在网上搜到的一种算法是利用自增长变量进行排序，然后再根据位置序号取。感觉有些复杂了，还是group_concat来的省事些
1. 按顺序聚合，逗号分隔，并计数
group_concat( number order by number asc)
2. 根据逗号拆分，判断奇偶数去截取中间位置的那个数
具体代码如下：
SELECT
doctor_name doctor, -- 分组
count(1) patientNum, -- 总数
group_concat(dnt order by dnt asc),
substring_index(SUBSTRING_INDEX(group_concat(dnt order by dnt asc),‘,‘,(count(1)+1) div 2),‘,‘,-1) dnt,
case when count(1)%2=1 then substring_index(SUBSTRING_INDEX(group_concat(dnt order by dnt asc),‘,‘,(count(1)+1) div 2),‘,‘,-1) else (substring_index(SUBSTRING_INDEX(group_concat(dnt order by dnt asc),‘,‘,(count(1)+2) div 2),‘,‘,-1) +
substring_index(SUBSTRING_INDEX(group_concat(dnt order by dnt asc),‘,‘,count(1) div 2),‘,‘,-1))/2
end mid_dnt
FROM
(
SELECT distinct doctor_name, record_id, dnt
from rp_green_channel_patient_detaile
where dnt is not null
AND visit_day >= ‘2020-03-30‘
AND visit_day <= ‘2020-06-27‘
) AS a
group by doctor_name
原文：https://www.cnblogs.com/skyEva/p/13362371.html