Background Though significant progress in disease elimination has been made over the past decades, trachoma is the leading infectious cause of blindness globally. Further efforts in trachoma elimination are paradoxically being limited by the relative rarity of the disease, which makes clinical training for monitoring surveys difficult. In this work, we evaluate the plausibility of an Artificial Intelligence model to augment or replace human image graders in the evaluation/diagnosis of trachomatous inflammation—follicular (TF). Methods We utilized a dataset consisting of 2300 images with a 5% positivity rate for TF. We developed classifiers by implementing two state-of-the-art Convolutional Neural Network architectures, ResNet101 and VGG16, and applying a suite of data augmentation/oversampling techniques to the positive images. We then augmented our data set with additional images from independent research groups and evaluated performance. Results Models performed well in minimizing the number of false negatives, given the constraint of the low numbers of images in which TF was present. The best performing models achieved a sensitivity of 95% and positive predictive value of 50–70% while reducing the number images requiring skilled grading by 66–75%. Basic oversampling and data augmentation techniques were most successful at improving model performance, while techniques that are grounded in clinical experience, such as highlighting follicles, were less successful. Discussion The developed models perform well and significantly reduce the burden on graders by minimizing the number of false negative identifications. Further improvements in model skill will benefit from data sets with more TF as well as a range in image quality and image capture techniques used. While these models approach/meet the community-accepted standard for skilled field graders (i.e., Cohen’s Kappa >0.7), they are insufficient to be deployed independently/clinically at this time; rather, they can be utilized to significantly reduce the burden on skilled image graders.

Background: The clinical characterization of the functional status of active wounds remains a considerable challenge that at present, requires excision of a tissue biopsy. In this pilot study, we use a convolutional Siamese neural network architecture to predict the functional state of a wound using digital photographs of wounds in a canine model of volumetric muscle loss (VML). Materials and Methods: Images of volumetric muscle loss injuries and tissue biopsies were obtained in a canine model of VML. Gene expression profiles for each image were obtained using RNAseq. These profiles were then converted to functional profiles using a manual review of validated gene ontology databases. A Siamese neural network was trained to regress functional profile expression values as a function of the data contained in an extracted image segment showing the surface of a small tissue biopsy. Network performance was assessed in a test set of images using Mean Absolute Percentage Error (MAPE). Results: The network was able to predict the functional expression of a range of functions based with a MAPE ranging from ~5% to ~50%, with functions that are most closely associated with the early-state of wound healing to be those best-predicted. Conclusions: These initial results suggest promise for further research regarding this novel use of ML regression on medical images. The regression of functional profiles, as opposed to specific genes, both addresses the challenge of genetic redundancy and gives a deeper insight into the mechanistic configuration of a region of tissue in wounds. As this preliminary study focuses on the first 14 days of wound healing, future work will focus on extending the training data to include longer time periods which would result in additional functions, such as tissue remodeling, having a larger presence in the training data.

Long COVID is recognized as a significant consequence of SARS-COV2 infection. While the pathogenesis of Long COVID is still a subject of extensive investigation, there is considerable potential benefit in being able to predict which patients will develop Long COVID. We hypothesize that there would be distinct differences in the prediction of Long COVID based on the severity of the index infection, and use whether the index infection required hospitalization or not as a proxy for developing predictive models. We divide a large population of COVID patients drawn from the United States National Institutes of Health (NIH) National COVID Cohort Collaborative (N3C) Data Enclave Repository into two cohorts based on the severity of their initial COVID-19 illness and correspondingly trained two machine learning models: the Long COVID after Severe Disease Model (LCaSDM) and the Long COVID after Mild Disease Model (LCaMDM). The resulting models performed well on internal validation/testing, with a F1 score of 0.94 for the LCaSDM and 0.82 for the LCaMDM. There were distinct differences in the top 10 features used by each model, possibly reflecting the differences in type and amount of pathophysiological data between the hospitalized and non-hospitalized patients and/or reflecting different pathophysiological trajectories in the development of Long COVID. Of particular interest was the importance of Plant Hardiness Zone in the feature set for the LCaMDM, which may point to a role of climate and/or sunlight in the progression to Long COVID. Future work will involve a more detailed investigation of the potential role of climate and sunlight, as well as refinement of the predictive models as Long COVID becomes increasingly parsed into distinct clinical phenotypes.

No abstract