Perhaps something along these lines:
select
s.id, sum_virality, ( ( (sum_likes) + ( sum_comments) ) * ( 1 / ( 1 + s.created_at / 3600000 ) ) )as response
from spreads s
inner join (select t.spread_id, DATE_FORMAT(t.created_at,'%Y-%m-%d %H') created_date, sum(t.virality) sum_virality, sum(t.likes) sum_likes, sum(t.comments) sum_comments
from spreadstimeline t
where t.spread_id = s.id
group by created_date) as history
group by
s.id