14 | New Scientist | 22 June 2019
News
Protective suits can
stop the spread of
Ebola via body fluids
AT LEAST half of all Ebola
outbreaks may go unrecognised,
and more surveillance is needed
to identify them. Those are the
conclusions of a new analysis
of data from the largest Ebola
outbreak in history.
The study has been published
amid the second-largest Ebola
outbreak in history, which is
currently affecting the Democratic
Republic of the Congo. Since that
outbreak began in August last
year, the virus has infected
about 2000 people, and caused
1400 deaths. Last week, a case
was confirmed across the border,
in Uganda.
Ebola starts with similar
symptoms to the flu – fever
and chills, muscle pain and
headache – but it often ends in
internal and external bleeding,
and death. The largest outbreak
killed more than 11,000 people in
West Africa between 2013 and 2016.
However, such epidemics
are relatively unlikely because
Ebola doesn’t necessarily spread
between many people each time
it “spills over” into the human
population from bats or other
animals. Emma Glennon at
the University of Cambridge
and her colleagues suspect
that, when it enters the
human population, Ebola often
dies out before it can grow
into a medical emergency.
To test the idea, Glennon’s
team used computer modelling
informed by data sets related to
the West Africa epidemic. After
taking into account the factors
that determine the likelihood
of the virus passing from person
to person, they estimated that
more than half of spillover
events go undetected, implying
that well over 100 outbreaks
of Ebola have never been picked
up by medical monitoring.
One of their models suggested
that the undetected spillover
figure might be as high as
83 per cent (PLoS Neglected
Tropical Diseases, doi.org/c7bq).
Single cases are likely to be the
most common outbreak size, but
only two countries have reported
these so far, they say. Ian Mackay
at the University of Queensland
in Australia says identifying
outbreaks early when only a few
people have been infected is vital,
because it is easier and cheaper
to contain them before dozens
of people have been infected.
He says the study also sheds
light on curious trends that
virologists have noticed in
the decades since Ebola was
discovered. “We have seen
studies come out over the years
from the 1980s onward showing
indications that some people have
antibodies [to the virus], but we
haven’t seen that to be a part of
ongoing outbreaks,” says Mackay.
This could be explained
by small outbreaks of Ebola
going unreported.
“If we’re seeing half or even
up to 80 per cent [of spillover
events] are missed, that might
contribute to these strange
antibody results,” Mackay says. ❚
Ebola going undetected
More than half of all outbreaks of the deadly virus are missed
Infectious diseases
Ruby Prosser Scully
RE
UT
ER
S/G
OR
AN
TO
MA
SE
VIC
MEET Irony Man. This
30-centimetre-high robot
combines nonchalant facial
expressions with deadpan
delivery in a way its inventors hope
will make it seem more natural. By
using irony in its dialogue, they say
it could be used to break bad news
gently and persuade people to do
things they otherwise wouldn’t do.
“Think of a robot in the role of a
lifestyle adviser that feels the user
should be more active and has
to convey this message without
appearing rude,” says Elisabeth
André at the University of Augsburg,
Germany, who worked on Irony Man.
The technology is based on a
computer program that analyses
the best response the robot could
make in a normal conversation. It
then flips words and adds emphasis
like overstatement to introduce
irony, that is to say, when the
intended meaning of a phrase
is opposite to the literal meaning.
So, in a conversation about the
weather, the phrase “I hate raining.
I usually have a bad mood when it
rains” becomes “Super! I utterly
love raining”, with elongated stress
on the word “utterly” and a smile.
In tests, a dozen students said
they preferred conversations with
Irony Man to those with a standard
robot. They rated it consistently
higher on qualities described as
likeable, appealing and motivating.
The work was presented at
the International Conference on
Autonomous Agents and Multiagent
Systems in Montreal.
The next step is to find a way
to make Irony Man bite its tongue.
“The robot is not yet able to
determine whether and when
there is a good moment to employ
irony,” André says. “It may happen
that the robot generates a funny
utterance, but the user is irritated.”
Imagine the possibilities: Great!
You put a totally expected item
in the bagging area. Eye roll. ❚
David Adam
A robot is using irony
and now people
like it more
Machine learning
Irony Man
doesn’t know
when ironic
quips are
inappropriate
HC
M-
LA
B,^ A
UG
SB
UR
G^ U
NIV
ERS
ITY
1400
people have died so far in the
current Ebola outbreak