中文 | English
中文 | En

Introduction

Since 2012, the nation has successively introduced policies to promote the integrated development of health care and big data. From the "Healthy China" strategy to the "two centenary" goals, big data of health care has become an innovative need of national strategy, and the core technology and the key tools needed for national rejuvenation. In this context, the first China ECG AI Contest (CEAC), co-sponsored by School of Clinical Medicine and Institute of Data Science, Tsinghua University, and several key hospitals, came into being and was officially launched on December 20, 2018. The contest was highly valued by leaders of Tianjin Municipal Health Commission and Hebei Provincial Health Commission. The total prize money of the contest is expected to reach one million RMB.

The contest aims to promote the development of China's ECG AI and provide open source data and algorithm platforms for ECG AI researchers.

Registration Closed

Highlights

More than15,000high-quality resting ECG records

354participating teams worldwide

Three rounds-final hackathon36hours

Application trial of contest data

The contest includes a total of 1,000 routine ECGs, with 600 records in the training set and 400 records in the test set. This contest will ensure the safety of medical data from the perspective of data security. The ECG data used in the contest has been all desensitized. The data of the training set data contains disease diagnosis annotation, and all data cannot be traced back to ensure the data security. Moreover, JD Cloud provides data security to ensure that data cannot be downloaded and leaked to protect patients' privacy. If you also want to make an algorithm model based on the real medical data of this contest, or have a product demand for contest solutions, you can submit a form to apply.

相关解决方案：云桌面

Apply Now

Voice from competitors

中国心电智能大赛初赛开源与总结

首届中国心电智能大赛项目，初赛要求为设计一个自动识别心电图波形算法。笔者使用Keras框架设计了基于Conv1D结构的模型，并且开源了代码作为Baseline。内容包括数据预处理，模型搭建，网络训练，模型应用等，此Baseline采用最简单的一维卷积达到了88%测试准确率。有多支队伍在笔者基线代码基础上调优取得了优异成绩，顺利进入复赛。

AI从心电智能大赛到医疗应用 – 越光医疗一等奖算法落地之路

此文应首届“中国心电智能大师赛”举办方京东云和清华数据院的邀请，作为大赛的获奖单位代表，以及越光医疗CTO，分享和探讨越光医疗在心电智能诊断领域的理解，应用与实践成果。

View More

Prize settings

First prize:

  • 800,000RMB in bonus
  • Award certificates

Second prize:

  • 100,000RMB in bonus
  • Award certificates

Third prize:

  • 86,000RMB in bonus
  • Award certificates

Finalist award:

  • 14,000RMB in bonus
  • Award certificates

Contest moments

Organizers

清华大学
清数
京东云
1

Recruitment stage

January 1st-March 31st, 2019

2

Recruitment stage

April 8-20, 2019

3

Semifinals

May 5-June 10, 2019

4

Final stage

July 26th-August 11th, 2019

5

Presentation and sharing stage

September-November 2019

1

Recruitment stage

1.1-3.31

2

Recruitment stage

4.8-4.20

3

Semifinals

5.5-6.10

4

Final stage

7.2-8.11

5

Presentation and sharing stage

9.1-11.6

Contest review

Recruitment stage (registration, training)
  • Time: January 1st-March 31st, 2019
  • Registration: Volunteer teams will automatically enter the preliminary stage after completing the registration on the official website. At that time, the competition organizing committee will issue an online preliminary start notice;
  • Training: During the recruitment period, in order to help everyone understand the competition-related information such as the schedule and competition system, the CEAC Organizing Committee will also organize online and offline promotion and interactive activities such as university sharing, expert forums and other online and offline promotion;
Preliminary stage
  • Time: April 8-20, 2019
  • For all the participating teams, the professional qualifications of the competition will be screened. The top 100 teams from the successfully registered teams will be selected to be officially qualified for the competition based on the results of the topics of the competition. Teams can download data through the competition platform, debug algorithms locally, and submit results online. There is no limit to the place of entry.
Semifinals
  • Time: May 5-June 10, 2019
  • Teams that successfully enter the semi-finals upload codes through the competition system and run the program online. Each team can apply for up to three test set results evaluations, and there is no limit to the location of the competition. According to the competition system, the competition system will select the highest score as the entry score and finally select the team with a score greater than or equal to 0.85 to advance to the final. The organizing committee will review the results submitted by the participating teams, review the codes and technical solutions submitted by the promoted teams, identify and eliminate teams suspected of cheating, and fill in the vacant places after promotion.
Final stage
  • Time: July 26th-August 11th, 2019
  • Location: Online + Jingjin Gaocun Science and Technology Innovation Park, Wuqing District, Tianjin
  • The offline part of the finals will be conducted in a 36-hour non-stop and fully enclosed manner. The teams that qualify for the finals must report relevant information of team members in advance;
  • Selection method: The final results comprehensively consider algorithm performance, time and space complexity and innovation, etc., the competition awards will be selected and grand awards and achievement sharing meeting will be held.
Presentation and sharing stage
  • Time: September-November 2019
  • Venue: Beijing Great Wall Conference
  • The award ceremony of the first China ECG AI Contest will be held at the 2019 Beijing Great Wall Conference. At the same time, the winning team will be invited to communicate, cooperate, and share results with leading companies and innovators in various industries around the world.
Recruitment stage (registration, training)
  • Time: January 1st-March 31st, 2019
  • Registration: Volunteer teams will automatically enter the preliminary stage after completing the registration on the official website. At that time, the competition organizing committee will issue an online preliminary start notice;
  • Training: During the recruitment period, in order to help everyone understand the competition-related information such as the schedule and competition system, the CEAC Organizing Committee will also organize online and offline promotion and interactive activities such as university sharing, expert forums and other online and offline promotion;
Preliminary stage
  • Time: April 8-20, 2019
  • For all the participating teams, the professional qualifications of the competition will be screened. The top 100 teams from the successfully registered teams will be selected to be officially qualified for the competition based on the results of the topics of the competition. Teams can download data through the competition platform, debug algorithms locally, and submit results online. There is no limit to the place of entry.
Semifinals
  • Time: May 5-June 10, 2019
  • Teams that successfully enter the semi-finals upload codes through the competition system and run the program online. Each team can apply for up to three test set results evaluations, and there is no limit to the location of the competition. According to the competition system, the competition system will select the highest score as the entry score and finally select the team with a score greater than or equal to 0.85 to advance to the final. The organizing committee will review the results submitted by the participating teams, review the codes and technical solutions submitted by the promoted teams, identify and eliminate teams suspected of cheating, and fill in the vacant places after promotion.
Final stage
  • Time: July 26th-August 11th, 2019
  • Location: Online + Jingjin Gaocun Science and Technology Innovation Park, Wuqing District, Tianjin
  • The offline part of the finals will be conducted in a 36-hour non-stop and fully enclosed manner. The teams that qualify for the finals must report relevant information of team members in advance;
  • Selection method: The final results comprehensively consider algorithm performance, time and space complexity and innovation, etc., the competition awards will be selected and grand awards and achievement sharing meeting will be held.
Presentation and sharing stage
  • Time: September-November 2019
  • Venue: Beijing Great Wall Conference
  • The award ceremony of the first China ECG AI Contest will be held at the 2019 Beijing Great Wall Conference. At the same time, the winning team will be invited to communicate, cooperate, and share results with leading companies and innovators in various industries around the world.

Competition Rules

Preliminary Rules

Semifinal Rules

Final Rules

Participants

This competition will recruit teams from all over the world, including teams from universities, hospitals, entrepreneurial teams and other personnel who are interested in the development of China's ECG AI Contest. Priority is given to encouraging cross-medical teams to participate. Each team has no more than 3 participants.

Prize settings

First prize:

  • 800,000RMB in bonus
  • Award certificates

Second prize:

  • 100,000RMB in bonus
  • Award certificates

Third prize:

  • 86,000RMB in bonus
  • Award certificates

Finalist award:

  • 14,000RMB in bonus
  • Award certificates

Contest format

  • 1. The participating teams use artificial intelligence algorithms to identify and classify the standardized ECG signal provided by the contest; the competition provides standard training data sets for contestants to train algorithm models; provides standard test data sets for contestants to submit evaluation results and participate in the ranking. All the ECG data used in the competition has been desensitized, including training set and test set data. The training set data contains disease diagnosis annotations.
  • 2. Except for the preliminary stage, the participating teams need to submit algorithm programs on their own dedicated computing servers, train and verify ECG data online, and finally submit source code, executable files (such as model files and related support files) and run instructions that the performance of the algorithm can be evaluated by the competition committee.
  • 3. The competition organizing committee will objectively and fairly calculate the accuracy and performance of each competition team through a program, and comprehensively obtain the competition scores to rank and determine the promotion team. In the final stage, the final competition ranking will be selected through the competition scores combined with the technical solution defense. The organizing committee of the competition strictly respects the intellectual property rights of the participating teams and protects the algorithm programs through technical means. At the same time, the participating teams are also required to abide by the organizing committee's regulations on the use of ECG data and confidentiality.
# Team Name Awards Score Defense Results F1
Beat Beat Heart 一等奖 86.22 B 0.8608
星云 一等奖 85.76 A 0.8562
医工结合队 一等奖 88.28 B 0.8816
4 越光医疗 一等奖 85.48 A 0.8533
5 ECG小分队 二等奖 84.93 B 0.8477
6 机器不思议 二等奖 85.34 B 0.8519
7 视源医疗 二等奖 85.39 B 0.8524
8 心电诊断小分队 二等奖 83.57 A 0.8340
9 ustc-ecg 二等奖 84.85 B 0.8470
10 道格研究会 三等奖 81.35 A 0.8116
11 DeepHeart 三等奖 84.07 B 0.8391
12 点数成金 三等奖 82.31 C 0.8213
13 读心者 三等奖 83.30 B 0.8313
14 EVA 三等奖 83.59 C 0.8342
15 E心医心 三等奖 83.79 B 0.8363
16 newneu 三等奖 83.00 C 0.8409
17 容我想想 三等奖 84.24 C 0.8409
18 武汉中旗生物医疗 三等奖 84.36 B 0.8420
19 西瓜探险 三等奖 82.84 C 0.8266
20 BIT-BME 入围奖 77.91 B 0.7769
21 data_sense 入围奖 0 C 0
22 医电第一队 入围奖 82.26 C 0.8208
# Team Name Name F1 Status
机器不思议 陈搏 0.886738 晋级
landmind 韩伟 0.883947 晋级
星云 蔡文杰 0.879754 晋级
4 医工结合队 夏周睿 0.878938 晋级
5 红鲤鱼绿鲤鱼与驴 黄成斌 0.875693 晋级
6 DeepHeart 何润南 0.874636 晋级
7 读心者 熊袁鹏 0.873774 晋级
8 石原伊辽 贾东亚 0.872136 晋级
9 越光医疗 邢文辉 0.871788 晋级
10 DT_learning 李毅 0.871072 晋级
11 data_sense 朱佳兵 0.867107 晋级
12 医电第一队 李雪梅 0.865307 晋级
13 一命二运三风水 陈龙庭 0.865225 晋级
14 容我想想 杨珊 0.862571 晋级
15 西瓜探险 严浩文 0.862259 晋级
16 ustc-ecg 张静 0.861657 晋级
17 点数成金 周书毅 0.860354 晋级
18 Beat Beat Heart 张驰名 0.856906 晋级
19 道格研究会 王轲 0.855028 晋级
20 EVA 邸跃 0.85501 晋级
21 BIT-BME 葛传斌 0.854603 晋级
22 newneu 祖春山 0.853812 晋级
23 喵喵喵 李全池 0.851159 晋级
111 DenseCare toysoldier 机器不思议
626小分队 DingFuRen TP-link 临时组队
996-ICU dofuer99 ustc-ecg 梦之队
Adam DT_learning xBD 喵喵喵
AIecg2.0 DunDun zoncareDL 黏黏有瑜
AL-H ecg_AI 宾北计算 青春三人行
ANT-TRIO ECG共同体 从你的全世界路过 容我想想
ANT-YYZ e-med 大连万达 睿仁医疗
ASDFPOI EVA 大龙 深思科技
Beat Beat Heart GD－ECG 道格研究会 什么队
Beyond ictECG 点数成金 石原伊辽
Adam DT_learning xBD 喵喵喵
BigDream ITS-DS-CD 独自上场 曙光会
BIT group JQI-MLAI-LAB 读心者 数创医疗 AlphaECG
BIT-BME landmind 福大农大省立医院 腾云心电
BlackLeopard NEU-AI 歌瑞米娅 提莫队长
BME826 newneu 共致开源 天桥神经网络调参摆摊
BPBP NULL小队 海星通 拖油瓶小组
bty SCNUCloudLab 翰纬医疗AI 我不是佩琪
data_sense Tesla_X 红鲤鱼绿鲤鱼与驴 我是真的萌萌哒
DeepHeart TorchBoys 红医心电 我心
雾凇代表队 小可爱 小幸运队 斜月三星
心电信号分析与智能诊断 心电智能 心狗深度学习 心立方
星云 轩宇信息医工结合团队 雪践煖鸢 一命二运三风水
心电信号分析与智能诊断 医电第一队 医工001 医工结合队
医工先锋 医数结合 移动健康智慧医疗 越光医疗
张氏姐妹花

Papers

Machine Learning and Medical Engineering for Cardiovascular Health and Intravascular Imaging and Computer Assisted Stenting

The participating teams of this competition submit and write together

See Details

Notebooks

中国心电智能大赛初赛开源与总结

首届中国心电智能大赛项目，初赛要求为设计一个自动识别心电图波形算法。笔者使用Keras框架设计了基于Conv1D结构的模型，并且开源了代码作为Baseline。内容包括数据预处理，模型搭建，网络训练，模型应用等，此Baseline采用最简单的一维卷积达到了88%测试准确率。有多支队伍在笔者基线代码基础上调优取得了优异成绩，顺利进入复赛。

AI从心电智能大赛到医疗应用 – 越光医疗一等奖算法落地之路

此文应首届“中国心电智能大师赛”举办方京东云和清华数据院的邀请，作为大赛的获奖单位代表，以及越光医疗CTO，分享和探讨越光医疗在心电智能诊断领域的理解，应用与实践成果。

View More