Chris Sainsbury

Chris Sainsbury, Feng Dong, Andreas Karwath

Autoregressive models can predict clinical events, but generating patient-conditioned multi-step trajectories that respond to intervention tokens and testing whether those responses preserve known pharmacological associations has received limited attention. We present FlatASCEND, a 14.5M-parameter autoregressive clinical sequence model using flat composite tokens and a zero-inflated log-normal time head. Standard distributional metrics (Jaccard 0.889-0.954) do not distinguish FlatASCEND from trivial baselines; the model's value lies in conditional generation from patient-specific prefixes. A prompt-shuffle ablation shows patient-specific conditioning amplifies mechanistic pharmacological effects (2.0-2.2x for steroid to glucose, diuretic to potassium) while leaving confounding-driven associations unchanged (0.9x for insulin to glucose). An incident-user framework assesses directional consistency against prior pharmacological knowledge on MIMIC-IV (N=500 per comparison): 4/10 recover correct mechanistic directions, 2 reproduce treatment-context associations, 4 are incorrect (9/10 significant, Wilcoxon p<0.05). This pattern - partial recovery under residual confounding - is consistent with learned observational associations without causal distinction. Direct preference optimisation with surrogate reward destroys all correct associations (3/3 to 0/3), illustrating reward exploitation when reward and evaluation share an outcome domain. Generative evidence is strongest for short-horizon ICU data; outpatient temporal fidelity is weaker (median 10 vs 154 days on INSPECT), and zero-shot cross-site transfer degrades without adaptation.

Chris Sainsbury, Feng Dong, Andreas Karwath

Sparse autoencoders (SAEs) have been applied to large language models and protein language models, but not systematically to electronic health record (EHR) foundation models. We train TopK SAEs on FlatASCEND, a 14.5-million-parameter autoregressive clinical sequence model, at all 10 residual stream extraction points on INSPECT (outpatient) and MIMIC-IV (ICU). SAE decomposition reveals progressive abstraction across transformer depth: layer-0 features are near-perfect token detectors (45.7% singleton), while layer-6 features span approximately 30 token types across multiple clinical categories (0.5% singleton). Under full-sequence simple linear probes, SAE features outperform dense representations for discrete event prediction (mortality) while dense representations outperform for continuous magnitude prediction (length of stay) - a probe-level representational phenomenon that does not extend to clinically relevant leakage-safe windows, where dense representations match or exceed SAE features across all tested settings (eICU-CRD 48-hour AUC: SAE 0.871 versus dense 0.880; base model zero-shot, SAE dictionaries trained on eICU activations; MIMIC-IV: 0.836 versus 0.914; INSPECT 1-year/3-year: 0.697 versus 0.800). A delta-mode intervention method reduces SAE perturbation noise by 86x, enabling cleaner feature-level experiments, though the resulting perturbation effects are larger than random controls in 3 of 4 conditions but not formally significant. Feature reproducibility across random seeds is 21%, and individual features should be interpreted as illustrative rather than stable.

Chris Sainsbury, Andreas Karwath

We present ASCENDgpt, a transformer-based model specifically designed for cardiovascular risk prediction from longitudinal electronic health records (EHRs). Our approach introduces a novel phenotype-aware tokenization scheme that maps 47,155 raw ICD codes to 176 clinically meaningful phenotype tokens, achieving 99.6\% consolidation of diagnosis codes while preserving semantic information. This phenotype mapping contributes to a total vocabulary of 10,442 tokens - a 77.9\% reduction when compared with using raw ICD codes directly. We pretrain ASCENDgpt on sequences derived from 19402 unique individuals using a masked language modeling objective, then fine-tune for time-to-event prediction of five cardiovascular outcomes: myocardial infarction (MI), stroke, major adverse cardiovascular events (MACE), cardiovascular death, and all-cause mortality. Our model achieves excellent discrimination on the held-out test set with an average C-index of 0.816, demonstrating strong performance across all outcomes (MI: 0.792, stroke: 0.824, MACE: 0.800, cardiovascular death: 0.842, all-cause mortality: 0.824). The phenotype-based approach enables clinically interpretable predictions while maintaining computational efficiency. Our work demonstrates the effectiveness of domain-specific tokenization and pretraining for EHR-based risk prediction tasks.