执行计划右上角有个显示方式选择文本显示

可以试试
/+FILTER_PUSH_DOWN(1) VIEW_PULLUP_FLAG(1)/
比如 select /*+ FILTER_PUSH_DOWN(1) VIEW_PULLUP_FLAG(1) */ * from view_follow where id ='条件'

能否将开窗函数从视图里去掉？
这个开窗序号目的是为了查询结果排序，那放到查询里是否可以？
比如：

--原视图修改
create or replace view view_follow as
(
SELECT t.*
      --,ROW_NUMBER() OVER ( PARTITION BY idcard, disease, TO_CHAR(TIME, 'YYYYMMDD') ORDER BY center ASC) AS rank
  FROM (SELECT t.*
。。。。。。

--查询脚本
SELECT *
      ,ROW_NUMBER() OVER ( PARTITION BY idcard, disease, TO_CHAR(TIME, 'YYYYMMDD') ORDER BY center ASC) AS rank
  FROM view_follow 
 WHERE ID = 1

是期望这样的执行计划嘛？

select t.*, rownumber() ...
from
select t.*, concat(...)
from (
...A... -- 走索引 扫描 [id,id]
union all
...B... -- 走索引 扫描 [id,id]
union all
...C... -- 走索引 扫描 [id,id]
) t
) t

这就错了，和原语句查询的结果不一样。根本原因在视图存在窗口函数有分组计算，如果走索引提前把数据筛选了再做开窗函数，计算的结果就错了。举个例子，查询一个人的工资信息（包含自己的工资，所在部门的平均薪资，所在在部门的薪资排行or公司的平均薪资，在整个公司的平均薪资排行），这得访问薪资表所有记录才能得出正确结果。
因此，这种需求下优化器选择全扫描是正确的，如果强行指定id列索引也能查询出正确结果，只是需要通过id回表绕了个弯。当然，每次全表扫描也是不合适的，尤其是查询频率高的SQL。
还是有加速的解决办法：
1、新增3个表的覆盖索引--用于快速计算rank

create index idx_ai_follow_1129         on ai_follow(id,idcard,disease,TIME);
create index idx_center_follow_hbp_1129 on center_follow_hbp(id,idcard,TIME);
create index idx_center_follow_dm_1129  on center_follow_dm(id,idcard,TIME);

2、新建视图--前半部分t1计算出每个id的rank，后半部分用于关联表找到id对应明细，建议ID列设计为聚集索引（无大字段情况时）

create view view_follow as
SELECT t2.*, t1.rank
from (
	SELECT t1.id,
	ROW_NUMBER() OVER ( 
	   PARTITION BY t1.idcard, 
					t1.disease, 
					TO_CHAR(t1.TIME, 'YYYYMMDD') 
		   ORDER BY t1.center ASC) AS rank
	FROM (SELECT af.id,
				 af.idcard,
				 af.disease,
				 af.time,
				 0 AS center
			FROM ai_follow AS af
		 UNION ALL
		 SELECT cfh.id,
				cfh.idcard,
				1 AS "disease",
				cfh.time,
				1 AS center
		   FROM center_follow_hbp AS cfh
		 UNION ALL
		 SELECT cfd.id,
				cfd.idcard,
				2 AS "disease",
				cfd.time,
				1 AS center
		   FROM center_follow_dm AS cfd
	) AS t
) t1 
join 
(
SELECT t.*, 
	 concat(idcard, '-', disease, '-', TO_CHAR(t.time, 'YYYYMMDD')) AS agg_idt
FROM (SELECT af.id,
			 af.idcard,
			 af.disease,
			 af.way,
			 af.weeks,
			 af.time,
			 af.plan_time,
			 NULL AS next_time,
			 af.utime,
			 unix_timestamp(af.utime) * 1000 AS utimestamp,
			 af.status,
			 af.cid,
			 af.ctime,
			 af.sync_state,
			 af.sync_time,
			 af.sync_user_id,
			 NULL AS org_code,
			 NULL AS org_name,
			 0 AS center
		FROM ai_follow AS af
	 UNION ALL
	 SELECT cfh.id,
			cfh.idcard,
			1 AS "disease",
			cfh.way,
			cfh.weeks,
			cfh.time,
			NULL AS plan_time,
			cfh.next_time,
			cfh.utime,
			cfh.utimestamp,
			cfh.status,
			NULL AS cid,
			NULL AS ctime,
			NULL AS sync_state,
			NULL AS sync_time,
			NULL AS sync_useid,
			cfh.org_code,
			cfh.org_name,
			1 AS center
	   FROM center_follow_hbp AS cfh
	 UNION ALL
	 SELECT cfd.id,
			cfd.idcard,
			2 AS "disease",
			cfd.way,
			cfd.weeks,
			cfd.time,
			NULL AS plan_time,
			cfd.next_time,
			cfd.utime,
			cfd.utimestamp,
			cfd.status,
			NULL AS cid,
			NULL AS ctime,
			NULL AS sync_state,
			NULL AS sync_time,
			NULL AS sync_useid,
			cfd.org_code,
			cfd.org_name,
			1 AS center
	   FROM center_follow_dm AS cfd) AS t
) t2
on t2.id = t1.id
;

3、查询

select /*+ view_pullup_flag(1) filter_push_down(1) enable_hash_join(0) */ * from view_follow where id = ?

以上，缺少环境未经验证（可能存在一些小错误需纠正），但提供了一个思路。您可以验证下效果，预计10s这样是可以的。