TwinWeaver: An LLM-Based Foundation Model Framework for Pan-Cancer Digital Twins
Listen
Radio episode about this paper
Transcript
Introduction to the show: ident: AI Radio. Generated commentary on the latest Artificial Intelligence papers.
Tom: We're looking at the paper: "TwinWeaver: An LLM-Based Foundation Model Framework for Pan-Cancer Digital Twins".
Jane: Here's the gist of the paper. Precision oncology requires forecasting clinical events and trajectories, yet modeling sparse, multimodal clinical time series remains a critical challenge TwinWeaver introduces an open-source framework that serializes longitudinal patient histories into text, enabling unified event prediction as well as forecasting with large language models This framework is used to build Genie Digital Twin (GDT) on 93,054 patients across 20 cancer types 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 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 Framework Architecture TwinWeaver is a general framework for constructing LLM-based patient digital twins from longitudinal, multi-modal data by fine-tuning LLMs on procedurally generated text representations of longitudinal patient journeys The framework serializes multi-modal EHR data into text to train the Genie Digital Twin (GDT) pan-cancer foundation model GDT jointly forecasts continuous biomarkers and predicts landmark clinical events Data Serialization and Training The training pipeline preprocesses longitudinal data, generates input-target pairs by splitting patient trajectories at critical time points, converts the pairs into text prompts, and fine-tunes a pretrained LLM with them The input to the model is defined as Xp(t) = (sp, Hp(t)), where Hp(t) =
hpi hpi ∈ Hp, hpi,timestamp ≤ t: The target Yp(t) is defined as a set of future predictions of various tasks The framework transforms the structured patient history Xp(t) and multi-task targets Yp(t) into a single, human-readable text prompt Forecasting and Event Prediction Tasks The first task type is time-series forecasting, used to predict frequently measured values such as blood biomarkers or vital signs The forecasting target Yforecast p (t forecast, V ′) comprises future values for the sampled variables V′ up to the forecasting horizon t forecast = t + ∆t forecast with ∆t forecast ∈ N + The second task is predicting a patient’s event status at a future time point using the landmarking framework The set of event labels is Y =
occurred, not occurred, censored: Performance and Generalization GDT consistently outperforms forecasting baselines across all indications, achieving a median MASE of 0.867 (IQR 0.186) aggregated across all variables GDT shows statistically significant improvements across the majority of modeled clinical outcomes for survival, progression, and therapy switching GDT demonstrates robust ranking capabilities in survival, progression, and switching therapy tasks The model also generalizes to out-of-distribution clinical trials, matching trained baselines at zero-shot and surpassing them with fine-tuning Reasoning Extension The TwinWeaver framework enables an interpretable clinical reasoning extension by fine-tuning LLMs on a synthetic corpus, followed by alignment via Group Relative Policy Optimization (GRPO) This extension provides transparent, biologically grounded context for hypothesis generation by extracting structured chains-of-thought conditioned on patient history and ground-truth outcomes Keyword analysis confirms that GDT generated reasoning chains correspond to clinically relevant features and genetic contexts Clinical Trial Performance In out-of-distribution clinical trial tasks, fine-tuned GDT shows strong performance, achieving the lowest MASE in blood biomarker forecasting (median MASE 0.883) and highest C-Indexes for survival and progression events The model demonstrates strong generalization in cold start scenarios where historical data is sparse Overall, our findings suggest that the TwinWeaver framework transforms pretrained LLMs using specific fine-tuning into clinical prediction engines, exemplified with GDT The ability of GDT to generalize to new clinical trials as well as its pan-cancer capabilities offers a promising solution for low-data scenarios Future work should focus on systematic hyperparameter optimization and integrating modalities such as histopathology images TwinWeaver indicates that LLMs can be adapted successfully as a platform for patient digital twins Software and Data Sources All framework and model code can be found at http://github.com/MendenLab/TwinWeaver The FH-FMI CGDB data that support the findings of this study were originated by and are the property of Flatiron Health, Inc.
Tom: and Foundation Medicine, Inc.
Jane: Anonymized records for individual patients across more than one data source external to Roche cannot, and should not, be linked due to a potential increase in risk of patient re-identification The study was funded by F.
Tom: Hoffmann-La Roche and the European Union’s Horizon 2020 Research and Innovation Programme Author Contributions N.M.
Paper summary: Tom: So we're looking at this paper called TwinWeaver: An LLM-Based Foundation Model Framework for Pan-Cancer Digital Twins, and basically, the main idea is that modeling those messy, sparse patient histories—the multimodal time series data—is really hard. This paper introduces TwinWeaver as this open-source framework that serializes all that longitudinal patient history into text.
Jane: That sounds like a big move because if you can turn patient journeys into text, it opens up a whole new way to use large language models for making predictions and forecasting. What's the core claim they are making about this framework?
Tom: They claim that by using TwinWeaver to build something called Genie Digital Twin or GDT on ninety-three thousand patients across twenty cancer types, you can actually get better at both predicting continuous biomarkers and forecasting those discrete clinical events.
Jane: So, the implication is that this isn't just a clever data processing trick; it's a way to use these models to do unified event prediction and forecasting using large language models. How does the paper back up that claim with some actual numbers?
Lu: The framework trains a pre-trained large language model by fine-tuning it on procedurally generated text representations of patient journeys, which they call the Genie Digital Twin foundation model. This process involves transforming structured longitudinal data into text prompts and then fine-tuning the LLM with those pairs.
Meng: Wait, so they are generating their own training data through this serialization process? What does that mean for the cost and complexity of building these digital twins at scale?
Lalam: The framework transforms the structured patient history Xp(t) and multi-task targets Yp(t) into a single, human-readable text prompt, which is how the model learns to connect everything together across different cancer types.
Tom: And the results are pretty strong. They say GDT significantly reduces forecasting error, achieving a median Mean Absolute Scaled Error or MASE of zero point eight seven compared to zero point nine seven for the strongest time-series baseline they tested, which is statistically significant at a p value less than zero point zero zero one <ref:2601.20906#pg1,MASE of 0.87 compared to 0.97 for the strongest time>.
Paper summary: Jane: That reduction from zero point nine seven down to zero point eight seven sounds substantial when you're dealing with sparse data and trying to predict future clinical events, right <ref:2601.20906#pg1>? What else are they showing us regarding how this digital twin performs for risk assessment?
Lu: Beyond forecasting, they also show that GDT improves risk stratification, achieving an average concordance index or C-index of zero point seven zero three across survival, progression, and therapy switching tasks <ref:2601.20906#pg1,C-index of 0.703 across survival, progression, and therapy switching tasks>. That's compared to the best baseline which scored a C-index of zero point six six two <ref:2601.20906#pg1>.
Meng: A C-index of zero point seven is decent for risk stratification in this context, but I wonder how robust these results are when you try to deploy this at scale on real patient data that isn't perfectly preprocessed like in their study <ref:2601.20906#pg1>.
Tom: They test the generalization a bit there too, and they find that GDT shows robust ranking capabilities across survival, progression, and switching therapy tasks. Plus, it generalizes to out-of-distribution clinical trials by matching trained baselines at zero shot and surpassing them with fine tuning.
Jane: So it suggests that this approach isn't just good for the specific data they used in their study; it has some real promise for applying these digital twins to new clinical situations or different patient populations. What about how this framework helps clinicians actually think?
Lu: They extended the reasoning capability by fine-tuning LLMs on a synthetic corpus, and then aligning that with group relative policy optimization or GRPO. This creates an interpretable clinical reasoning extension that gives transparent, biologically grounded context for generating hypotheses by extracting structured chains of thought conditioned on patient history and ground truth outcomes.
Lalam: Keyword analysis confirms that the reasoning chains generated by GDT correspond to clinically relevant features and genetic contexts, which is important because it means the model isn't just spitting out random text; it's linking the prediction back to known biology.
Paper summary: Meng: From an engineering standpoint, I’m interested in how they handle the inference part when you actually want a prediction for a new patient. They generate independent completions and average them for forecasting, which sounds like it adds a layer of stability to those predictions.
Tom: And for event prediction, they use a two-step post-processing strategy involving conditioning followed by isotonic calibration using the pool adjacent violators algorithm or PAVA to recover a consistent cumulative incidence function for individual patient analysis without needing retraining. That’s a solid engineering detail there.
Jane: So, while the performance metrics look strong across all those tasks—forecasting and event prediction—the paper also points out some limitations regarding its application in new clinical trials, especially concerning cold start scenarios where historical data is sparse.
Lu: Exactly. They note that while it performs well when fine-tuned, it still shows strong performance in out-of-distribution clinical trial tasks, specifically achieving the lowest MASE of zero point eight eight three in blood biomarker forecasting and highest C-indexes for survival and progression events there <ref:2601.20906#pg1>.
Tom: It sounds like the key thing here is that TwinWeaver proves that these large language models can be adapted successfully as a platform for patient digital twins, especially because they show this ability to generalize across different cancer types within one pan-cancer model.
Jane: So, looking at the whole picture of TwinWeaver: An LLM-Based Foundation Model Framework for Pan-Cancer Digital Twins, it seems like the authors are demonstrating a unified way to tackle the challenge of sparse, multimodal clinical time series by turning history into text for better prediction and risk stratification.
Tom: Yeah, it’s about taking this complex longitudinal data and giving the large language model a structured way to understand it so it can forecast what's coming next while also helping clinicians make better risk decisions based on that combined understanding.
Jane: And the implication is that we might see models like GDT being used not just for prediction, but for generating reasoned explanations about patient trajectories in a way that feels more grounded in clinical context.
Conclusion: Tom: So, TwinWeaver is this open source framework for building digital twins using large language models across all kinds of cancer data.
Jane: It basically takes all that messy patient history and turns it into text so the model can actually learn from it.
Tom: The authors, N.M., E.V.P., J.W., D.M., M.B., L.V, V, S and H are the team behind this thing; they built the whole system from scratch with a focus on using these LLMs for patient trajectories across twenty different cancer types at once—pan-cancer modeling is key here for them.
Lu: The core idea they present is serializing those longitudinal patient histories into text to train a foundation model, the Genie Digital Twin or GDT.
Meng: What I see as important from an engineering standpoint is that they're fine-tuning a large language model on these generated text representations of the patient journeys.
Lalam: Right, and they use this process to jointly forecast continuous measurements like blood markers and also predict specific clinical events.
Jane: The conclusion of the paper suggests that this framework shows how large language models can be adapted successfully to become actual engines for clinical prediction.
Tom: They show GDT does a really good job generalizing to new patient data, even in those out-of-distribution clinical trials, which is pretty cool for real-world use.
Lu: It also includes this reasoning extension using group relative policy optimization or GRPO, which gives the model a way to explain its predictions with structured chains of thought.
Jane: So it suggests that we're moving toward these LLMs being used not just as pattern matchers, but as tools that can actually generate grounded clinical reasoning.
Tom: Exactly; the implications here are pretty huge because if this works reliably across so many cancer types, it opens up a new way to handle the complexity of patient data in oncology.
Meng: The caveat is that they still point out cold start scenarios where there's very little historical data, and we need more work on optimizing the hyper-parameters for those sparse cases.
Jane: So while they’ve shown strong results, the next step for this technology has to be making it even better at handling those initial, sparsely populated patient profiles.
Computational Sciences Center of Excellence, Roche Penzberg Germany Computational Health Center Helmholtz Munich Department of Biology Ludwig Maximilian University of Munich Early Development Oncology Roche Innovation Center Zurich Computational Sciences Center of Excellence Genentech South San Francisco Department of Computer Science Stanford University Center for Data Science New York University Computational Sciences Center of Excellence Roche Basel Department of Biochemistry and Pharmacology Bio21 Molecular Science and Biotechnology
cs.LG
Submitted: 2026-01-28
Updated: 2026-02-09
Journal ref: Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:85853-85893, 2026
Code: https://github.com/endenLab/TwinWeaver
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Importance score: 98/100
The gist: Precision oncology requires forecasting clinical events and trajectories, yet modeling sparse, multimodal clinical time series remains a critical challenge TwinWeaver introduces an open-source…
Key concepts
- TwinWeaver Framework
- A general framework designed to build LLM-based patient digital twins. It works by taking multi-modal clinical data (like lab results and history) and converting it into a standardized text format that fine-tunes large language models for patient understanding.
- Genie Digital Twin (GDT)
- A pan-cancer foundation model built using TwinWeaver. GDT learns from serialized patient journeys to jointly forecast continuous biomarkers and predict specific clinical events, such as disease progression or therapy switching, across many different cancer types.
- Data Serialization
- The process of converting complex, structured longitudinal patient data into a single, human-readable text prompt. This textual representation is the input used to train the LLM so it can learn from the entire patient history at once.
- Forecasting and Event Prediction Tasks
- Two main prediction goals: 1) Time-series forecasting for continuous values like blood biomarkers, predicting future measurements. 2) Landmark event prediction, classifying whether a specific clinical event (like survival or progression) will occur in the future.
Terminology
Summary
Precision oncology requires forecasting clinical events and trajectories, yet modeling sparse, multimodal clinical time series remains a critical challenge TwinWeaver introduces an open-source framework that serializes longitudinal patient histories into text, enabling unified event prediction as well as forecasting with large language models This framework is used to build Genie Digital Twin (GDT) on 93,054 patients across 20 cancer types GDT significantly reduces forecasting error, achieving a median Mean Absolute Scaled Error (MASE) of 0. p. 087 compared to 0. p. 097 for the strongest time-series baseline Furthermore, GDT improves risk stratification, achieving an average concordance index (C-index) of 0. p. 0703 across survival, progression, and therapy switching tasks, surpassing the best baseline of 0. p. 0662
Framework Architecture
TwinWeaver is a general framework for constructing LLM-based patient digital twins from longitudinal, multi-modal data by fine-tuning LLMs on procedurally generated text representations of longitudinal patient journeys The framework serializes multi-modal EHR data into text to train the Genie Digital Twin (GDT) pan-cancer foundation model GDT jointly forecasts continuous biomarkers and predicts landmark clinical events
Data Serialization and Training
The training pipeline preprocesses longitudinal data, generates input-target pairs by splitting patient trajectories at critical time points, converts the pairs into text prompts, and fine-tunes a pretrained LLM with them The input to the model is defined as Xp(t) = (sp, Hp(t)), where Hp(t) = [hpi hpi ∈ Hp, hpi,timestamp ≤ t] The target Yp(t) is defined as a set of future predictions of various tasks The framework transforms the structured patient history Xp(t) and multi-task targets Yp(t) into a single, human-readable text prompt
Forecasting and Event Prediction Tasks
The first task type is time-series forecasting, used to predict frequently measured values such as blood biomarkers or vital signs The forecasting target Yforecast p (t forecast, V ′) comprises future values for the sampled variables V′ up to the forecasting horizon t forecast = t + ∆t forecast with ∆t forecast ∈ N + The second task is predicting a patient’s event status at a future time point using the landmarking framework The set of event labels is Y = [occurred, not occurred, censored]
Performance and Generalization
GDT consistently outperforms forecasting baselines across all indications, achieving a median MASE of 0.867 (IQR 0. p. 6186) aggregated across all variables GDT shows statistically significant improvements across the majority of modeled clinical outcomes for survival, progression, and therapy switching GDT demonstrates robust ranking capabilities in survival, progression, and switching therapy tasks The model also generalizes to out-of-distribution clinical trials, matching trained baselines at zero-shot and surpassing them with fine-tuning
Reasoning Extension
The TwinWeaver framework enables an interpretable clinical reasoning extension by fine-tuning LLMs on a synthetic corpus, followed by alignment via Group Relative Policy Optimization (GRPO) This extension provides transparent, biologically grounded context for hypothesis generation by extracting structured chains-of-thought conditioned on patient history and ground-truth outcomes Keyword analysis confirms that GDT generated reasoning chains correspond to clinically relevant features and genetic contexts
Clinical Trial Performance
In out-of-distribution clinical trial tasks, fine-tuned GDT shows strong performance, achieving the lowest MASE in blood biomarker forecasting (median MASE 0.883) and highest C-Indexes for survival and progression events The model demonstrates strong generalization in cold start scenarios where historical data is sparse
Conclusion
Overall, our findings suggest that the TwinWeaver framework transforms pretrained LLMs using specific fine-tuning into clinical prediction engines, exemplified with GDT The ability of GDT to generalize to new clinical trials as well as its pan-cancer capabilities offers a promising solution for low-data scenarios Future work should focus on systematic hyperparameter optimization and integrating modalities such as histopathology images TwinWeaver indicates that LLMs can be adapted successfully as a platform for patient digital twins
Software and Data Sources
All framework and model code can be found at http://github.com/MendenLab/TwinWeaver The FH-FMI CGDB data that support the findings of this study were originated by and are the property of Flatiron Health, Inc. p. 8 p. 8 and Foundation Medicine, Inc. p. 8 p. 8 Anonymized records for individual patients across more than one data source external to Roche cannot, and should not, be linked due to a potential increase in risk of patient re-identification The study was funded by F. p. 8 p. 8 Hoffmann-La Roche and the European Union’s Horizon 2020 Research and Innovation Programme
Author Contributions
N.M. conceived and executed the project, while E. p. 8 p. 8V.P., J.W., and D.M. contributed to methodology and implementation of the software framework M. p. 8B., L.V.V., S.M., and H.W performed data preparation, experimental analysis and visualization RRE, FSE, and MPM supervised, designed and directed the project
References
Ansari et al. Chronos: Learning the language of time series is cited in reference 10 Cox's model is referenced in reference 10 Becker et al. An enhanced prognostic score for overall survival of patients with cancer derived from a large realworld cohort is cited in reference 2 Bedi et al. p. 9 How to interpret ‘zero-shot’ results from generative EHR models is cited in reference 10 Bordukova et al. p. 9 Generative artificial intelligence empowers digital twins in drug discovery and clinical trials is cited in reference 10 Guo et al. p. 9 Ehr foundation models improve robustness in the presence of temporal distribution shift is cited in reference 2 Gruver et al. p. 9 Large language models are zero-shot time series forecasters is cited in reference 10 Hemadri et al. p. 9 Oncoreason: Structuring clinical reasoning in llms for robust and interpretable survival prediction is cited in reference 10 Hoo et al. p. 9 The tabular foundation model tabpfn outperforms specialized time series forecasting models based on simple features is cited in reference 2 Hussain et al. p. 9 Joint ai-driven event prediction and longitudinal modeling in newly diagnosed and relapsed multiple myeloma is cited in reference 2
Additional References
Kamel Boulos and Zhang P. Digital twins: from personalised medicine to precision public health is cited in reference 10 Katzman et al. p. 9 Deepsurv: personalized treatment recommender system using a cox proportional hazards deep neural network is cited in reference 2 Latimer et al. p. 9 Nice dsu technical support document 16: adjusting survival time estimates in the presence of treatment switching is cited in reference 10 Makarov et al. p. 1 Large language models forecast patient health trajectories enabling digital twins is cited in reference 10 Orcutt et al. p. 0 Evaluating generalizability of oncology trial results to real-world patients using machine learning-based trial emulations is cited in reference 10 Orr et al. p. 0 A kinetic model of bone marrow neutrophil production that characterizes late phenotypic maturation is cited in reference 10 Pellegrini et al. p. 7 From ehrs to patient pathways: Scalable modeling of longitudinal health trajectories with llms is cited in reference 10 Pukelsheim. p. 0 The three sigma rule is cited in reference 10 Shao et al. p. 5 Tumor mutational burden as a predictive biomarker for non-small cell lung cancer treated with immune checkpoint inhibitors of PD-1/PD-l1 is cited in reference 2 Shao et al. p. 7 Deepseekmath: Pushing the limits of mathematical reasoning in open language models is cited in reference 10 Zheng et al. p. 0 Prognostic implications of STK11 with different mutation status and its relationship with tumor-infiltrating immune cells in non-small cell lung cancer is cited in reference 2 Zhuang et al. Advancing precision oncology through modeling of longitudinal and multimodal data is cited in reference 2
Data Sources
This study used the US-based deidentified Flatiron Health-Foundation Medicine Breast Cancer, Colorectal Cancer, Pancreatic Cancer, Ovarian Cancer, Metastatic Prostate Cancer, Gastric and Esophageal Cancer, Endometrial Cancer, Melanoma, Advanced Urothelial Carcinoma Carcinoma Clinico-Genomic Database (FH-FMI CGDB) Clinical data from the Flatiron Health Research Database are linked to genomic data derived from FMI’s comprehensive genomic profiling (CGP) tests in the FH-FMI CGDB by deterministic matching
Methodological Details
The training pipeline transforms raw multi-modal inputs into a structured textual representation of longitudinal patient trajectories at weekly resolution Numeric and categorical variables are aggregated using their mean and mode, respectively The loss function is defined as L(θ) = − X N i=k log qθ(xi x1:i−1) where qθ is the LLM parameterized by θ The final ranking score of the event E is taken from the ‘occurred’ class probability using a log-likelihood score derived from softmax normalization The final risk score is taken from the ‘occurred’ class probability using log-likelihoods, which provide continuous scores required for downstream evaluation metrics
Inference Post-Processing
For forecasting inference, M independent completions are generated and then averaged to produce a stable mean prediction Yˆp forecast = 1/M PM m=1 Yˆp m For landmark event prediction, the final ranking score is taken from the ‘occurred’ class probability using log-likelihoods To recover a consistent Cumulative Incidence Function (CIF) for individual patient analysis, we employ a two-step post-processing strategy involving conditioning followed by isotonic calibration using the Pool Adjacent Violators Algorithm (PAVA) This ensures clinical interpretability without retraining the model
Evaluation Metrics
Forecasting performance is primarily evaluated using the Mean Absolute Scaled Error (MASE), where MASE < 1 indicates that the model provides superior forecasts to simply carrying forward the last known value For landmark event tasks, we assess risk ranking using the Inverse Probability of Censoring Weighting (IPCW) Concordance Index (C-index) A C-index of 0.5 implies random ranking, while 1. p. 2 p. 20 implies perfect discrimination The Brier Score measures the mean squared difference between the predicted probabilities and the actual outcomes
Extended Results
GDT with supervised fine-tuning (SFT) outperforms baselines on out-of-distribution clinical trial tasks, achieving the lowest MASE in blood biomarker forecasting GDT demonstrates improved risk stratification in predicting survival, progression, and switching therapy The model's performance is enhanced when incorporating richer patient histories for survival predictions The reasoning extension provides a calculated trade-off for clinical utility by providing transparent, biologically grounded context for hypothesis generation GDT consistently improves performance over single-indication models when trained on the pan-cancer dataset
Synthetic Example
A synthetic patient example demonstrates how the framework handles input and generates structured reasoning chains, including patient summary, key factors, mechanistic analysis, and confounding factors The model generates a structured reasoning chain that corresponds to clinically relevant features and genetic contexts This capability allows for hypothesis generation by providing different explanations for a patient’s response in clinical practice
Example Forecast Predictions
The example forecast predictions show the longitudinal trajectory of different blood-based markers across various tumor indications, with GDT capturing some shapes or values even when the average prediction is incorrect The longitudinal trajectory of different blood-based markers are shown for a variety of tumor indications
Data Table Summary
Table 3 provides a comprehensive breakdown of evaluated forecasting variables by LOINC code across all cancer indications, showing the coverage and relevance of the model across diverse patient populations The table details which variables are included for each specific cancer type The table shows that many variables are included for multiple indications, indicating the pan-cancer nature of GDT
Forecasting Example
In the example forecast predictions, the longitudinal trajectory of different blood-based markers are shown for a variety of tumor indications Measured values are displayed in grey and predicted values in blue The longitudinal trajectory of different blood-based markers are shown for a variety of tumor indications The Longitudinal trajectory of different blood-based markers are shown for a variety of tumor indications
In conclusion, results with GDT demonstrate the effectiveness of the TwinWeaver framework, indicating that LLMs can be adapted successfully as a platform for patient digital twins TwinWeaver indicates that LLMs can be adapted successfully as a platform for patient digital twins The study concludes that GDT demonstrates the effectiveness of the TwinWeaver framework The results with GDT demonstrate the effectiveness of the TwinWeaver framework
Software and Data
All framework and model code can be found at http://github.com/MendenLab/TwinWeaver The FH-FMI CGDB data that support the findings of this study were originated by and are the property of Flatiron Health, Inc. and Foundation Medicine, Inc. Anonymized records for individual patients across more than one data source external to Roche cannot, and should not, be linked due to a potential increase in risk of patient re-identification The study was funded by F. Hoffmann-La Roche and the European Union’s Horizon 2020 Research and Innovation Programme
N.M. conceived and executed the project, while E.V.P., J.W., and D.M contributed to methodology and implementation of the software framework M. p. 8B., L.V.V., S.M., and H.W performed data preparation, experimental analysis and visualization RRE, FSE, and MPM supervised, designed and directed the project
Forecasting performance is primarily evaluated using the Mean Absolute Scaled Error (MASE), where MASE < 1 indicates that the model provides superior forecasts to simply carrying forward the last known value For landmark event tasks, we assess risk ranking using the Inverse Probability of Censoring Weighting (IPCW) Concordance Index (C-index) A C-index of 0.5 implies random ranking, while 1.
Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.
Improvements for AI systems
-
Foundation models can be adapted into clinical prediction engines by fine-tuning them on procedurally generated text representations of longitudinal patient journeys, which allows them to
jointly forecast continuous biomarkers and discrete clinical event prediction.
-
The system can perform time series forecasting by generating multiple text-based completions from the model’s predictive distribution and averaging them to produce a stable mean prediction, achieving a median Mean Absolute Scaled Error (MASE) of 0.830 across the top 30 most time-varying variables in real-world data.
-
The system can improve risk stratification by predicting patient event status using landmarking, achieving an average concordance index (C-index) of 0.703 across survival, progression, and therapy switching tasks, surpassing the best baseline of 0.662.
-
The system demonstrates zero-shot generalization to clinical trials and can surpass trained baselines with fine-tuning, achieving a median MASE of 0.75–0.88 in event prediction and an average C-index of 0.672 versus 0.648 for event prediction.
-
The framework enables an interpretable clinical reasoning extension by using a reinforcement learning-based approach with a deterministic reward function defined by the negative Mean Absolute Error (MAE) to ground generated rationales in empirical accuracy, providing
transparent, biologically grounded context for hypothesis generation.
Sources
- AutoGluon-Tabular: Robust and Accurate AutoML for Structured Data
- OncoReason: Structuring Clinical Reasoning in LLMs for Robust and Interpretable Survival Prediction
- EHR2Path: Comprehensive Pathway-Level Modeling of Longitudinal Patient Trajectories from Multimodal Electronic Health Records
- DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models
- Qwen3 Technical Report
Related papers
- FlexMoE: Learning One-for-All Multi-Budget Expert Pruning for Flexible MoE Deployment
- OptiMind: Improving Small Language Models for Optimization Modeling through Formulation Guidance and Multi-Turn Learning
- Staying in the Sweet Spot: Boost Reasoning Evolution via Responsive Rollout Difficulty Manipulation
- Examining the Impact of Data and Training Methods on Emergent and Subliminal Misalignment
- Rethinking Black-Box On-Policy Distillation: A Rubric-based Analysis
- (Mis)generalization of Helpful-Only Fine-tuning