While traditional routers broadcast wifi from a single access point, mesh routers can consist of several access points that communicate with one another to more efficiently route data to and from multiple devices. I bought a TP-Link Deco M9 Plus mesh router system. Although it comes with 3 units, I felt that I only needed 2 units and will use the 3rd unit as a backup. I attached the 1st unit via ethernet cable to my NETGEAR modem-router (which I’m essentially using only as a modem now) on the 2nd floor, and I installed the 2nd unit near the kitchen on the 1st floor.

Setup was a breeze using the TP-Link iOS app. I helped a friend install a mesh router and had all kinds of problems, but the TP-Link instructions were very clear, and configuration options were intuitive to understand and/or pre-populated with default options that work for most situations. Because the mesh router must be assigned its own network name and password, and because I didn’t want to re-connect all of our devices (we have more than 10 internet-enabled devices) to a new network configuration, I first renamed the network on my NETGEAR router and then assigned the original network name and password to my new TP-Link mesh router. All our our devices then automatically connected to the mesh network. Boom. I love it when things work. My dead spots are gone, and I can even get a moderate wifi signal (usually 2 out of 3 bars) from the underground garage.

Interestingly, I noticed that internet speeds seemed to be faster with the mesh network, so I decided to perform a head-to-head comparison between my traditional NETGEAR router and my new TP-Link mesh router. Using the iMac in my 2nd floor office which is located about 15 feet away from both routers (more or less equidistant), I fired up Safari 13.1 on macOS 10.14.6 and ran 6 sequential speed tests on Speedtest, alternating between my traditional and mesh router networks. I then calculated averages for ping time, download speeds, and upload speeds. Here are the results:

Seq	Router	Ping (ms)	Down (Mbps)	Up (Mbps)	Speedtest Result
1	Traditional	25	8.34	2.12	https://www.speedtest.net/result/9429818874
2	Mesh	17	52.82	1.65	https://www.speedtest.net/result/9429824134
3	Traditional	24	3.63	3.62	https://www.speedtest.net/result/9429830017
4	Mesh	19	41.99	1.97	https://www.speedtest.net/result/9429833684
5	Traditional	17	13.60	1.76	https://www.speedtest.net/result/9429837451
6	Mesh	20	51.53	1.36	https://www.speedtest.net/result/9429841785
Avg	Traditional	22.00	8.52	2.50	N/A (calculated average)
Avg	Mesh	18.67	48.78	1.66	N/A (calculated average)

Although the sample size is small, the mesh router appears to have download speeds that are faster by more than a factor of 5. Another way to look at the results is that I used to get 50-60 Mbps download speeds with my traditional modem-router, but it’s no longer running at peak performance. Curiously, the upload speeds appear to be slightly slower with the mesh network as compared with the traditional router, although it may be premature to conclude whether or not there is a statistically significant difference. Maybe I’ll run some more speed tests later to find out.

In summary, it appears that there has been degradation of performance in my traditional modem-router which adds a data point to the assertion that routers slow down with age. Additionally, if you have dead spots anywhere in your home, there is a good chance that a mesh network will solve your wifi woes. I’m certainly happy with mine.

Is Your State Doing Enough COVID-19 Testing?

Most of the United States has some kind of stay at home order in effect as a social distancing response to the COVID-19 pandemic, and many states have begun to gradually re-open businesses or are planning to do so in the near future. Some key factors that should drive the decision to nudge our lives closer to normal include getting the infection under control (seeing a declining number of new cases), having a contact tracing solution in place, and having adequate testing capacity.

It is well known that the US has had inadequate means to test for SARS-CoV-2 since the initial outbreak, and the problem was not adequately addressed as the infections grew to pandemic proportions. A recent analysis shows that more than half of US states aren’t doing enough COVID-19 testing, based on data from the COVID Tracking Project.

Based on the same data, the folks from NPR have created a graphic to illustrate how each state is performing in terms of current daily testing versus its minimum target.

The visualization displays the size of the outbreak for each state, as represented by deaths per 100,000 people, with a current peak value of 132 deaths per 100,000 people in New York. It also displays current daily testing versus a minimum target threshold that is needed by May 15, 2020, and this is measured as tests per 100,000 individuals. Most states are well below their May 15 target thresholds. Finally, the visualization shows the positive test ratio which is the percent of tests that come back with a positive result, and whose desired target is 10% or less.

How is your state doing?

Shelter in Place Index

On Wednesday, March 11, 2020, the World Health Organization declared that COVID-19 had reached pandemic proportions. Shortly thereafter, different parts of the United States began rolling out social distancing and other recommendations for nonpharmaceutical interventions. How did our nation respond to shelter in place orders? The folks at SafeGraph provide a dashboard view of their Shelter in Place Index (a.k.a. Stay at Home Index):



The dashboard visualizes the change in the percentage of people staying home as compared with baseline measures. The “stay at home” determinations are based on data from more than 45 million smartphones which SafeGraph asserts is a representative sample of Americans.

The index represents whether someone stays at home or not. It doesn’t factor in the distance traveled from home because “one does not need to travel long distances to undermine social-distancing and enable viral transmission” according to SafeGraph. “Home” is defined as “ the most common nighttime location in recent months identified to a precision of about 100 square meters.” SafeGraph provides transparency about their methodology and data schema for those of you who want to take a closer look.

While Americans appear to have increasingly stayed at home from mid-March to a peak in mid-April, it appears that we have relaxed our adherence to shelter in place recommendations in the past couple weeks as discussed in this article. For a graphic comparison, see this rolling 3-day average of new confirmed COVID-19 cases in the United States:



We appear to be holding steady with the number of new cases per day. The obvious conclusions are that (1) at least the number of daily cases is not increasing and (2) it would be better if we started to see a decline. We are still in the midst of a global pandemic, and we still have a lot of work to do before we get things under control.