TwinWeaver: An LLM-Based Foundation Model Framework for Pan-Cancer Digital Twins
This work addresses the problem of predicting clinical outcomes for cancer patients, offering a scalable and interpretable solution that improves forecasting accuracy and risk stratification, though it is incremental as it builds on existing LLM and time-series methods.
The paper tackles the challenge of forecasting clinical events and trajectories from sparse, multi-modal clinical time series in precision oncology by introducing TwinWeaver, an LLM-based framework that serializes patient histories into text, and uses it to build Genie Digital Twin (GDT) on 93,054 patients across 20 cancer types, achieving a median MASE of 0.87 compared to 0.97 for baselines and an average C-index of 0.703 versus 0.662.
Precision oncology requires forecasting clinical events and trajectories, yet modeling sparse, multi-modal clinical time series remains a critical challenge. We introduce TwinWeaver, an open-source framework that serializes longitudinal patient histories into text, enabling unified event prediction as well as forecasting with large language models, and use it to build Genie Digital Twin (GDT) on 93,054 patients across 20 cancer types. In benchmarks, GDT significantly reduces forecasting error, achieving a median Mean Absolute Scaled Error (MASE) of 0.87 compared to 0.97 for the strongest time-series baseline (p<0.001). Furthermore, GDT improves risk stratification, achieving an average concordance index (C-index) of 0.703 across survival, progression, and therapy switching tasks, surpassing the best baseline of 0.662. GDT also generalizes to out-of-distribution clinical trials, matching trained baselines at zero-shot and surpassing them with fine-tuning, achieving a median MASE of 0.75-0.88 and outperforming the strongest baseline in event prediction with an average C-index of 0.672 versus 0.648. Finally, TwinWeaver enables an interpretable clinical reasoning extension, providing a scalable and transparent foundation for longitudinal clinical modeling.