Background and Purpose

In modern medicine, artificial intelligence (AI) and big data are completely changing the diagnosis and treatment methods of traditional medicine. With the continuous deepening of the application of large-scale deep learning models in medical data processing, it is possible to assist doctors in making more accurate and rapid diagnoses. Through this competition, we aim to bring together outstanding AI researchers and developers from around the world to explore and implement intelligent medical diagnosis solutions based on electrocardiogram (ECG) data to promote the widespread application of AI technology in the medical field.

Competition Topic

Topic Description

The theme of this competition is "Auxiliary ECG Medical Diagnosis Based on AI". Participants need to use time series deep learning/machine learning tools to process and analyze the provided ECG data, and ultimately achieve disease prediction and diagnosis.

Dataset

This competition will provide two sets of data sets:

1. MIT-BIH+PTB-XL public database: contains more than 100,000 ECG data that have been cleaned, unified in format, and labeled. Participants can use this set of data sets for model training and verification. This data set is widely used in the field of electrocardiogram analysis and has high quality and reliability.
   
   
2. Anonymized clinical data: contains 100 ECG data without preprocessing and labeling. Participants need to learn how to normalize data and predict diseases based on the MIT-BIH+PTB-XL dataset and apply it to this set of clinical data. This competition ensures the security of medical data from the perspective of data security. The ECG data used in the competition have been desensitized, all data cannot be traced, and data cannot be downloaded or leaked to protect the privacy of patients.

Competition tasks

Participants need to complete the following tasks:

1. Data normalization: Use public datasets to train models and normalize clinical data.
   
2. Disease prediction: Predict diseases for clinical data based on the trained model, and the predicted labels are the same as the public dataset.
3. Result report: Make an EXCEL report of the prediction results, including the prediction results and related indicators for each data.
4. Results presentation: Make a PPT presentation of the results, and introduce the algorithm, model, prediction results and innovations in detail.
5. Code submission: Upload the complete code to the designated cloud platform and ensure that the code can run on the platform.

Submission content

Participants need to submit the following content:

1. Clinical data prediction results: Including the prediction results and related analysis of each data, submitted in EXCEL format.
   
2. Results presentation PPT: Detailed introduction of algorithms, models, prediction results and innovations to facilitate the judges' understanding and review.
3. Code files: Contains all the codes for implementing algorithms and predictions, and provides necessary operating instructions, which are uploaded to the designated cloud platform.

Evaluation criteria

The final evaluation of the competition will be comprehensively judged based on the following criteria:

1. Accuracy: The accuracy of clinical data prediction results.
   
2. Operability: The operability and practicality of the algorithm in the actual medical environment.
3. Visualization: The visualization and display effect of the results.
4. Privacy protection and security: The algorithm takes into account data security-related risks in practical applications.

Eligibility

This competition is open to the world. Participants are not limited to academic qualifications, majors and nationalities. All individuals and teams interested in AI medical algorithms are welcome to participate.

Schedule

The specific schedule of the competition is as follows:

• Submission deadline: August 28, 2024
  
• Review period: August 28, 2024 - August 31, 2024
• Result announcement time: September 1, 2024
• Offline Finals: September 20, 2024 (Haizhu District, Guangzhou)
• Award Ceremony: September 21, 2024 (Haizhu District, Guangzhou)

Awards

The total cash prize pool for this competition is 3.78 million RMB, provided by organizers of the Pazhou Algorithm Competition. In addition, there are other rewards such as AWS Cloud Credit.

The specific prize distribution is as follows (all prizes are before tax):

• Champion (1 winner): 100K RMB in cash and 175K RMB in AWS Cloud Credit provided by the Pazhou Algorithm Competition Organizing Committee, along with an award certificate.
• Runner-up (1 winner): 50K RMB in cash and 105K RMB in AWS Cloud Credit provided by the Pazhou Algorithm Competition Organizing Committee, along with an award certificate.
• Third Place (1 winner): 20K RMB in cash and 30K RMB in AWS Cloud Credit provided by the Pazhou Algorithm Competition Organizing Committee, along with an award certificate.

Furthermore, this competition track offers the following additional incentives:

The top scorer in each region during the semi-finals will receive a 20K RMB (before tax) cash reward from the Pazhou Algorithm Competition organizers.

The champion of this track's finals will have the opportunity to participate in the Pazhou Algorithm Competition Grand Finals, competing alongside champions from 14 other tracks for an additional 400K RMB cash prizes and 100K RMB AWS Cloud Credit. The overall champion team will also receive 5 million RMB in special support, a Guangzhou Talent Green Card, and a Pazhou CBD residence permit.

All winning teams will have the opportunity to receive up to 10 million RMB in start-up (project) implementation support from the Pazhou Algorithm Competition Organizing Committee, as well as priority investment of no less than 3 million RMB from venture capital institutions. Members of the winning teams will also enjoy a green service channel for talent settlement.

Notes

1. Originality: Participants must ensure that the submitted works are original and must not plagiarize or use other people's works. Any plagiarism found will result in disqualification.
2. Authorization: All submitted works will be used for academic research and non-commercial purposes. Participants must agree to authorize the organizer to use their works.
3. Running instructions: Please ensure that the submitted code can be run on the designated cloud platform, and provide detailed running instructions for the judges to review.
4. Data description: The anonymized clinical data provided in this competition can only be used for this competition and cannot be used for other commercial or private purposes.
5. Contestants need to submit the code and the prediction results on the test set within the specified time according to the layout requirements of the sample code (necessary). In order to ensure the fairness of the competition, the organizing committee will test the code. The organizing committee also respects the intellectual property rights of each participating team and will not disclose the relevant code after the competition.

We look forward to your participation and jointly promote the development and application of AI technology in the medical field. If you have any questions, please feel free to contact the organizing committee of the competition: ting.tan@aifc.ngo. I wish you excellent results!

背景和目的

在现代医疗中，人工智能（AI）和大数据技术正在彻底改变传统医疗的诊断和治疗方式。随着大规模深度学习模型在医疗数据处理中的应用不断深入，辅助医生进行更加准确和快速的诊断成为可能。通过这次赛题，我们旨在汇聚全球优秀的AI研究者和开发者，探索和实现基于心电图（ECG）数据的智能医疗诊断解决方案，以推动AI技术在医疗领域的广泛应用。

大赛题目

题目描述

本阶段赛题的主题是“基于ECG数据的智能医疗诊断”。参赛者需要使用时间序列的深度学习/机器学习工具，对提供的ECG数据进行处理和分析，最终实现疾病的预测与诊断。

数据集

本阶段赛题将提供两套数据集：

1. MIT-BIH+PTB-XL公开数据库：包含十万多条清洗完毕、格式统一、并且打好标签的ECG数据。参赛者可以使用这套数据集进行模型的训练和验证。该数据集被广泛应用于心电图分析领域，具有高质量和高可靠性。
   
   
2. 匿名化临床数据：包含一百条未经预处理和打标签的ECG数据，参赛者需要根据MIT-BIH+PTB-XL数据集学习如何进行数据的归一化和疾病预测，并将其应用于这套临床数据。本次比赛从数据安全角度保证医疗数据安全，竞赛所用的心电数据均已采取脱敏化处理，所有数据不可溯，数据不可下载外泄，保护患者的隐私。

参赛任务

参赛者在本阶段需完成以下任务：

1. 数据归一化：使用公开数据集训练模型，并对临床数据进行归一化处理。
   
2. 疾病预测：根据训练的模型对临床数据进行疾病预测，预测的标签与公开数据集相同。
3. 结果报告：制作预测结果的EXCEL报告，包括每条数据的预测结果和相关指标。
4. 成果展示：制作成果展示PPT，详细介绍算法、模型、预测结果和创新点。
5. 代码提交：将完整的代码上传至指定的云端平台，并确保代码可以在平台上运行。

提交内容

参赛者需提交以下内容：

1. 临床数据预测结果：包括每条数据的预测结果和相关分析，以EXCEL格式提交。
   
2. 成果展示PPT：详细介绍算法、模型、预测结果和创新点，便于评委理解和评审。
3. 代码文件：包含所有实现算法和预测的代码，并提供必要的运行说明，上传至指定的云端平台。

评价标准

大赛的最终评价将基于以下标准进行综合评判：

1. 准确率：临床数据预测结果的准确率。
   
2. 可操作性：算法在实际医疗环境中的可操作性和实用性。
3. 可视化程度：结果的可视化程度和展示效果。
4. 隐私保护与安全性：算法在实际应用中考虑到数据安全相关隐患。

时间安排

本阶段赛题的具体时间安排如下：

• 方案提交截止日期：2024年8月28日
  
• 评审期：2024年8月28日-2024年8月31日
• 晋级名单公布时间：2024年9月1日
• 线下决赛：2024年9月20日（广州市海珠区）
• 颁奖大会：2024年9月21日（广州市海珠区）

奖励

大赛设有丰厚的奖励，本届大赛现金总奖金池为378万元，由琶洲算法大赛组织方提供。此外还有算力劵等其他奖励。

具体奖金分布如下（所有奖金均为税前）：

• 冠军（1名）：琶洲算法大赛组委会提供的10万元人民币现金奖金和17.5万元算力券，颁发获奖证书。
• 亚军（1名）：琶洲算法大赛组委会提供的5万元人民币现金奖金和10.5万元算力券，颁发获奖证书。
• 季军（1名）：琶洲算法大赛组委会提供的2万元人民币现金奖金和3万元算力券，颁发获奖证书。

此外，本赛道还提供以下额外激励：

复赛期间各赛区成绩第一名将获得琶洲算法大赛赛事方提供的 2 万元人民币（税前）现金奖励。

本赛道决赛冠军将有机会参加琶洲算法大赛总决赛，与14个赛道的冠军同台竞技，争夺额外40万元人民币现金奖励和10万元算力券。总冠军团队还将获得500万元专项扶持、广州市人才绿卡、琶洲CBD户口。

所有获奖团队将有机会获得琶洲算法大赛组委会提供的最高1000万元的创业（项目）落地扶持和不少于300万元的创投机构资金优先投资。获奖团队成员还将享受人才落地绿色服务通道，并有机会与知名医疗机构和AI公司合作，共同推进AI医疗技术的发展。

注意事项

1. 原创性：参赛者需保证所提交的作品为原创，不得抄袭或使用他人作品。任何发现的抄袭行为将取消参赛资格。
2. 使用授权：所有提交的作品将用于学术研究和非商业用途，参赛者需同意授权主办方使用其作品。
3. 运行说明：请确保提交的代码可以在指定的云端平台运行，并提供详细的运行说明，以便评委进行评审。
4. 数据说明：本次比赛提供的匿名化临床数据仅可用于本次比赛使用，不可用于其他商业用途或私人用途。
5. 参赛选手需要按照样例代码的布局要求，在规定时间内提交代码和在测试集上的预测结果 （缺一不可） 。为了保证比赛的公平公正，组委会对代码进行测试。组委会同时尊重各参赛队伍的知识产权，在赛后不会公开相关代码。

我们期待您的参与，共同推动AI技术在医疗领域的发展和应用。如果有任何问题，请随时联系赛题组委会ting.tan@aifc.ngo。祝您取得优异的成绩！