Imperial College London

Cross-species single-cell ageing-state classifier that transfers mouse age labels to human HSC and CD8+ T cells, reaching 0.953 held-out AUROC.

Yeast sequence-to-expression model predicting strand-specific RNA-seq coverage across a 5 kb multi-gene window at 10 bp resolution.

Bacterial promoter annotation and expression prediction from a 1.8M-parameter transformer pretrained on 9M gammaproteobacterial regulatory regions.

Gradient-free protein design framework that treats engineering as Monte Carlo sampling of a user-defined energy landscape over pretrained models.

Multimodal foundation model for cardiac biosignals, pretrained by masked modeling on ECG, PPG, and clinical text from ~1.7 million individuals.

EEG foundation model whose codebook tokenizer encodes Fourier phase on the unit circle, gaining six points of balanced accuracy over LaBraM.

Whole-heart segmentation foundation model for CT and MRI, pretrained self-supervised on unlabeled cardiac scans with an xLSTM-UNet backbone.

2B-parameter medical vision-language model that uses reinforcement learning to show interpretable reasoning for radiology visual question answering.

Graph deep learning framework fusing frozen protein language model embeddings with structure graphs to predict per-residue flexibility in antibodies.

Autoregressive genomic foundation models from 20M to 1B parameters that solve ten DNA tasks at once and map sequences to text and images.

Cell Painting generative model encoding lab, batch, and well position as causal variables, predicting mechanism and target for unseen compounds.

Vision-language chat model for 3D chest CT volumes, answering free-form questions and drafting radiology report findings from a frozen 3D encoder.

Cardiac MR vision foundation model self-supervised on 36 million images, fine-tuned for segmentation, view classification and pathology detection.

Universal brain lesion segmentation for multi-modal brain MRI, using a Mixture of Modality Experts to span diverse modalities and lesion types.

Vision-language model for 3D chest CT that aligns whole volumes with radiology reports for zero-shot abnormality detection and case retrieval.

Vision-language pretraining for 3D CT volumes, aligning scans with their radiology reports for zero-shot classification, retrieval, and segmentation.

Metabolite-likeness scoring ranks any chemical structure by its distance from a learned hypersphere of known endogenous metabolites.