Projects
Electrocardiography based acute cardiac diseases detection system
Participants

Shih-Hua Lin
Adviser

Chin Lin
Technical PI

Chin-Sheng Lin
Clinical PI

Wen-Hui Fang
Community PI

Shi-Hung Tsai
Emergency PI

Chia-Cheng Lee
IT manager

Sy-Jou Chen
Emergency physician

Wen-Yu Lin
Cardiologist

Ban-Yen Liu
Cardiologist

Wen-Cheng Liu
Cardiologist

Da-Wei Chang
Cardiologist

Yu-Lan Liu
Cardiologist

Chiao-Chin Lee
Cardiologist

Wei-Ting Liu
Cardiologist

Yu-Cheng Chen
Cardiologist

Chia-Jung Hsu
Engineer
Summary

Reference:
1npj Digital Medicine, 5(1):8.
2JMIR Med Inform, 8(3):e15931.
3EuroIntervention, 17(9):765-773.
4J Pers Med, 11(11):1149.
5Unpublished data.
6Can J Cardiol, 38(2):160-168.
7J Pers Med, 12(7):1150.
8Unpublished data.
9Eur J Trauma Emerg Surg, https://doi.org/10.1007/s00068-022-01904-3.
10Int J Environ Res Public Health, 18(7), 3839.
11J Endocr Soc, 5(9), bvab120.
12We shared our works to the Discovery.
Algorithm development
Participants

Chin Lin
Adviser

Yu-Sheng Luo
PhD student
Summary
Deep learning is a data-driven algorithm, so it is necessary to understand the limitations of medical data and design suitable algorithms to build an accurate medical artificial intelligence system. We consider that main differences between medical field and other fields and solutions are as follows:
1. Unsupervised learning aided supervised learning to solve the rare data problem: Because the medical data involves manual operations and expensive instruments, the amount of labeled data is scarce compared with other industries. Moreover, the incidence and prevalence of more fatel diseases is lower compared with other common diseases. We need to face this important limitation rather than selecting sufficient data for research. Transfer learning is a common solution in deep learning training that can effectively reduce the number of samples required. Our main work is to use a large amount of unlabeled data for unsupervised algorithms, and apply them to supervised learning.1
2. To design special prediction functions and loss functions to solve the problem of missing data: The data source is usually based on observational design in the medical field, so it often faces the missing data problem. For example, the typical follow up cohort will face the problem of right-censored data, and all the tests may not be performed every time. The survival analysis in the statistics gives us a good inspiration. By cutting the data into fine-grained time periods, we effectively use the information of all samples. We apply the masking idea to our sample for deep learning training in conventional medical data. It will solve the gap to directly apply traditional deep learning technology in the medical field.
3. Combination of multi-level statistical models and deep learning to create personalized artificial intelligence: The response of each indivial to the same treatment is different, so it is impossible to use one artificial intelligence models for accurately appling in everyone. Therefore, it is very important to establish a personalized artificial intelligence model, but it is difficult to train each person's data separately in medical field. The multi-level data analysis in statistics shows how to use the distribution assumptions to construct personalized predictions. Even if only one record also can be used to construct a personalized model, our research attempts to apply this concept to deep learning training. The new age medical artificial intelligence can dimamicly revise the prediction results.2
4. Causal relationship learning to confront the interference: Traditional medical researches focus on causality exploration, but the medical AI studies often ignored this issue. Since the powerful feature extraction ability from DLM, current AI often learns the spurious correlations to construct the prediction which reduce the accuracy in future external validation. We designed the matched training strategy to maximize the notable features learning by DLM, which significantly improves the accuracy in many tasks

Reference:
1Unpublished data.
2Unpublished data.
Electrocardiography based disease screening platform
Participants

Chin Lin
Technical PI

Yu-Juei Hsu
Clinical PI

Chin-Sheng Lin
Clinical PI

Wen-Hui Fang
Community PI

Ping-Huang Tsai
Nephrologist

Chiao-Hsiang Chang
Cardiologist

Hung-Yi Chen
Cardiologist

Chia-Jung Hsu
Engineer
Summary
Reference:
1J Pers Med, 12(3):455.
2Front Med, 9:870523.
3Unpublished data.
4J Pers Med, 11(8):725.
5Unpublished data.
6Unpublished data.
7Front Cardiovasc Med, 9:895201.
8Unpublished data.
9J Pers Med, 12(2):315.
10Unpublished data.
11Unpublished data.
12Unpublished data.
13Front Cardiovasc Med, 9:754909.
Medical image analysis
Participants

Chin Lin
Adviser

Yi-Ying Wu
Hematologist

Dung-Jang Tsai
Research fellow

Yu-Sheng Luo
PhD student

Ying-Chu Chen
MS student

Xin-An Lin
Engineer

Yu-Cheng Chen
Cardiologist

Hao-Chun Liao
BS student
Summary
Reference:
1JMIR Med Inform, 8(4):e15963.
2Can J Cardiol, S0828-282X(22)00004-6.
Medical Natural Language Processing
Participants

Chin Lin
Adviser

Wen-Hui Fang
Clinical expert

Dung-Jang Tsai
Research fellow

Yu-Sheng Luo
PhD student

Xin-An Lin
Engineer
Summary
Reference:
1J Med Internet Res, 19(11):e380.
2JMIR Med Inform, 7(3):e14499.
3Healthcare, 9(10):1298.