adding claims_hosp (hospital-admissions) flu signal #2159
Conversation
…id conventions and formatting other peoples code)
There was a problem hiding this comment.
claims local run just finished.
{"elapsed_time_in_seconds": 2083.01, "csv_export_count": 50112, "max_lag_in_days": 4, "oldest_final_export_date": "2025-10-19", "event": "Completed indicator run", "logger": "delphi_claims_hosp.run", "level": "info", "pid": 12584, "timestamp": "2025-10-23T21:41:29.131204Z"}
and looks like there are equal number of flu signal as covid which is 25056
and given the date difference of 2088 days inclusive gives us
25056 / 2088 = 12 (6 geo resolution * 2 time transformation )
I took a quick look and it seems that at least there are same amount of data and looking at one of the latest data the generate numbers seems to correlate with https://www.cdc.gov/respiratory-viruses/data/activity-levels.html
for 20250920 the nation flu was about 0.01 and covid was 0.3 which seems to be proportional to the data table.
since this is a high priority I approved it knowing that I only did superficial checks. Are we planning on going over the statistical analysis for the flu signal?
| max_dates.append(updater.output_dates[-1]) | ||
| n_csv_export.append(len(updater.output_dates)) | ||
|
|
||
| for numerator_name in ["Covid_like", "Flu1"]: |
There was a problem hiding this comment.
Non-blocking: do you know when the Flu1 column became available / do you know if existence of this column is flaky?
adds a set of flu signals to the output, taken from a previously unused column in the input files.
also includes an annoying amount of formatting changes to unrelated or barely related code, because the linter is being applied to these ancient files in drips and drops 😡 (look at the first commit for a more concise view of the functional changes)
a quick review would be appreciated so i can get this deployed to the staging server as soon as possible.