Stratifying risk for onset of type 1 diabetes using islet autoantibody trajectory clustering.

AIMS/HYPOTHESIS: Islet autoantibodies can be detected prior to the onset of type 1 diabetes and are important tools for aetiologic studies, prevention trials and disease screening. Current risk stratification models rely on the positivity status of islet autoantibodies alone, but additional autoantibody characteristics may be important for understanding disease onset. This work aimed to determine if a data-driven model incorporating characteristics of islet autoantibody development, including timing, type and titre, could stratify risk for type 1 diabetes onset. METHODS: Data on autoantibodies against GAD (GADA), tyrosine phosphatase islet antigen-2 (IA-2A) and insulin (IAA) were obtained for 1,415 children enrolled in The Environmental Determinants of Diabetes in the Young study with at least one positive autoantibody measurement from years 1 to 12 of life. Unsupervised machine learning algorithms were trained to identify clusters of autoantibody development based on islet autoantibody timing, type and titre. Risk for type 1 diabetes across each identified cluster was evaluated using time-to-event analysis. RESULTS: We identified 2-4 clusters in each year cohort that differed by autoantibody timing, titre and type. During the first 3 years of life, risk for type 1 diabetes onset was driven by membership in clusters with high titres of all three autoantibodies (1-year risk: 20.87-56.25%, 5-year risk: 67.73-69.19%). Type 1 diabetes risk transitioned to type-specific titres during ages 4 to 8, as clusters with high titres of IA-2A (1-year risk: 20.88-28.93%, 5-year risk: 62.73-78.78%) showed faster progression to diabetes compared with high titres of GADA (1-year risk: 4.38-6.11%, 5-year risk: 25.06-31.44%). The importance of high GADA titres decreased during ages 9 to 12, with clusters containing high titres of IA-2A alone (1-year risk: 14.82-30.93%) or both GADA and IA-2A (1-year risk: 8.27-25.00%) demonstrating increased risk. CONCLUSIONS/INTERPRETATION: This unsupervised machine learning approach provides a novel tool for stratifying risk for type 1 diabetes onset using multiple autoantibody characteristics. These findings suggest that age-dependent changes in IA-2A titres modulate risk for type 1 diabetes onset across 12 years of life. Overall, this work supports incorporation of islet autoantibody timing, type and titre in risk stratification models for aetiologic studies, prevention trials and disease screening.

A rare FGF5 candidate variant (rs112475347) for predisposition to nonsquamous, nonsmall-cell lung cancer.

A unique approach with rare resources was used to identify candidate variants predisposing to familial nonsquamous nonsmall-cell lung cancers (NSNSCLC). We analyzed sequence data from NSNSCLC-affected cousin pairs belonging to high-risk lung cancer pedigrees identified in a genealogy of Utah linked to statewide cancer records to identify rare, shared candidate predisposition variants. Variants were tested for association with lung cancer risk in UK Biobank. Evidence for linkage with lung cancer was also reviewed in families from the Genetic Epidemiology of Lung Cancer Consortium. Protein prediction modeling compared the mutation with reference. We sequenced NSNSCLC-affected cousin pairs from eight high-risk lung cancer pedigrees and identified 66 rare candidate variants shared in the cousin pairs. One variant in the FGF5 gene also showed significant association with lung cancer in UKBiobank. This variant was observed in 3/163 additional sampled Utah lung cancer cases, 2 of whom were related in another independent pedigree. Modeling of the predicted protein predicted a second binding site for SO4 that may indicate binding differences. This unique study identified multiple candidate predisposition variants for NSNSCLC, including a rare variant in FGF5 that was significantly associated with lung cancer risk and that segregated with lung cancer in the two pedigrees in which it was observed. FGF5 is an oncogenic factor in several human cancers, and the mutation found here (W81C) changes the binding ability of heparan sulfate to FGF5, which might lead to its deregulation. These results support FGF5 as a potential NSNSCLC predisposition gene and present additional candidate predisposition variants.

Prioritization of infectious epitopes for translational investigation in type 1 diabetes etiology.

Molecular mimicry is one mechanism by which infectious agents are thought to trigger islet autoimmunity in type 1 diabetes. With a growing number of reported infectious agents and islet antigens, strategies to prioritize the study of infectious agents are critically needed to expedite translational research into the etiology of type 1 diabetes. In this work, we developed an in-silico pipeline for assessing molecular mimicry in type 1 diabetes etiology based on sequence homology, empirical binding affinity to specific MHC molecules, and empirical potential for T-cell immunogenicity. We then assess whether potential molecular mimics were conserved across other pathogens known to infect humans. Overall, we identified 61 potentially high-impact molecular mimics showing sequence homology, strong empirical binding affinity, and empirical immunogenicity linked with specific MHC molecules. We further found that peptide sequences from 32 of these potential molecular mimics were conserved across several human pathogens. These findings facilitate translational evaluation of molecular mimicry in type 1 diabetes etiology by providing a curated and prioritized list of peptides from infectious agents for etiopathologic investigation. These results may also provide evidence for generation of infectious and HLA-specific preclinical models and inform future screening and preventative efforts in genetically susceptible populations.

Sequential data mining of infection patterns as predictors for onset of type 1 diabetes in genetically at-risk individuals.

Infections are implicated in the etiology of type 1 diabetes mellitus (T1DM); however, conflicting epidemiologic evidence makes designing effective strategies for presymptomatic screening and disease prevention difficult. Considering the temporality and combination in which infections occur may provide valuable insights into understanding T1DM etiology but is rarely studied due to limited longitudinal datasets and insufficient analytical techniques. The objective of this work was to demonstrate a computational approach to classify the temporality and combination of infections in presymptomatic T1DM. We present a sequential data mining pipeline that leverages routinely collected infectious disease data from a prospective cohort study, the Environmental Determinants of Diabetes in the Young (TEDDY) study, to extract, interpret, and compare infection sequences. We then utilize this pipeline to assess risk for developing presymptomatic biomarkers of islet autoimmunity and clinical onset of T1DM. Overall, we identified 229 significant sequential rules that increased the risk for developing presymptomatic biomarkers of islet autoimmunity or clinical onset of T1DM. Multiple significant sequential rules involving varicella increased the risk for all presymptomatic biomarker-specific outcomes, while a single significant sequential rule involving parasites significantly increased risk for T1DM. Significant sequential rules involving respiratory illnesses were differentially represented among the presymptomatic biomarkers of islet autoimmunity and clinical onset of T1DM. Risk for T1DM was significantly increased by a single episode of sixth disease at 12 months, representing the only single-event sequence that increased disease risk. Together, these findings provide the first insights into the timing and combination of infections in T1DM etiology, which may ultimately lead to personalized disease screening and prevention strategies. The sequential data mining pipeline developed in this work demonstrates how temporal data mining can be used to address clinically meaningful questions. This method can be adapted to other presymptomatic factors and clinical conditions.

Molecular dynamics study of the mechanical properties of drug loaded model systems: A comparison of a polymersome with a bilayer.

In this work we implement a new methodology to study structural and mechanical properties of systems having spherical and planar symmetries throughout Molecular Dynamics simulations. This methodology is applied here to a drug delivery system based in polymersomes, as an example. The chosen model drug was the local anesthetic prilocaine due to previous parameterization within the used coarse grain scheme. In our approach, mass density profiles (MDPs) are used to obtain key structural parameters of the systems, and pressure profiles are used to estimate the curvature elastic parameters. The calculation of pressure profiles and radial MPDs required the development of specific methods, which were implemented in an in-house built version of the GROMACS 2018 code. The methodology presented in this work is applied to characterize poly(ethylene oxide)-poly(butadiene) polymersomes and bilayers loaded with the model drug prilocaine. Our results show that structural properties of the polymersome membrane could be obtained from bilayer simulations, with significantly lower computational cost compared to whole polymersome simulations, but the bilayer simulations are insufficient to get insights on their mechanical aspects, since the elastic parameters are canceled out for the complete bilayer (as consequence of the symmetry). The simulations of entire polymersomes, although more complex, offer a complementary approach to get insights on the mechanical behavior of the systems.

K-means cluster analysis of cooperative effects of CO, NO2, O3, PM2.5, PM10, and SO2 on incidence of type 2 diabetes mellitus in the US.

Air pollution (AP) has been shown to increase the risk of type 2 diabetes mellitus, as well as other cardiometabolic diseases. AP is characterized by a complex mixture of components for which the composition depends on sources and metrological factors. The US Environmental Protection Agency (EPA) monitors and regulates certain components of air pollution known to have negative consequences for human health. Research assessing the health effects of these components of AP often uses traditional regression models, which might not capture more complex and interdependent relationships. Machine learning has the capability to simultaneously assess multiple components and find complex, non-linear patterns that may not be apparent and could not be modeled by other techniques. Here we use k-means clustering to assess the patterns associating PM2.5, PM10, CO, NO2, O3, and SO2 measurements and changes in annual diabetes incidence at a US county level. The average age adjusted annual decrease in diabetes incidence for the entire US populations is -0.25 per 1000 but the change shows a significant geographic variation (range: -17.2 to 5.30 per 1000). In this paper these variations were compared with the local daily AP concentrations of the pollutants listed above from 2005 to 2015, which were matched to the annual change in diabetes incidence for the following year. A total of 134,925 daily air quality observations were included in the cluster analysis, representing 125 US counties and the District of Columbia. K-means successfully clustered AP components and indicated an association between exposure to certain AP mixtures with lower decreases on T2D incidence.

A role for the MEGF6 gene in predisposition to osteoporosis.

Osteoporosis is a common skeletal disorder characterized by deterioration of bone tissue. The set of genetic factors contributing to osteoporosis is not completely specified. High-risk osteoporosis pedigrees were analyzed to identify genes that may confer susceptibility to disease. Candidate predisposition variants were identified initially by whole exome sequencing of affected-relative pairs, approximately cousins, from 10 pedigrees. Variants were filtered on the basis of population frequency, concordance between pairs of cousins, affecting a gene associated with osteoporosis, and likelihood to have functionally damaging, pathogenic consequences. Subsequently, variants were tested for segregation in 68 additional relatives of the index carriers. A rare variant in MEGF6 (rs755467862) showed strong evidence of segregation with the disease phenotype. Predicted protein folding indicated the variant (Cys200Tyr) may disrupt structure of an EGF-like calcium-binding domain of MEGF6. Functional analyses demonstrated that complete loss of the paralogous genes megf6a and megf6b in zebrafish resulted in significant delay of cartilage and bone formation. Segregation analyses, in silico protein structure modeling, and functional assays support a role for MEGF6 in predisposition to osteoporosis.

Gene co-expression network analysis for identifying modules and functionally enriched pathways in SCA2.

Spinocerebellar ataxia type 2 (SCA2) is an autosomal dominant neurodegenerative disease caused by CAG repeat expansion in the ATXN2 gene. The repeat resides in an encoded region of the gene resulting in polyglutamine (polyQ) expansion which has been assumed to result in gain of function, predominantly, for the ATXN2 protein. We evaluated temporal cerebellar expression profiles by RNA sequencing of ATXN2Q127 mice versus wild-type (WT) littermates. ATXN2Q127 mice are characterized by a progressive motor phenotype onset, and have progressive cerebellar molecular and neurophysiological (Purkinje cell firing frequency) phenotypes. Our analysis revealed previously uncharacterized early and progressive abnormal patterning of cerebellar gene expression. Weighted Gene Coexpression Network Analysis revealed four gene modules that were significantly correlated with disease status, composed primarily of genes associated with GTPase signaling, calcium signaling and cell death. Of these genes, few overlapped with differentially expressed cerebellar genes that we identified in Atxn2-/- knockout mice versus WT littermates, suggesting that loss-of-function is not a significant component of disease pathology. We conclude that SCA2 is a disease characterized by gain of function for ATXN2.

The Effects of Calcium on Lipid-Protein Interactions and Ion Flux in the Cx26 Connexon Embedded into a POPC Bilayer.

Gap junctions provide a communication pathway between adjacent cells. They are formed by paired connexons that reside in the plasma membrane of their respective cell and their activity can be modulated by the bilayer composition. In this work, we study the dynamic behavior of a Cx26 connexon embedded in a POPC lipid bilayer, studying: the membrane protein interactions and the ion flux though the connexon pore. We analyzed extensive atomistic molecular dynamics simulations for different conditions, with and without calcium ions. We found that lipid-protein interactions were mainly mediated by hydrogen bonds. Specific amino acids were identified forming hydrogen bonds with the POPC lipids (ARG98, ARG127, ARG165, ARG216, LYS22, LYS221, LYS223, LYS224, SER19, SER131, SER162, SER219, SER222, THR18 and TYR97, TYR155, TYR212, and TYR217). In the presence of calcium ions, we found subtle differences on the HB lifetimes. Finally, these MD simulations are able to identify and explain differential chlorine flux through the pore depending on the presence or absence of the calcium ions and its distribution within the pore.

Comprehensive methodology to monitor longitudinal change patterns during EHR implementations: a case study at a large health care delivery network.

OBJECTIVE: To test a systematic methodology to monitor longitudinal change patterns on quality, productivity, and safety outcomes during a large-scale commercial Electronic Health Record (EHR) implementation. MATERIALS AND METHODS: Our method combines an interrupted time-series design with control sites and 41 consensus outcomes including quality (11 measures), productivity (20 measures), and safety (10 measures). The intervention consisted of a phased commercial EHR implementation at a large health care delivery network. Four medium-size hospitals and 39 clinics from 5 geographic regions implementing the new EHR were compared against a parallel control consisting of one medium-size and one large hospital and 10 clinics that had not implemented the new EHR at the time of this study. We collected monthly data from February 2013 to July 2017. RESULTS: The proposed methodology was successfully implemented and significant changes were observed in most measured variables. A significant change attributable to the intervention was observed in 12 (29%) measures in three or more regions; in 32 (78%) measures in two or more regions; and in 40 (98%) measures in at least one region. A similar pattern (i.e., same impact in three or more regions) was detected for nine (22%) measures, a mixed pattern (i.e., same impact in two regions, and different impact in other regions) was detected for nine (22%) measures, and an inconsistent pattern (i.e., did not detect the same impact across regions) was detected for 23 (56%) measures. DISCUSSION: Using a formal methodology to assess changes in a set of consensus measures, we detected various patterns of impact and mixed time-sensitive effects. With an increasing adoption of EHR systems, it is critical for health care organizations to systematically monitor their EHR implementations. The proposed method provides a robust and consistent approach to monitor EHR implementations longitudinally allowing for continuous monitoring after the system becomes stable in order to avoid unexpected effects. CONCLUSION: Our results and methodology can guide the broader medical and informatics communities by informing what and how to continuously monitor EHR impact on quality, productivity, and safety.

Mechanical properties of drug loaded diblock copolymer bilayers: A molecular dynamics study.

In this work, we present results of coarse-grained simulations to study the encapsulation of prilocaine (PLC), both neutral and protonated, on copolymer bilayers through molecular dynamics simulations. Using a previously validated membrane model, we have simulated loaded bilayers at different drug concentrations and at low (protonated PLC) and high (neutral PLC) pH levels. We have characterized key structural parameters of the loaded bilayers in order to understand the effects of encapsulation of PLC on the bilayer structure and mechanical properties. Neutral PLC was encapsulated in the hydrophobic region leading to a thickness increase, while the protonated species partitioned between the water phase and the poly(ethylene oxide)-poly(butadiene) (PBD) interface, relaxing the PBD region and leading to a decrease in the thickness. The tangential pressures of the studied systems were calculated, and their components were decomposed in order to gain insights on their compensation. In all cases, it is observed that the loading of the membrane does not significantly decrease the stability of the bilayer, indicating that the system could be used for drug delivery.

Development and classification of a robust inventory of near real-time outcome measurements for assessing information technology interventions in health care.

OBJECTIVE: To develop and classify an inventory of near real-time outcome measures for assessing information technology (IT) interventions in health care and assess their relevance as perceived by experts in the field. MATERIALS AND METHODS: To verify the robustness and coverage of a previously published inventory of measures and taxonomy, we conducted semi-structured interviews with clinical and administrative leaders from a large care delivery system to collect suggestions of outcome measures that can be calculated with data available in electronic format for near real-time monitoring of EHR implementations. We combined these measures with the most commonly reported in the literature. We then conducted two online surveys with subject-matter experts to collect their perceptions of the relevance of the measures, and identify other potentially relevant measures. RESULTS: With input from experienced health care leaders and informaticists, we developed an inventory of 102 outcome measures. These measures were classified into a taxonomy of commonly used measures around the categories of quality, productivity, and safety. Safety measures were rated as most relevant by subject-matter experts, especially those measuring medication processes. Clinician satisfaction and measures assessing mean time to complete tasks and time spent on electronic documentation were also rated as highly relevant. DISCUSSION: By expanding the coverage of our previously published inventory and taxonomy, we expect to help providers, health IT vendors and researchers to more effectively and consistently monitor the impact of EHR implementations in near real-time, and report more standardized outcomes in future studies. We identified several measures not commonly assessed by previous studies of IT implementations, especially those of safety and productivity, which deserve more attention from the broader informatics community. CONCLUSION: Our inventory of measures and taxonomy will help researchers identify gaps in their measurement approaches and report more standardized measurements of IT interventions that could be shared among researchers, hopefully facilitating comparison across future studies and increasing our understanding of the impact of IT interventions in health care.

Diblock copolymer bilayers as model for polymersomes: A coarse grain approach.

This paper presents a new model for polymersomes developed using a poly(ethylene oxide)-poly(butadiene) diblock copolymer bilayer. The model is based on a coarse-grained approach using the MARTINI force field. Since no MARTINI parameters exist for poly(butadiene), we have refined these parameters using quantum mechanical calculations and molecular dynamics simulations. The model has been validated using extensive molecular dynamics simulations in systems with several hundred polymer units and reaching up to 6 µs. These simulations show that the copolymer coarse grain model self-assemble into bilayers and that NPT and NPNγT ensemble runs reproduce key structural and mechanical experimental properties for different copolymer length chains with a similar hydrophilic weight fraction.

Health information technology adoption: Understanding research protocols and outcome measurements for IT interventions in health care.

OBJECTIVE: To classify and characterize the variables commonly used to measure the impact of Information Technology (IT) adoption in health care, as well as settings and IT interventions tested, and to guide future research. MATERIALS AND METHODS: We conducted a descriptive study screening a sample of 236 studies from a previous systematic review to identify outcome measures used and the availability of data to calculate these measures. We also developed a taxonomy of commonly used measures and explored setting characteristics and IT interventions. RESULTS: Clinical decision support is the most common intervention tested, primarily in non-hospital-based clinics and large academic hospitals. We identified 15 taxa representing the 79 most commonly used measures. Quality of care was the most common category of these measurements with 62 instances, followed by productivity (11 instances) and patient safety (6 instances). Measures used varied according to type of setting, IT intervention and targeted population. DISCUSSION: This study provides an inventory and a taxonomy of commonly used measures that will help researchers select measures in future studies as well as identify gaps in their measurement approaches. The classification of the other protocol components such as settings and interventions will also help researchers identify underexplored areas of research on the impact of IT interventions in health care. CONCLUSION: A more robust and standardized measurement system and more detailed descriptions of interventions and settings are necessary to enable comparison between studies and a better understanding of the impact of IT adoption in health care settings.

Crystal Structure Prediction from First Principles: The Crystal Structures of Glycine.

Here we present the results of our unbiased searches of glycine polymorphs obtained using the Genetic Algorithms search implemented in Modified Genetic Algorithm for Crystals coupled with the local optimization and energy evaluation provided by Quantum Espresso. We demonstrate that it is possible to predict the crystal structures of a biomedical molecule using solely first principles calculations. We were able to find all the ambient pressure stable glycine polymorphs, which are found in the same energetic ordering as observed experimentally and the agreement between the experimental and predicted structures is of such accuracy that the two are visually almost indistinguishable.

Structure prediction of polyglutamine disease proteins: comparison of methods.

BACKGROUND: The expansion of polyglutamine (poly-Q) repeats in several unrelated proteins is associated with at least ten neurodegenerative diseases. The length of the poly-Q regions plays an important role in the progression of the diseases. The number of glutamines (Q) is inversely related to the onset age of these polyglutamine diseases, and the expansion of poly-Q repeats has been associated with protein misfolding. However, very little is known about the structural changes induced by the expansion of the repeats. Computational methods can provide an alternative to determine the structure of these poly-Q proteins, but it is important to evaluate their performance before large scale prediction work is done. RESULTS: In this paper, two popular protein structure prediction programs, I-TASSER and Rosetta, have been used to predict the structure of the N-terminal fragment of a protein associated with Huntington's disease with 17 glutamines. Results show that both programs have the ability to find the native structures, but I-TASSER performs better for the overall task. CONCLUSIONS: Both I-TASSER and Rosetta can be used for structure prediction of proteins with poly-Q repeats. Knowledge of poly-Q structure may significantly contribute to development of therapeutic strategies for poly-Q diseases.

iBIOMES Lite: summarizing biomolecular simulation data in limited settings.

As the amount of data generated by biomolecular simulations dramatically increases, new tools need to be developed to help manage this data at the individual investigator or small research group level. In this paper, we introduce iBIOMES Lite, a lightweight tool for biomolecular simulation data indexing and summarization. The main goal of iBIOMES Lite is to provide a simple interface to summarize computational experiments in a setting where the user might have limited privileges and limited access to IT resources. A command-line interface allows the user to summarize, publish, and search local simulation data sets. Published data sets are accessible via static hypertext markup language (HTML) pages that summarize the simulation protocols and also display data analysis graphically. The publication process is customized via extensible markup language (XML) descriptors while the HTML summary template is customized through extensible stylesheet language (XSL). iBIOMES Lite was tested on different platforms and at several national computing centers using various data sets generated through classical and quantum molecular dynamics, quantum chemistry, and QM/MM. The associated parsers currently support AMBER, GROMACS, Gaussian, and NWChem data set publication. The code is available at https://github.com/jcvthibault/ibiomes .

A domain analysis model for eIRB systems: addressing the weak link in clinical research informatics.

Institutional Review Boards (IRBs) are a critical component of clinical research and can become a significant bottleneck due to the dramatic increase, in both volume and complexity of clinical research. Despite the interest in developing clinical research informatics (CRI) systems and supporting data standards to increase clinical research efficiency and interoperability, informatics research in the IRB domain has not attracted much attention in the scientific community. The lack of standardized and structured application forms across different IRBs causes inefficient and inconsistent proposal reviews and cumbersome workflows. These issues are even more prominent in multi-institutional clinical research that is rapidly becoming the norm. This paper proposes and evaluates a domain analysis model for electronic IRB (eIRB) systems, paving the way for streamlined clinical research workflow via integration with other CRI systems and improved IRB application throughput via computer-assisted decision support.

A Scalable and Extensible Logical Data Model of Electronic Health Record Audit Logs for Temporal Data Mining (RNteract): Model Conceptualization and Formulation.

BACKGROUND: Increased workload, including workload related to electronic health record (EHR) documentation, is reported as a main contributor to nurse burnout and adversely affects patient safety and nurse satisfaction. Traditional methods for workload analysis are either administrative measures (such as the nurse-patient ratio) that do not represent actual nursing care or are subjective and limited to snapshots of care (eg, time-motion studies). Observing care and testing workflow changes in real time can be obstructive to clinical care. An examination of EHR interactions using EHR audit logs could provide a scalable, unobtrusive way to quantify the nursing workload, at least to the extent that nursing work is represented in EHR documentation. EHR audit logs are extremely complex; however, simple analytical methods cannot discover complex temporal patterns, requiring use of state-of-the-art temporal data-mining approaches. To effectively use these approaches, it is necessary to structure the raw audit logs into a consistent and scalable logical data model that can be consumed by machine learning (ML) algorithms. OBJECTIVE: We aimed to conceptualize a logical data model for nurse-EHR interactions that would support the future development of temporal ML models based on EHR audit log data. METHODS: We conducted a preliminary review of EHR audit logs to understand the types of nursing-specific data captured. Using concepts derived from the literature and our previous experience studying temporal patterns in biomedical data, we formulated a logical data model that can describe nurse-EHR interactions, the nurse-intrinsic and situational characteristics that may influence those interactions, and outcomes of relevance to the nursing workload in a scalable and extensible manner. RESULTS: We describe the data structure and concepts from EHR audit log data associated with nursing workload as a logical data model named RNteract. We conceptually demonstrate how using this logical data model could support temporal unsupervised ML and state-of-the-art artificial intelligence (AI) methods for predictive modeling. CONCLUSIONS: The RNteract logical data model appears capable of supporting a variety of AI-based systems and should be generalizable to any type of EHR system or health care setting. Quantitatively identifying and analyzing temporal patterns of nurse-EHR interactions is foundational for developing interventions that support the nursing documentation workload and address nurse burnout.

Uncertainty in Breast Cancer Risk Prediction: A Conformal Prediction Study of Race Stratification.

The use of Artificial Intelligence (AI) in medicine has attracted a great deal of attention in the medical literature, but less is known about how to assess the uncertainty of individual predictions in clinical applications. This paper demonstrates the use of Conformal Prediction (CP) to provide insight on racial stratification of uncertainty quantification for breast cancer risk prediction. The results presented here show that CP methods provide important information about the diminished quality of predictions for individuals of minority racial backgrounds.