林嶔博士於2016年四月博士畢業，主要專長在於洞悉資料型態並發展適合的演算法。目前主要的研究方向在於結合當前最火熱的深度學習技術與傳統的統計學，並應用於時間序列分析、電腦視覺、自然語言處理等任務上，試圖在醫學場域內建構精準的電腦輔助醫療系統與技術。
- Nov 2017 – Recent
Artificial intelligence for electrocardiogram analysis
Deep Learning, Decision support, Computer Aided Diagnosis, Wearable Device ECG12Net and its specific training process have been developed for ECG recognition.More than 50 clinical diagnoses supporting via ECG12Net which is better than cardiologists.The model has been deployed in hospital, and further application in out-of-hospital is conducting.Morphologies not in textbook has been identified and used to teach physicians.
- Jul 2018 – Recent
Deep learning algorithm design for medical data
Deep Learning, Statistics, Mathematics, Epidemiology Unsupervised learning aided deep neural network training to solve the rare data problem. Matched deep learning training strategy for learning causality.To design special prediction functions and loss functions to solve the problem of missing data.Multi-level statistical models enhanced deep learning for personalized artificial intelligence.
- Jul 2018 – Recent
Smart hospital development
Deep Learning, Computer Vision, Natural Language Processing
The self-developed annotation system is applied to service more than 30 researches.Automatic training process is used in X-ray, CT, MRI, endoscope, pathology, etc.Initial diagnosis and calculation for reducing physician loading.
The natural language processing system is applied to simplify routine works.
