Lauren Erdman

INTRODUCTION: Emerging evidence implicates eosinophils as important modulators of disease activity and therapeutic response in ulcerative colitis (UC). Automated image analysis provides a scalable and reproducible approach to their evaluation, overcoming the limitations of manual quantification. METHODS: We developed a U-Net-based deep learning model to automate eosinophil detection, using whole-slide images from 2 pediatric sites. Pathologist-annotated 500 × 500-pixel patches served as training data; model performance was evaluated using area under the receiver operator curve (AUROC), precision, recall, and F1 score and benchmarked against SAU-Net, YOLO-v8, and HoVer-Next. The validated algorithm was applied to rectal biopsies from treatment-naïve children with moderate-to-severe UC in the PROTECT inception cohort (n = 221). Eosinophil metrics (density in cells/mm ² and peak high-power field counts) were analyzed for associations with baseline clinical, histologic, and endoscopic features and 1-year clinical outcomes including anti-tumor necrosis factor α (TNFα) escalation and corticosteroid-free clinical remission. Predictive models were evaluated with logistic regression and machine learning approaches using 5-fold cross-validation. RESULTS: The segmentation model achieved high accuracy (test AUROC = 0.94, F1 score = 0.86) and strong concordance with expert counts (ρ = 0.87–0.89). Higher eosinophil density and counts correlated with basal lymphoid aggregates ( P < 0.001) and predicted week 4 and corticosteroid-free remission ( P = 0.03 and P = 0.01, respectively). Lower eosinophil counts predicted escalation to anti-TNFα therapy ( P = 0.04). Incorporation of eosinophil metrics into Predicting Response to Standardized Pediatric Colitis Therapy (PROTECT) clinical models maintained predictive accuracy for anti-TNFα escalation (AUROC = 0.79; accuracy = 0.76). DISCUSSION: Deep learning-based eosinophil quantification has the potential to transform pediatric UC care by serving as a reproducible biomarker that informs risk stratification and to guide therapy.

Artificial intelligence (AI) has the potential to transform healthcare for all patients. Yet, there are disproportionately fewer paediatric AI studies and US Food and Drug Administration approvals relative to adults, indicating less effort focused on AI for children. Here, given that innovations in medical AI are accelerated by community-driven research on public datasets, we hypothesized that the disparity in AI for paediatrics is tied to the lack of public paediatric medical imaging data to support their development and evaluation. To that end, we systematically identified and reviewed 203 datasets, revealing that 33% of datasets lacked metadata on patient ages, and when available, children represented less than 2% of patients. To illustrate how a lack of paediatric data can lead to harmful algorithmic bias, we trained models to classify adult cardiomegaly and evaluated them on healthy children. We found a consistent pattern of age-related bias, reproducible across four large-scale public chest X-ray datasets. These findings suggest that the lack of public paediatric data hinders the development of safe AI for children, producing a landscape of adult-first and adult-only AI models with unknown patterns of bias in children.

Implantable neurotechnologies are increasingly used to reduce seizure burden in pediatric epilepsy. Vagus nerve stimulation (VNS), the most common option, is effective for only half of patients, with no means to predict outcome prior to surgery. As a result, many children undergo invasive and costly procedures without benefit. Although T1-weighted magnetic resonance imaging (T1w) is routinely acquired presurgically and may capture structural brain differences relevant to treatment outcome, its high dimensionality relative to sample sizes has limited its utility in predictive modelling. To address this challenge, we present VQ-VNS, a deep representation learning model to predict VNS outcome based on preoperative T1w (n = 263). First, we present data from the largest paediatric VNS cohort (n = 1046), wherein presurgical clinical data could not predict response (AUC 0.54,p > 0.99). Next, VQ-VNS was pretrained on 7433 T1w images to learn compact anatomical representations enabling its classifier to predict VNS response (AUC = 0.73,p = 0.007). Model predictions localized to serotonin-rich brain regions and inferred large-scale disruptions in network connectivity among non-responders. This biologically interpretable predictor based on routine structural imaging improves upon current clinical decision-making.

A ‘silent trial’ refers to the prospective, noninterventional testing of artificial intelligence (AI) models in the intended clinical setting without affecting patient care or institutional operations. The silent evaluation phase has received less attention than in silico algorithm development or formal clinical evaluations, despite its increasing recognition as a critical phase. There are no formal guidelines for performing silent AI evaluations in healthcare settings. We conducted a scoping review to identify silent AI evaluations described in the literature and to summarize current practices for performing silent testing. We screened the PubMed, Web of Science and Scopus databases for articles fitting our criteria for silent AI evaluations, or silent trials, published from 2015 to 2025. A total of 891 articles were identified, of which 75 met the criteria for inclusion in the final review. We found wide variance in terminology, description and rationale for silent evaluations, leading to substantial heterogeneity in the reported information. Overwhelmingly, the papers reported measurements of area under the curve and similar metrics of technical performance. Far fewer studies reported verification of outputs against an in situ clinical ground truth; when reported, the approaches varied in comprehensiveness. We noted less discussion of sociotechnical components, such as stakeholder engagement and human–computer interaction elements. We conclude that there is an opportunity to bring together diverse evaluative practices (for example, from data science, human factors and other fields) if the silent evaluation phase is to be maximally effective. These gaps mirror challenges in the effective translation of AI tools from computer to bedside and identify opportunities to improve silent evaluation protocols that address key needs.

Background. Artificial intelligence (AI) is transforming healthcare, but its benefits have not been equitably distributed, with children being particularly overlooked. Only 17% of FDA-approved medical AI devices are labeled for pediatric use. We hypothesized that this disparity may be due to a fundamental data gap in pediatric medical imaging. Methods. To test this hypothesis, we performed a systematic review of 180 publicly available medical imaging datasets to assess pediatric data representation. To identify the primary data sources used for methods development, we first surveyed papers from a machine learning imaging conference. Finally, we evaluated the performance of adult-trained chest radiograph models when applied to pediatric populations to quantify potential age-related bias. Results . Our systematic review found that children represent less than 1% of the data in public medical imaging datasets. The majority of machine learning conference papers we surveyed relied on publicly available data for model development. Furthermore, we found that adult-trained chest radiograph models exhibit significant age bias when applied to pediatric populations, with higher false positive rates in younger children. Discussion. This study highlights the urgent need for increased pediatric representation in publicly accessible medical datasets. Our findings suggest that the lack of pediatric data may contribute to the scarcity of AI tools for children and the poor performance of adult-trained models in this population. We provide actionable recommendations for researchers, policymakers, and data curators to address this age equity gap and mitigate the potential harms of AI systems not trained on pediatric patients.