【数据科学赛】Brain Age Prediction Challenge #回归问题 #时间序列 #计算机辅助诊断

CompHub 实时聚合多平台的数据类(Kaggle、天池…)和OJ类(Leetcode、牛客…）比赛。本账号会推送最新的比赛消息，欢迎关注！

更多比赛信息见 CompHub主页 或 点击文末阅读原文

---

以下内容摘录自比赛主页

Part1赛题介绍

题目

Brain Age Prediction Challenge

举办平台

CondaLab

主办方

背景

计算精神病学是一种新方法，其中算法不仅用于管理和组织数据，还用于了解患者隐藏的生理和行为信号。它允许计算机辅助诊断 (computer aided diagnosis，CAD) 以及通过生成模型更深入地了解病情本身。CAD侧重于识别受试者的可测量的生物学或生理学特征，需要有规范数据来做支持。

一个有前景的应用是使用大脑MRI数据预测脑龄，即通过从受试者的神经成像数据推断受试者的年龄。

Part2时间安排

Part3奖励机制

$1,600

Part4赛题描述

The problem is a regression problem. Each subject is characterized by time series of EEG recording, with eyes opened and eyes closed. You must predict the age of the individual.

You are given for training 2400 raw EEG files (MNE format) from 1200 subjects, each with recording of 129 electrodes, sampled at 500 Hz for 20s (eyes open) or 40s (eyes close). They hold data matrix of dimension 1200 x 2 x 129 x 10000 (20000 for eye closed) and an array of 1200 age values, one for each subject. You must train a model which predicts the age, and then your model will be evaluated on a test dataset of 400 subjects (800 EEG files) where the age is unknow to you.

---