Our research in NLP for Health focuses on transforming unstructured clinical documents into structured, actionable intelligence.
We developed advanced deep learning architectures for automated medical coding, patient outcome prediction, and complex biomedical reasoning, ensuring that healthcare providers can leverage the full depth of electronic health records.
Key Objectives
Automated Medical Coding
Implementing state-of-the-art neural networks to map clinical notes to standardized ICD codes, addressing challenges like label imbalance and document noise.
Information Extraction
Leveraging graph embeddings and weak supervision to identify medical entities and detect adverse drug events across diverse clinical texts.
Biomedical Reasoning
Developing logical query models and multitask hypernetworks to predict patient outcomes and reason about complex drug-drug interactions.
Publications
A Unified Review of Deep Learning for Automated Medical Coding
Shaoxiong Ji, Xiaobo Li, Wei Sun, Hang Dong, Ara Taalas, Yijia Zhang, Honghan Wu, Esa Pitkänen, and Pekka Marttinen
TransFOL: A Logical Query Model for Complex Relational Reasoning in Drug-Drug Interaction
Junkai Cheng, Yijia Zhang, Hengyi Zhang, Shaoxiong Ji, and Mingyu Lu
Patient Outcome and Zero-shot Diagnosis Prediction with Hypernetwork-guided Multitask Learning
Shaoxiong Ji and Pekka Marttinen
Weak Supervision and Clustering-Based Sample Selection for Clinical Named Entity Recognition
Wei Sun, Shaoxiong Ji, Tuulia Denti, Hans Moen, Oleg Kerro, Antti Rannikko, Pekka Marttinen, and Miika Koskinen
Multitask Balanced and Recalibrated Network for Medical Code Prediction
Wei Sun, Shaoxiong Ji, Erik Cambria, and Pekka Marttinen
Automated Clinical Coding: What, Why, and Where We Are?
Hang Dong, Matúš Falis, William Whiteley, Beatrice Alex, Joshua Matterson, Shaoxiong Ji, Jiaoyan Chen, and Honghan Wu
Contextualized Graph Embeddings for Adverse Drug Event Detection
Ya Gao, Shaoxiong Ji, Tongxuan Zhang, Prayag Tiwari, and Pekka Marttinen
Medical Code Assignment with Gated Convolution and Note-Code Interaction
Shaoxiong Ji, Shirui Pan, and Pekka Marttinen