Research presented at the 2022 World Cancer Congress shows that deep learning algorithm outperformed radiologists in detecting head and neck cancer spread. The algorithm was able to detect small areas of cancer spread that were not detected by the radiologists. This is a potentially important finding, as early detection of cancer spread can lead to better outcomes for patients. Head and neck cancers kill about 400,000 people a year worldwide. Radiation with or without chemotherapy or surgery followed by chemotherapy is the most common treatment strategy.

Equally impressive to their algorithm's performance is their effort to validate their technique with images from multiple institutions, addressing the challenge of generalizability that many machine learning–based diagnostics face.2 However, their work raises a fundamental question that should be considered as algorithms begin to perform tasks that could previously be performed only by clinicians. Is the algorithm being asked to perform exactly the same task as its human counterpart? This question has important implications for evaluating the relative performance of the algorithm as well as for assessing the clinical significance of its findings. Initially, it appears that the algorithm and the radiologists are given the same task: to identify ENE in lymph nodes from CT scans.

Wikipedia is a huge opportunity for machine learning, being the largest semi-structured base of knowledge available. Because of this, many works examine its contents, and focus on structuring it in order to make it usable in learning tasks, for example by classifying it into an ontology. Beyond its textual contents, Wikipedia also displays a typical graph structure, where pages are linked together through citations. In this paper, we address the task of integrating graph ( i.e. structure) information to classify Wikipedia into a fine-grained named entity ontology (NE), the Extended Named Entity hierarchy. To address this task, we first start by assessing the relevance of the graph structure for NE classification. We then explore two directions, one related to feature vectors using graph descriptors commonly used in large-scale network analysis, and one extending flat classification to a weighted model taking into account semantic similarity. We conduct at-scale practical experiments, on a manually labeled subset of 22,000 pages extracted from the Japanese Wikipedia. Our results show that integrating graph information succeeds at reducing sparsity of the input feature space, and yields classification results that are comparable or better than previous works.

Extranodal extension (ENE) is a well-established poor prognosticator and an indication for adjuvant treatment escalation in patients with head and neck squamous cell carcinoma (HNSCC). Identification of ENE on pretreatment imaging represents a diagnostic challenge that limits its clinical utility. We previously developed a deep learning algorithm that identifies ENE on pretreatment computed tomography (CT) imaging in patients with HNSCC. We sought to validate our algorithm performance for patients from a diverse set of institutions and compare its diagnostic ability to that of expert diagnosticians.

Efficient Natural Evolution Strategies (eNES) is a novel alternative to conventional evolutionary algorithms, using the natural gradient to adapt the mutation distribution. Unlike previous methods based on natural gradients, eNES uses a fast algorithm to calculate the inverse of the exact Fisher information matrix, thus increasing both robustness and performance of its evolution gradient estimation, even in higher dimensions. Additional novel aspects of eNES include optimal fitness baselines and importance mixing (a procedure for updating the population with very few fitness evaluations). The algorithm yields competitive results on both unimodal and multimodal benchmarks.