Scientific Discovery & Drug Design - Weekly Roundup

Robust RL policies for adaptive cancer therapy achieve superior performance and generalization by treating patients as hidden contexts, outperforming existing methods even with noisy biomarkers and shifting disease dynamics.

A foundation model now surpasses the Cox model on survival analysis, even for small clinical cohorts, by predicting survival curves in-context.

Fine-tuning wind-field super-resolution models with persistent homology and merge tree losses dramatically improves topological structure reconstruction, outperforming standard methods and even the pretrained model.

AI weather models now forecast tropical cyclone tracks with superior probabilistic accuracy to leading physics-based systems, even while struggling with intensity.

A standard CNN technosignature detector misses 91% of genuine signals when they deviate slightly from the training morphology, and is fooled by terrestrial interference as often as it finds real signals.

Robust RL policies for adaptive cancer therapy achieve superior performance and generalization by treating patients as hidden contexts, outperforming existing methods even with noisy biomarkers and shifting disease dynamics.

A foundation model now surpasses the Cox model on survival analysis, even for small clinical cohorts, by predicting survival curves in-context.

Fine-tuning wind-field super-resolution models with persistent homology and merge tree losses dramatically improves topological structure reconstruction, outperforming standard methods and even the pretrained model.

AI weather models now forecast tropical cyclone tracks with superior probabilistic accuracy to leading physics-based systems, even while struggling with intensity.

A standard CNN technosignature detector misses 91% of genuine signals when they deviate slightly from the training morphology, and is fooled by terrestrial interference as often as it finds real signals.

A handful of chemically diverse unlabeled molecules can boost QSAR model performance more than massive, randomly sampled datasets.

A novel temporal graph framework achieves over 250x fewer FLOPs and 20x fewer parameters than Transformers/LSTMs for early gastrointestinal cancer risk stratification from sparse EHRs.

Causal direction can be identified by minimizing the Dirichlet energy of one variable treated as a signal on a kernel graph of the other, bypassing regression and independence tests.

Deep learning architectures designed for semantic understanding can now solve complex problems in algebraic topology, identifying topologically equivalent finite group actions on Riemann surfaces.

An adaptive acquisition function learned online via temporal difference learning discovers optimal perturbation experiments 2x more effectively than fixed methods.

Generative models can inject crucial scientific knowledge into GP posteriors, recovering hidden modes and improving predictive accuracy where traditional GPs fail.

Hyperspectral image denoising and destriping now achieve state-of-the-art accuracy without any parameter tuning, outperforming methods that require true noise levels and running orders of magnitude faster.

Spectral decomposition via a learnable Hodge-Laplacian unlocks a 47% reduction in interior MSE for steady-PDE surrogates on coarse meshes.

Scientific agent swarms can achieve state-of-the-art performance and generalize to new tasks by evolving through feedback-guided mutation and complementary recombination of hypotheses and code.

Novel disease trajectories discovered from EHRs can be effectively prioritized using a two-level semantic sparsity and reliability framework, leading to significant improvements in clinical forecasting accuracy.

A single pretrained tabular foundation model can solve diverse power grid analysis tasks, from power flow to contingency screening, without fine-tuning.

Training-free neural emulators achieve solver-level accuracy and out-of-distribution robustness by directly encoding known polynomial physics.

Branching motifs in EHRs offer structured trajectory context that primarily enhances long-range temporal forecasting, rather than uniformly improving endpoint discrimination or survival prediction.

Continual learning for molecular communication achieves robust performance estimation across evolving experimental complexities by intelligently synthesizing past data.

Generative models for PDEs can now recover fine-scale details and maintain long-horizon stability by learning residual distributions and aligning multiple trajectories.

Neural operators learn optimal derivative orders strictly below the PDE order, revealing a data-driven spectral regularization mechanism.

RNA inverse design is fundamentally a topology-and-geometry co-design problem, and TANGO achieves a 52.7% improvement in secondary-structure fidelity by explicitly modeling this.

A single turbulent flow realization, when intelligently patched, can unlock extreme-scale super-resolution, outperforming models trained on multiple full snapshots.

Staged PINNs recover power converter parameters with 5% accuracy for L/C and 10% for ESR, outperforming simultaneous training by an order of magnitude in waveform RMSE.

Continuous Markov chains can now have their pseudo-spectral gap estimated with minimax optimality, unlocking tighter concentration bounds and improved regret guarantees in bandit settings.

Conditional flow models hallucinate non-additive scientific interactions even when achieving high aggregate fidelity, failing to generalize interaction structure when condition patterns diverge.

Generative CT super-resolution can now reconstruct anatomically accurate vascular networks by explicitly guiding diffusion models with refined 3D vascular structures.

Diffusion models can learn lattice field theories without ever seeing real data, matching HMC performance.

Pretraining chemical language models yields diminishing returns, often underperforming classical baselines and even random initialization, suggesting current approaches capture little true chemistry.

LLMs can now learn to progressively optimize molecules across multiple turns by explicitly conditioning on past feedback, achieving substantial gains in similarity-constrained success rates.

Long-horizon agents can now achieve GPT-6 Astra and Claude Opus-level performance on complex tasks by learning from dense, progressive rewards derived from latent world specifications.

Neural networks now jointly infer CRN structure and kinetics from complex time-series data, outperforming traditional ABC methods.

Autonomous research agents can achieve state-of-the-art performance and efficiency by coupling rapid, parallel evidence assessment with strategic, deliberative planning.

Fine-grained, foundation-model-derived semantic representations unlock state-of-the-art whole-body PET/CT registration without task-specific fine-tuning.

Quantum state discrimination errors are reduced by 2x by encoding auxiliary qubit readout signals and dynamically calibrating neural decoders to the evolving circuit environment.

Manufacturing AI can now learn to avoid costly negative transfer by selectively leveraging prior knowledge, achieving 20% lower carbon footprint prediction error with minimal target data.

A single protein tokenizer now captures sequence, structure, and function simultaneously, unlocking joint generation and zero-shot retrieval across modalities.

Quantum state preparation can achieve four orders of magnitude lower infidelities for specific states by embracing risk-seeking optimization, rather than solely optimizing for average fidelity.

A theoretically grounded EI algorithm for continuous simulation optimization achieves simple-regret guarantees by learning unknown heteroscedastic noise and optimizing incumbent/recommendation strategies based on regret bounds.

Physically informed GNNs can accurately forecast advection-diffusion dynamics on irregular grids, outperforming existing methods and enhancing solar power predictions.

A conditional energy prior explicitly guides latent space towards perturbation-specific signals, unlocking an 83.2% improvement in perturbation recognition for cellular response prediction.

Generative AI and learner-facing applications are now the dominant themes in medical education research, but empirical evaluation and responsible integration remain critical unmet needs.