ID modeling periods, rates, and fractions?

[chendaniely]

Go through the papers to get the range of incubation periods used in the papers. This will help create a distribution of values that can be used to parameterize epi models. This was from today's call with Lauren Childs.

Get a dataset of values along with the paper metadata along with the geographic location the incubation period is for.

[chendaniely]

#51 finds the statements in the papers that give the incubation period for covid19.

What to do next: if that statement references a paper we need to see if the dataset also has that paper. We want to create a list of all the primary sources where the incubation period is given. This also helps us see which sources are used.

Worst case scenario all the papers are referencing the same paper.

[chendaniely]

From Lauren:

Text Data Mining Words
Key words and rough definitions:

• ”incubation period”
period of time between infection and appearance of symptoms
• ”infectiousness period”
period of time when individuals are able to transmit
• ”recovery rate”
how quickly individuals recover, may depend on severity of infection
• ”case fatality ratio”
fraction of reported cases that end in death
• ”asymptomatic fraction”
fraction of infections that do not develop symptoms
• ”hospitalized fraction”
fraction of cases that lead to hospitalization
It would be great to know what values or ranges are being reported for these
and what population is the source (location, age, etc).

[chendaniely]

papers that mention the current covid pandemic looking for sentences that mention "incubation period" along with "day" (to filter out staments about web lab incubation periods)

• point: the point estimate given in the sentence (e.g., median, mean, value they used)
• lower/upper: a range provided by the paper (e.g., confidence interval, credible interval, range). some papers only provided max time, so that was put in as an 'upper' value

[chendaniely]

after all the filtering for covid19 papers and sentences. not all the keywords existed in the papers provided

	keyword	sentence_count
sent-incubation_period_day	85.0
sent-recovery_rate	12.0
sent-case_fatality_ratio	3.0
sent-infectiousness_period	NaN
sent-asymptomatic_fraction	NaN
sent-hospitalized_fraction	NaN

[chendaniely]

• location for values as case fatality ratio

[chendaniely]

Paper counts after fixing some regex covid19 mentioned papers

Originally

filter_words_lower_regex = [
    r"covid\s?-?\s?19", # covid-19
    r"sars-?cov-?2",    # sars-cov-2
    r"ncov",            # ncov
    r"coronavirus",     # coronavirus
]

changed to account for matches in the middle of words

filter_words_lower_regex = [
    r".*covid\s?-?\s?19.*", # covid-19
    r".*sars-?cov-?2.*",    # sars-cov-2
    r".*ncov.*",            # ncov
    r".*coronavirus.*",     # coronavirus
]

keyword	sentence_count
sent-incubation_period_day	376.0
sent-recovery_rate	65.0
sent-case_fatality_ratio	37.0
sent-hospitalized_fraction	1.0
sent-infectiousness_period	NaN
sent-asymptomatic_fraction	NaN

[chendaniely]

"asymptomatic" and "fraction"
"asymptomatic" and "proportion"
"asymptomatic" and "ratio"
"infectious" and "period" possibly with "day"
"case fatality rate" (vs "case fatality ratio")
"hospitalized" and "fraction" and "cases"
"hospitalized" and "proportion" and "cases"
"latent period" possibly with "day"