Search

BACKGROUND Microscopic inflammation has been shown to be an important indicator of disease activity in ulcerative colitis (UC). However, manual histologic scoring is semi-quantitative and subject to interobserver variation, and AI-based solutions often lack interpretability. Here we report two distinct quantitative approaches to predict disease activity scores and histological remission using AI-powered digital pathology. Both the random forest classifier (RFC) and graph neural network (GNN) further provide explainability and biological insight by identifying histological features informing model predictions. METHODS Convolutional neural networks (CNNs) were developed using >162k annotations on 820 WSI of H&E-stained colorectal biopsies for pixel-level identification of tissue regions (e.g. crypt abscesses, erosion/ulceration) and cell types (e.g. neutrophils, plasma cells). All WSI were scored by 5 board-certified pathologists using the Nancy Histological Index (NHI) to establish consensus ground truth. A rich, quantitative set of human interpretable features that capture CNN predictions of the tissue region and cell type across each WSI was extracted and used to train a RFC to predict slide-level NHI score. To test the hypothesis that tissue region spatial relationships and cellular composition can inform AI-based predictions of disease activity, a separate GNN was trained, using nodes defined by spatially-resolved CNN model-generated outputs, to predict NHI score. The RFC and GNN also predicted histologic remission (NHI RESULTS The RFC and GNN both predicted histologic remission with high accuracy (weighted kappa 0.87 and 0.85, respectively). Both models also identified histologic features relevant to disease activity predictions. Some features are well established, e.g. infiltrated epithelium or neutrophil cell features distinguish cases with histologic remission. The models also identified features beyond those assessed by the NHI, e.g. area proportion of basal plasmacytosis associated with predictions of NHI 2 and 3. Other features not previously implicated in UC disease activity were also identified, e.g. intraepithelial lymphocytes differentiate cases with NHI 3. CONCLUSIONS We report quantitative and interpretable AI-powered approaches for UC histological assessment. CNN identification of UC histology was used as input to two distinct disease activity classifiers that showed strong concordance with consensus pathologist scoring. Both approaches provide interpretable features that explain model predictions and that may be used to inform biomarker selection and clinical development efforts.

Ulcerative colitis (UC) is a chronic inflammatory bowel disease that is characterized by a relapsing and remitting course. Appropriate assessment of disease activity is critical for adequate treatment decisions. In addition to endoscopic mucosal healing, histologic remission is emerging as a treatment target and a key factor in the evaluation of disease activity and therapeutic efficacy. However, there is no standardized definition of histologic remission, limiting the utility of histologic scoring, and manual pathologist evaluation is subject to intra-and inter-observer variability. Machine learning approaches are increasingly being developed to aid pathologists in accurate and reproducible scoring of histology, and can enable sensitive assessment of clinically relevant features. Here we report a proof-of-concept study using the PathAI platform to develop ML models for identification and quantification of UC histological features directly from hematoxylin and eosin (H&E)-stained whole slide images. Model-predicted histological features were used to quantify tissue area proportions and cell count proportions and densities, which correlated with disease severity and pathologist-assigned Nancy Histological Index (NHI) scores. Moreover, using multivariate analysis based on selected model-predicted histological features, we were able to accurately predict NHI scores, with a weighted kappa (k=0.93) and Spearman correlation (ρ=0.93, p

Background Health research apps often do not focus on usability as a design priority. This is problematic when the population of interest is disproportionately underrepresented as users of mobile apps, especially observed with aging older adults (> = 75). Challenges with the adoption of health information technology (HIT) among this group are exacerbated by poor design and user interface/experience (UI/UX) choices. This protocol describes the testing and evaluation process of one HIT app for the family-based collaboration platform InfoSAGE. Methods We aim to recruit twenty subjects from both informal family-caregivers and aging older adults to examine the usability of the InfoSAGE mobile medication manager. Participants will be audio and visually recorded, in addition to the use of screen capture recordings, while ‘thinking aloud’ as they complete eight common use-case scenarios. Multiple independent reviewers will code video and audio recordings for thematic analysis and use problems will be evaluated. Success and failure of each scenario will be determined by completion of sub-events. Time-to-complete analysis will be used to ascertain the learning curve associated with the app. Discussion Frequently observed problem areas will be used as the basis of further evolution of the app, and will further inform generalized recommendations for the design of HIT apps for research and public use. This study aims to improve the model of development for dual user populations with dissimilar technological literacy to improve retention and use. Results of this study will form the foundation of a design framework for mobile health apps.

Objective Adherence to antiretroviral therapy (ART) is paramount to successful long-term suppression of human immunodeficiency virus (HIV). For poorly adherent patients with HIV, barriers to remaining adherent may be overcome by the implementation of targeted interventions delivered via mobile devices. This systematic review is focused specifically on mobile phone technologies to deliver adherence interventions in HIV/acquired immunodeficiency syndrome (AIDS) populations. Methods This review (PROSPERO #CRD42017065131) systematically extracted data from published literature from five databases on mobile phone interventions to improve adherence to ART for HIV. The reported studies had been conducted between 2007 and 2017. Risk of bias was assessed using the Cochrane method ranking each criterion as low, high, or unclear risk of bias. Results Of the 835 articles returned, we identified 26 randomized controlled trials (RCTs), retrospective and prospective cohort trials, or mixed method studies with a comparison group that fit criteria for inclusion. No standard measure of adherence was consistent throughout the examined studies, and assessments by self-report, pill counting, and medication event monitoring system (MEMS) were utilized. The studies reported mixed results, with 17 reporting significant improvements to adherence, 3 reporting improvements without supplying p-values, and 6 reporting no significant change or a reduction in adherence. Conclusion The mixed nature of the results exemplifies the need for more comprehensive approaches and larger scale trials to confirm results observed in limited cohort sizes. To better retain satisfactory adherence within the HIV population, and especially in low-resource settings, we recommend that future interventions incorporate multiple strategies: mobile-based reminders, social support structures, and personalized content.