TEDDY: A Pediatric Foundation Model for Risk Forewarning from ICD-Coded Diagnostic Histories
Researchers introduced TEDDY, a 1.84‑million‑parameter transformer trained on 73 million ICD‑10 codes from 1.6 million children, to forecast future diagnoses and visit timing.
- TEDDY is a transformer model trained on 73 million ICD‑10 codes from 1.6 million children.
- It predicts future diagnoses and visit timing before codes are recorded.
- The model demonstrates the feasibility of pediatric foundation models for early clinical warning.
- Its release may spur further AI research focused on child health trajectories.
A team of researchers released TEDDY (Temporal Event Decoder for Disease in Youth), a decoder‑style transformer designed for pediatric electronic health records. The model contains 1.84 million parameters and was trained on roughly 73 million ICD‑10 diagnosis codes drawn from 1.6 million patients at a single children's hospital.
TEDDY captures longitudinal diagnosis trajectories and the timing of clinical visits, allowing it to predict upcoming diagnosis codes before they appear in a patient's record. The authors evaluated the model on held‑out data, demonstrating its ability to forecast both the type and timing of future health events.
The work highlights the untapped potential of generative foundation models in pediatric care, where developmental patterns differ from adult populations. By leveraging large‑scale coded health data, TEDDY could enable earlier interventions and more personalized treatment plans.
While the model is modest in size compared with massive language models, its specialized focus and extensive training data make it a notable contribution to AI‑driven healthcare research.
Provides a concrete example of applying transformer decoders to longitudinal health data.
Shows a potential tool for hospitals to improve early detection and care planning.
Highlights emerging value in AI‑enabled pediatric health analytics.
Offers a research case study on domain‑specific foundation models.
Illustrates how AI can help anticipate health issues in children.
- ICD-10
- International Classification of Diseases, 10th revision, a coding system for medical diagnoses.
- Transformer
- A neural network architecture that uses attention mechanisms to process sequences.
- Decoder transformer
- A transformer variant that generates outputs conditioned on input sequences, often used for prediction tasks.
Xi Jinping sets out China’s goal to be global AI leader - Financial Times
QFireNet: A Quantum-Enhanced U-Net for Wildfire Segmentation from Sentinel-2 Imagery
Certified Domain Consistency for Multi-Domain Retrieval: Label-Free Per-Domain Contamination Control with Conformal Risk Guarantees
Inference-Time Concept Suppression and Video-Centric Evaluation for Text-to-Video Models
KeyFrame-Compass: Towards Comprehensive Evaluation of Keyframe-Conditioned Video Generation
China’s Xi says AI ‘should not be a solo performance by a single country’ - Al Jazeera
Chinese President Xi Jinping stated that artificial intelligence development should not be dominated by a single country, calling for broader international cooperation.
Organize your curiosity: Generative AI tools prove adept at structuring volumes of information - Editor and Publisher
Generative AI tools are effective in structuring large volumes of information, helping to organize and make sense of complex data. This development has significant implications for various industries and applications.
SeeSE3: Emergence of 3D Space in Vision Features
A new paper titled SeeSE3 investigates whether vision foundation models inherently represent 3D Euclidean space. The authors introduce probes to measure the relationship between visual features and Euclidean transformations.
3D Lane Detection with Odometry for High-Speed Vehicle Racing
Researchers introduce a new dataset for 3D lane detection in high-speed racing and compare various approaches to this challenging problem.
SD-MAR: Multi-image Analytical Reasoning via Synthetic Data and Reinforcement Learning
Researchers introduce SD-MAR, a method using synthetic data and reinforcement learning to enhance multi-image analytical reasoning in Vision Language Models.
MixCompress: Mixture of Experts for Variable Rate Learned Image Compression
Researchers propose MixCompress, a new framework for variable rate learned image compression, addressing limitations of existing methods.