The Economist - USA (2021-11-13)

80 Science & technology TheEconomistNovember13th 2021

fore  they  begin  to  show  obvious  symp­

toms.  George  Stothart  at  the  University  of

Bath in Britain leads a team that has devel­

oped  very  quick  passive  tests  that  hold

great  promise  for  detecting  cognitive  im­

pairment  early.  These  tests  use  electroen­

cephalographic  caps  which  are  worn  on

the head to compare brainwave responses

to a series of images. The caps are relatively

cheap and the test can be conducted using

a tablet computer.  

Taking the exam

Meanwhile,  machine­learning  and  aiare

enabling  big  improvements  in  cognitive

testing,  which,  like  other  sorts  of  pencil­

and­paper  exams,  has  been  prone  to  cul­

tural and educational bias, and to a “learn­

ing bias” (lessened accuracy as practice im­

proves the participants’ results). Cognetiv­

ity  Neurosciences,  for  example,  a  firm

launched by two academics at the Univer­

sity of Cambridge, produces an “integrated

cognitive  assessment”  that  has  already

been deployed by some regions of Britain’s

National  Health  Service.  It  has  also  been

approved by America’s Food and Drug Ad­

ministration (fda) and by regulators in the

European Union. 

The Cognetivity test relies on a series of

flashing  images,  among  which  some  ani­

mals  have  been  embedded  for  the  person

being  tested  to  identify—rather  like  the

captchatests  used  by  some  websites  to

weed out robots trying to log on, only these

are conducted at rapid speeds. So rapid, in­

deed,  that  the  tests  cannot  be  conducted

remotely, but are done on tablets in memo­

ry clinics or at doctors’ surgeries. Already,

however, a variety of do­it­yourself cogni­

tive tests of various standards are available

online,  and  some  of  these  could  become

important diagnostic tools.

Even before the explosion in the use of

ai,  scientists  could  detect  evidence  of  de­

mentia from how people use words. A stu­

dy  in  2011,  for  example,  found  clear  retro­

spective  evidence  in  the  writings  in  her

40s and 50s of Iris Murdoch, a novelist, of

the Alzheimer’s she was to die with in 1999,

aged  79.  The  research  arm  of  ibm,  a  com­

puting  giant,  has  used  data  from  the  Fra­

mingham  heart  study,  which  has  tracked

three  generations  of  people  in  a  town  in

Massachusetts  since  1948  to  improve

knowledge  of  cardiovascular  health.  Ajay

Royyuru, who heads ibm’s health­care and

life­sciences  research,  says  that  studying

the  use  of  language  by  participants  in  the

study suggests that changes over time can

be  used  to  predict  which  of  them  will  ac­

quire dementia, seven and a half years be­

fore they are diagnosed even with mci.

Such  data—and  the  massive  amounts

people  compile  every  day  on  their  smart­

phones,  using  various  services  such  as

messaging  or  navigation—could  help  en­

able  much  earlier  detection  of  dementia.

Thismightbepossiblewithapps,although

theethicsofanynon­consensualdiagno­

sisandthewillingnessofpeopleto use

suchservicesareanothermatter.Oneway

oranother,though,PaolaBarbarino,chief

executiveofAlzheimer’sDiseaseInterna­

tional,anadvocacygroup,expectsa “tsu­

nami”ofdemandfortreatment.

For health­care systems around the

world allthis maypose moreproblems

thanitsolves.Themainoneisthelackof

proven treatments. In Junethefda ap­

provedAduhelm,thefirstdrugtotreatAlz­

heimer’s.A monoclonalantibodyshownto

reduce accumulations of beta­amyloid,

ithassofarbeenlittleused,becauseitis

expensiveandinsurersarereluctanttoap­

prove reimbursement when there are

doubtsastowhetheritactuallyslowscog­

nitivedegeneration.Butitisthefirstina

queueofdrugs,forwhichAlzheimer’sspe­

cialistsholdoutgreathopes.

The seconddifficulty isin assessing

whenmcirequiresmedicalintervention.

AsDrGauthierpointsout,somementalde­

clineisindeedpartofageing.Peoplefind

waysofcoping.Tellingthedifferenceis

hardtodothroughbloodtestsortheuseof

AI. Itstillrequirestimeandhumaninter­

vention.Thisispartofthethirdandbig­

gestdifficulty:that,astheworldages,the

numberofpeoplewithdementiaisgoing

toriserapidly,tomorethan80mby 2030

andmorethan140mby2050.Eventoday

healthservicesarebuckling.Whoknows

howtheywillcopeinthedecadesahead.n

COP26

Are theclimategoalsdeadoralive?

“A

t theendofthecopweneedtobein

a positiontosay‘Wearestillontrack

tobewellbelow2 degrees,westillhavea

shotatthe1.5degrees.’Alloureffortsthis

week should be directed at that.” Thus

spokeFransTimmermans,aneucommis­

sioner,as thecop 26 climatesummit in

Glasgowrolledintoitsfinalweek.Unfor­

tunatelyforMrTimmermans,politicalre­

solveandclimateplansdonotcurrently

adduptothedemandinggoalssetoutby

the Paris agreement of limiting global

warmingto“wellbelow2°C”aboveprein­

dustrialaverages,letalonetokeepingitas

lowas1.5°C. Instead,thelatestnumber­

crunchingsuggeststhattheplansofthe 193

partiestotheagreementcollectivelycarve

a pathtoapproximately2.4°Cofwarming

bytheendofthecentury.

Thatisonlya modestimprovementon

wherethingsstoodwhentheParisagree­

mentwasbeingnegotiatedin2015.Under

thedealthatwasstruckatthetime,gov­

ernments offered up pledges to reduce

theirnationalemissions.Toilingawayin

thebackground, climate modellersesti­

matedthatthecumulativeconsequenceof

these pledges would be to bring about

roughly2.7°Cofwarmingby2100.Awareof

thedisconnectbetweenthisfigureandthe

agreement’soverallgoals,negotiatorssaid

they would present new, hopefully im­

proved decarbonisationplans everyfive

years,inthehopesthatthis“ratchetmech­

anism”wouldbringtheoverall1.5­2°CPar­

isgoalscloser.

Hencetheflurryofclimatepledgesthat

weremadeoverthepastyear.Theyfocus

onwhatwillbedonebytheendofthede­

cade,bywhichtimeglobalgreenhousegas

emissionsmustberoughlyhalfwhatthey

werein 2010 inordertohavea goodchance

oflimiting warming to1.5°C. So far, no

countryisontracktodothis,saysNiklas

HöhneofNewClimateInstitute,athink­

tank.DrHöhneispartofa consortiumof

researcherscalledClimateActionTracker,

whichplugsnationalclimatepoliciesand

pledgesintomodelsinordertogiveanidea

ofhowtheytranslateintotemperatures.

Thegroup’slatestresults(seechart),pub­

lishedonNovember9th,saythatif all 2030

decarbonisationplansweretobecarried

out as advertisedbut no furtherefforts

weremade,therewouldbea 68%chance

thatglobalaveragetemperaturesin 2100

wouldbebetween1.9°Cand3.0°Cwarmer

G LASGOW

TheloftygoalsmadeinParisseemlikebeingleftbehind

Uncertain heat

Source:ClimateActionTracker

Global emissions, gigatonnes of CO2 equivalent

*Projectedwarming by 2100

†Median value

-20

0

20

40

60

2000 20 40 60 80 2100

Historical

FORECAST

Tostaybelow 1.5°C

Current policies

2.5-2.9°C*

200 targets

2.4°C*†

Optimistic scenario,

including net-zero

targets 1.°C*†