Volume 5 Issue 8, August 2023

To protect the integrity of knowledge production, the training procedures of foundation models such as GPT-4 need to be made accessible to regulators and researchers. Foundation models must become open and public, and those are not the same thing.

Medical artificial intelligence needs governance to ensure safety and effectiveness, not just centrally (for example, by the US Food and Drug Administration) but also locally to account for differences in care, patients and system performance. Practical collaborative governance will enable health systems to carry out these challenging governance tasks, supported by central regulators.

Limited interpretability and understanding of machine learning methods in healthcare hinder their clinical impact. Imrie et al. discuss five types of machine learning interpretability. They examine medical stakeholders, highlight how interpretability meets their needs and emphasize the role of tailored interpretability in linking machine learning advancements to clinical impact.

Algorithmic super-resolution in the context of fluorescence microscopy is challenging due to the difficulty to reliably represent biological nanostructures in synthetically generated images. Bouchard and colleagues propose a deep learning model for live-cell imaging that can leverage auxiliary microscopy imaging tasks to guide and enhance reconstruction, while preserving the biological features of interest.

There are currently promising developments in deep learning for protein design, with applications in drug discovery and synthetic biology. For more efficient exploration of the design space, Wang et al. demonstrate a reinforcement learning method, EvoZero, for directed evolution in protein engineering towards desired functional or structure-related properties.

Out of the large number of neoepitopes, few elicit an immune response from the major histocompatibility complex. To predict which neoepitopes can be effective, Albert and colleagues present a method based on long short-term memory ensembles and transfer learning from immunogenicity assays.

To ensure that a machine learning model has learned the intended features, it can be useful to have an explanation of why a specific output was given. Slack et al. have created a conversational environment, based on language models and feature importance, which can interactively explore explanations with questions asked in natural language.

The tendency of machine learning algorithms to learn biases from training data calls for methods to mitigate unfairness before deployment to healthcare and other applications. Yang et al. propose a reinforcement-learning-based method for algorithmic bias mitigation and demonstrate it on COVID-19 screening and patient discharge prediction tasks.

Microscopic imaging and holography aim to decrease reliance on labelled experimental training data, which can introduce biases, be time-consuming and costly to prepare, and may lack real-world diversity. Huang et al. develop a physics-driven self-supervised model that eliminates the need for labelled or experimental training data, demonstrating superior generalization on the reconstruction of experimental holograms of various samples.

A challenging problem in deep learning consists in developing theoretical frameworks suitable to study generalization. Feng and colleagues uncover a duality relation between neuron activities and weights in deep learning neural networks, and use it to show that sharpness of the loss landscape and norm of the solution act together in determining its generalization performance.

Traditional feedback-state selection in robot learning is empirical and requires substantial engineering efforts. Yu et al. develop a quantitative and systematic state-importance analysis, revealing crucial feedback signals for learning locomotion skills.

Deep learning applied to live-cell images of patient-derived neurons aids predicting underlying mechanisms and gains insights into neurodegenerative diseases, facilitating the understanding of mechanistic heterogeneity. D’Sa and colleagues use patient-derived stem cell models, high-throughput imaging and machine learning algorithms to investigate Parkinson’s disease subtyping.