Skip to content

Commit 043fb03

Browse files
authored
Merge pull request #278 from cmu-delphi/change-signal-name
Change signal name
2 parents 8295ef9 + 4a1b347 commit 043fb03

File tree

1 file changed

+12
-13
lines changed

1 file changed

+12
-13
lines changed

docs/api/covidcast-signals/changehc.md

Lines changed: 12 additions & 13 deletions
Original file line numberDiff line numberDiff line change
@@ -21,8 +21,8 @@ percentage of COVID-related doctor's visits in a given location, on a given day.
2121

2222
| Signal | Description |
2323
| --- | --- |
24-
| `smoothed_cli` | Estimated percentage of outpatient doctor visits with confirmed COVID-19, based on data from Change Healthcare, smoothed in time using a Gaussian linear smoother |
25-
| `smoothed_adj_cli` | Same, but with systematic day-of-week effects removed; see [details below](#day-of-week-adjustment) |
24+
| `smoothed_covid` | Estimated percentage of outpatient doctor visits with confirmed COVID-19, based on data from Change Healthcare, smoothed in time using a Gaussian linear smoother |
25+
| `smoothed_adj_covid` | Same, but with systematic day-of-week effects removed; see [details below](#day-of-week-adjustment) |
2626

2727
## Table of contents
2828
{: .no_toc .text-delta}
@@ -57,41 +57,40 @@ Standard errors are not available for this data source.
5757
Due to changes in medical-seeking behavior on holidays, this data source has
5858
upward spikes in the fraction of doctor's visits that are COVID-related around
5959
major holidays (e.g. Memorial Day, July 4, Labor Day, etc.). These spikes are
60-
not necessarily indicative of a true increase of COVID-like illness in a
61-
location.
60+
not necessarily indicative of a true increase of COVID-19 in a location.
6261

6362
## Qualifying Conditions
6463

6564
We receive data on the following two categories of counts:
6665

6766
- Denominator: Daily count of all unique outpatient visits.
6867
- Covid: Daily count of all unique visits with primary ICD-10 code in any of:
69-
{U071, U072, B9721, or B9729}.
68+
{U07.1, B97.21, or B97.29}.
7069

7170
## Estimation
7271

7372
### COVID-Like Illness
7473

7574
For a fixed location $i$ and time $t$, let $Y_{it}$
7675
denote the Covid counts and let $N_{it}$ be the
77-
total count of visits (the *Denominator*). Our estimate of the CLI percentage is
78-
given by
76+
total count of visits (the *Denominator*). Our estimate of the COVID-19
77+
percentage is given by
7978

8079
$$
8180
\hat p_{it} = 100 \cdot \frac{Y_{it}}{N_{it}}
8281
$$
8382

8483
### Day-of-Week Adjustment
8584

86-
The fraction of visits due to CLI is dependent on the day of the week. On
85+
The fraction of visits due to COVID-19 is dependent on the day of the week. On
8786
weekends, doctors see a higher percentage of acute conditions, so the percentage
88-
of CLI is higher. Each day of the week has a different behavior, and if we do
87+
of COVID-19 is higher. Each day of the week has a different behavior, and if we do
8988
not adjust for this effect, we will not be able to meaningfully compare the
9089
doctor visits signal across different days of the week. We use a Poisson
9190
regression model to produce a signal adjusted for this effect.
9291

9392
We assume that this weekday effect is multiplicative. For example, if the
94-
underlying rate of CLI on each Monday was the same as the previous Sunday, then
93+
underlying rate of COVID-19 on each Monday was the same as the previous Sunday, then
9594
the ratio between the doctor visit signals on Sunday and Monday would be a
9695
constant. Formally, we assume that
9796

@@ -102,10 +101,10 @@ $$
102101
\end{aligned}
103102
$$
104103

105-
where $Y_{it}$ is the observed doctor visits percentage of CLI at time $t$,
104+
where $Y_{it}$ is the observed doctor visits percentage of COVID-19 at time $t$,
106105
$\text{wd}(t) \in \{0, \dots, 6\}$ is the day-of-week of time $t$,
107106
$\alpha_{\text{wd}(t)}$ is the corresponding weekday correction, and
108-
$\phi_t$ is the corrected doctor visits percentage of CLI at time $t$.
107+
$\phi_t$ is the corrected doctor visits percentage of COVID-19 at time $t$.
109108

110109
For simplicity, we assume that the weekday parameters do not change over time or
111110
location. To fit the $\alpha$ parameters, we minimize the following convex
@@ -126,7 +125,7 @@ adjusted count
126125

127126
$$\dot{Y}_{it} = Y_{it} / \alpha_{wd(t)}.$$
128127

129-
We then use these adjusted counts to estimate the CLI percentage as described
128+
We then use these adjusted counts to estimate the COVID-19 percentage as described
130129
above.
131130

132131
### Backwards Padding

0 commit comments

Comments
 (0)