TwinWeaver: An LLM-Based Foundation Model Framework for Pan-Cancer Digital Twins

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 errors, 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.

Cancer care could be more personalized if doctors could reliably forecast how a patient's disease will unfold — but real-world medical records are messy, mixing lab tests, treatments, genetic results, and diagnoses recorded at irregular intervals. Existing AI tools usually handle only a handful of variables, rely on rigid medical code lists that can't accommodate new drugs or mutations, or can't combine numerical forecasts (like predicting blood values) with event predictions (like survival or disease progression). We built TwinWeaver, a framework that rewrites a patient's medical history as plain text, so a large language model (the kind of AI behind chatbots) can read it like a story. Using this approach, we developed a model called Genie Digital Twin on records from 93,054 cancer patients across 20 cancer types. The same model forecasts future lab values, predicts clinical events, and can be extended to even explain its reasoning in everyday language. Genie Digital Twin outperforms specialized prediction tools, including on clinical trials it never saw during training. By providing accurate, transparent predictions, it could help personalize treatment decisions, accelerate drug development, and bring better forecasting to patients with rarer cancers where data is scarce.