sql - pg_trgm を高速化するには?

Question

テーブルを作成:

CREATE TABLE public.personal
    (
      id bigserial NOT NULL PRIMARY KEY,
      first_name character varying(255) NOT NULL,
      last_name character varying(255),
      middle_name character varying(255),
      deleted integer NOT NULL DEFAULT 0
    );

クエリ:

select last_name, first_name, middle_name, similarity(concat_ws(' ', last_name, first_name, middle_name), 'Smit') AS sml
from personal ORDER BY sml DESC LIMIT 100

このクエリを高速化するにはどうすればよいですか?