Search

Domain generalization is a ubiquitous challenge for machine learning in healthcare. Model performance in real-world conditions might be lower than expected because of discrepancies between the data encountered during deployment and development. Underrepresentation of some groups or conditions during model development is a common cause of this phenomenon. This challenge is often not readily addressed by targeted data acquisition and 'labeling' by expert clinicians, which can be prohibitively expensive or practically impossible because of the rarity of conditions or the available clinical expertise. We hypothesize that advances in generative artificial intelligence can help mitigate this unmet need in a steerable fashion, enriching our training dataset with synthetic examples that address shortfalls of underrepresented conditions or subgroups. We show that diffusion models can automatically learn realistic augmentations from data in a label-efficient manner. We demonstrate that learned augmentations make models more robust and statistically fair in-distribution and out of distribution. To evaluate the generality of our approach, we studied three distinct medical imaging contexts of varying difficulty: (1) histopathology, (2) chest X-ray and (3) dermatology images. Complementing real samples with synthetic ones improved the robustness of models in all three medical tasks and increased fairness by improving the accuracy of clinical diagnosis within underrepresented groups, especially out of distribution., (© 2024. The Author(s).)

In treating depression and anxiety, just over half of all clients respond. Monitoring and obtaining early client feedback can allow for rapidly adapted treatment delivery and improve outcomes. This study seeks to develop a state-of-the-art deep-learning framework for predicting clinical outcomes in internet-delivered Cognitive Behavioural Therapy (iCBT) by leveraging large-scale, high-dimensional time-series data of client-reported mental health symptoms and platform interaction data. We use de-identified data from 45,876 clients on SilverCloud Health, a digital platform for the psychological treatment of depression and anxiety. We train deep recurrent neural network (RNN) models to predict whether a client will show reliable improvement by the end of treatment using clinical measures, interaction data with the iCBT program, or both. Outcomes are based on total improvement in symptoms of depression (Patient Health Questionnaire-9, PHQ-9) and anxiety (Generalized Anxiety Disorder-7, GAD-7), as reported within the iCBT program. Using internal and external datasets, we compare the proposed models against several benchmarks and rigorously evaluate them according to their predictive accuracy, sensitivity, specificity and AUROC over treatment. Our proposed RNN models consistently predict reliable improvement in PHQ-9 and GAD-7, using past clinical measures alone, with above 87% accuracy and 0.89 AUROC after three or more review periods, outperforming all benchmark models. Additional evaluations demonstrate the robustness of the achieved models across (i) different health services; (ii) geographic locations; (iii) iCBT programs, and (iv) client severity subgroups. Results demonstrate the robust performance of dynamic prediction models that can yield clinically helpful prognostic information ready for implementation within iCBT systems to support timely decision-making and treatment adjustments by iCBT clinical supporters towards improved client outcomes., Competing Interests: AE, JP, DK, and DR are employees of SilverCloud Health. GD is a cofounder of SilverCloud Health and has a minority shareholding in the company. This does not alter our adherence to PLOS ONE policies on sharing data and materials., (Copyright: © 2023 Prasad et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.)

For many years, psychiatrists have tried to understand factors involved in response to medications or psychotherapies, in order to personalize their treatment choices. There is now a broad and growing interest in the idea that we can develop models to personalize treatment decisions using new statistical approaches from the field of machine learning and applying them to larger volumes of data. In this pursuit, there has been a paradigm shift away from experimental studies to confirm or refute specific hypotheses towards a focus on the overall explanatory power of a predictive model when tested on new, unseen datasets. In this paper, we review key studies using machine learning to predict treatment outcomes in psychiatry, ranging from medications and psychotherapies to digital interventions and neurobiological treatments. Next, we focus on some new sources of data that are being used for the development of predictive models based on machine learning, such as electronic health records, smartphone and social media data, and on the potential utility of data from genetics, electrophysiology, neuroimaging and cognitive testing. Finally, we discuss how far the field has come towards implementing prediction tools in real-world clinical practice. Relatively few retrospective studies to-date include appropriate external validation procedures, and there are even fewer prospective studies testing the clinical feasibility and effectiveness of predictive models. Applications of machine learning in psychiatry face some of the same ethical challenges posed by these techniques in other areas of medicine or computer science, which we discuss here. In short, machine learning is a nascent but important approach to improve the effectiveness of mental health care, and several prospective clinical studies suggest that it may be working already., (© 2021 World Psychiatric Association.)

Importance: The mechanisms by which engagement with internet-delivered psychological interventions are associated with depression and anxiety symptoms are unclear., Objective: To identify behavior types based on how people engage with an internet-based cognitive behavioral therapy (iCBT) intervention for symptoms of depression and anxiety., Design, Setting, and Participants: Deidentified data on 54 604 adult patients assigned to the Space From Depression and Anxiety treatment program from January 31, 2015, to March 31, 2019, were obtained for probabilistic latent variable modeling using machine learning techniques to infer distinct patient subtypes, based on longitudinal heterogeneity of engagement patterns with iCBT., Interventions: A clinician-supported iCBT-based program that follows clinical guidelines for treating depression and anxiety, delivered on a web 2.0 platform., Main Outcomes and Measures: Log data from user interactions with the iCBT program to inform engagement patterns over time. Clinical outcomes included symptoms of depression (Patient Health Questionnaire-9 [PHQ-9]) and anxiety (Generalized Anxiety Disorder-7 [GAD-7]); PHQ-9 cut point greater than or equal to 10 and GAD-7 scores greater than or equal to 8 were used to define depression and anxiety., Results: Patients spent a mean (SD) of 111.33 (118.92) minutes on the platform and completed 230.60 (241.21) tools. At baseline, mean PHQ-9 score was 12.96 (5.81) and GAD-7 score was 11.85 (5.14). Five subtypes of engagement were identified based on patient interaction with different program sections over 14 weeks: class 1 (low engagers, 19 930 [36.5%]), class 2 (late engagers, 11 674 [21.4%]), class 3 (high engagers with rapid disengagement, 13 936 [25.5%]), class 4 (high engagers with moderate decrease, 3258 [6.0%]), and class 5 (highest engagers, 5799 [10.6%]). Estimated mean decrease (SE) in PHQ-9 score was 6.65 (0.14) for class 3, 5.88 (0.14) for class 5, and 5.39 (0.14) for class 4; class 2 had the lowest rate of decrease at -4.41 (0.13). Compared with PHQ-9 score decrease in class 1, the Cohen d effect size (SE) was -0.46 (0.014) for class 2, -0.46 (0.014) for class 3, -0.61 (0.021) for class 4, and -0.73 (0.018) for class 5. Similar patterns were found across groups for GAD-7., Conclusions and Relevance: The findings of this study may facilitate tailoring interventions according to specific subtypes of engagement for individuals with depression and anxiety. Informing clinical decision needs of supporters may be a route to successful adoption of machine learning insights, thus improving clinical outcomes overall.

Background: We hypothesized that filaggrin (FLG) loss-of-function mutations modify the effect of allergen exposure on the development of allergic sensitization., Objective: We sought to determine whether early-life exposure to inhalant allergens increases the risk of specific sensitization and whether FLG mutations modulate these odds., Methods: In a population-based birth cohort we measured mite, cat, and dog allergen levels in dust samples collected from homes within the first year of life. Sensitization was assessed at 6 time points between infancy and age 16 years. Genotyping was performed for 6 FLG mutations., Results: In the longitudinal multivariable model (age 1-16 years), we observed a significant interaction between FLG and Fel d 1 exposure on cat sensitization, with the effect of exposure being significantly greater among children with FLG mutations compared with those without (odds ratio, 1.36; 95% CI, 1.02-1.80; P = .035). The increase in risk of mite sensitization with increasing Der p 1 exposure was consistently greater among children with FLG mutations, but the interaction did not reach statistical significance. Different associations were observed for dogs: there was a significant interaction between FLG and dog ownership, but the risk of sensitization to any allergen was significantly lower among children with FLG mutations who were exposed to a dog in infancy (odds ratio, 0.16; 95% CI, 0.03-0.86; P = .03)., Conclusions: FLG loss-of-function mutations modify the relationship between allergen exposure and sensitization, but effects differ at different ages and between different allergens., (Copyright © 2019 The Authors. Published by Elsevier Inc. All rights reserved.)

Background: Allergic diseases (eczema, wheeze and rhinitis) in children often present as heterogeneous phenotypes. Understanding genetic associations of specific patterns of symptoms might facilitate understanding of the underlying biological mechanisms., Objective: To examine associations between allergic disease-related variants identified in a recent genome-wide association study and latent classes of allergic diseases (LCADs) in two population-based birth cohorts., Methods: Eight previously defined LCADs between birth and 11 years: "No disease," "Atopic march," "Persistent eczema and wheeze," "Persistent eczema with later-onset rhinitis," "Persistent wheeze with later-onset rhinitis," "Transient wheeze," "Eczema only" and "Rhinitis only" were used as the study outcome. Weighted multinomial logistic regression was used to estimate associations between 135 SNPs (and a polygenic risk score, PRS) and LCADs among 6345 individuals from The Avon Longitudinal Study of Parents and Children (ALSPAC). Heterogeneity across LCADs was assessed before and after Bonferroni correction. Results were replicated in Manchester Asthma and Allergy Study (MAAS) (n = 896) and pooled in a meta-analysis., Results: We found strong evidence for differential genetic associations across the LCADs; pooled PRS heterogeneity P-value = 3.3 × 10 -14 , excluding "no disease" class. The associations between the PRS and LCADs in MAAS were remarkably similar to ALSPAC. Two SNPs (a protein-truncating variant in FLG and a SNP within an intron of GSDMB) had evidence for differential association (pooled P-values ≤ 0.006). The FLG locus was differentially associated across LCADs that included eczema, with stronger associations for LCADs with comorbid wheeze and rhinitis. The GSDMB locus in contrast was equally associated across LCADs that included wheeze., Conclusions and Clinical Relevance: We have shown complex, but distinct patterns of genetic associations with LCADs, suggesting that heterogeneous mechanisms underlie individual disease trajectories. Establishing the combination of allergic diseases with which each genetic variant is associated may inform therapeutic development and/or predictive modelling., (© 2019 The Authors. Clinical & Experimental Allergy published by John Wiley & Sons Ltd.)

Background: A critical window in infancy has been proposed, during which the microbiota may affect subsequent health. The longitudinal development of the oropharyngeal microbiota is under-studied and may be associated with early-life wheeze. We aimed to investigate the temporal association of the development of the oropharyngeal microbiota with early-life wheeze., Methods: A population-based birth cohort based in London, UK was followed for 24 months. We collected oropharyngeal swabs at six time-points. Microbiota was determined using sequencing of the V3-V5 region of the 16S rRNA-encoding gene. Medical records were reviewed for the outcome of doctor diagnosed wheeze. We used a time-varying model to investigate the temporal association between the development of microbiota and doctor-diagnosed wheeze., Findings: 159 participants completed the study to 24 months and for 98 there was complete sequencing data at all timepoints and outcome data. Of these, 26 had doctor-diagnosed wheeze. We observed significant increase in the abundance of Neisseria between 9 and 24 months in children who developed wheeze (p = 0∙003), while in those without wheezing there was a significant increment in the abundance of Granulicatella (p = 0∙012) between 9 and 12 months, and of Prevotella (p = 0∙018) after 18 months., Interpretation: A temporal association between the respiratory commensal Granulicatella and also Prevotella with wheeze (negative), and between Neisseria and wheeze (positive) was identified in infants prior to one year of age. This adds to evidence for the proposed role of the microbiota in the development of wheeze. FUND: Research funding from the Winnicott Foundation, Meningitis Now and Micropathology Ltd., (Copyright © 2019. Published by Elsevier B.V.)

Background: Current published asthma predictive tools have moderate positive likelihood ratios (+LR) but high negative likelihood ratios (-LR) based on their recommended cut-offs, which limit their clinical usefulness., Objective: To develop a simple clinically applicable asthma prediction tool within a population-based birth cohort., Method: Children from the Manchester Asthma and Allergy Study (MAAS) attended follow-up at ages 3, 8 and 11 years. Data on preschool wheeze were extracted from primary-care records. Parents completed validated respiratory questionnaires. Children were skin prick tested (SPT). Asthma at 8/11 years (school-age) was defined as parentally reported (a) physician-diagnosed asthma and wheeze in the previous 12 months or (b) ≥3 wheeze attacks in the previous 12 months. An asthma prediction tool (MAAS APT) was developed using logistic regression of characteristics at age 3 years to predict school-age asthma., Results: Of 336 children with physician-confirmed wheeze by age 3 years, 117(35%) had school-age asthma. Logistic regression selected 5 significant risk factors which formed the basis of the MAAS APT: wheeze after exercise; wheeze causing breathlessness; cough on exertion; current eczema and SPT sensitisation(maximum score 5). A total of 281(84%) children had complete data at age 3 years and were used to test the MAAS APT. Children scoring ≥3 were at high risk of having asthma at school-age (PPV > 75%; +LR 6.3, -LR 0.6), whereas children who had a score of 0 had very low risk(PPV 9.3%; LR 0.2)., Conclusion: MAAS APT is a simple asthma prediction tool which could easily be applied in clinical and research settings., (© 2018 The Authors. Clinical & Experimental Allergy Published by John Wiley & Sons Ltd.)

Background: There is a paucity of information about longitudinal patterns of IgE responses to allergenic proteins (components) from multiple sources., Objectives: This study sought to investigate temporal patterns of component-specific IgE responses from infancy to adolescence, and their relationship with allergic diseases., Methods: In a population-based birth cohort, we measured IgE to 112 components at 6 follow-ups during childhood. We used a Bayesian method to discover cross-sectional sensitization patterns and their longitudinal trajectories, and we related these patterns to asthma and rhinitis in adolescence., Results: We identified 1 sensitization cluster at age 1, 3 at age 3, 4 at ages 5 and 8, 5 at age 11, and 6 at age 16 years. "Broad" cluster was the only cluster present at every follow-up, comprising components from multiple sources. "Dust mite" cluster formed at age 3 years and remained unchanged to adolescence. At age 3 years, a single-component "Grass" cluster emerged, which at age 5 years absorbed additional grass components and Fel d 1 to form the "Grass/cat" cluster. Two new clusters formed at age 11 years: "Cat" cluster and "PR-10/profilin" (which divided at age 16 years into "PR-10" and "Profilin"). The strongest contemporaneous associate of asthma at age 16 years was sensitization to dust mite cluster (odds ratio: 2.6; 95% CI: 1.2-6.1; P < .05), but the strongest early life predictor of subsequent asthma was sensitization to grass/cat cluster (odds ratio: 3.5; 95% CI: 1.6-7.4; P < .01)., Conclusions: We describe the architecture of the evolution of IgE responses to multiple allergen components throughout childhood, which may facilitate development of better diagnostic and prognostic biomarkers for allergic diseases., (Copyright © 2018 The Authors. Published by Elsevier Inc. All rights reserved.)

Rationale: Immunophenotypes of antiviral responses, and their relationship with asthma, allergy, and lower respiratory tract infections, are poorly understood., Objectives: We characterized multiple cytokine responses of peripheral blood mononuclear cells to rhinovirus stimulation, and their relationship with clinical outcomes., Methods: In a population-based birth cohort, we measured 28 cytokines after stimulation with rhinovirus-16 in 307 children aged 11 years. We used machine learning to identify patterns of cytokine responses, and related these patterns to clinical outcomes, using longitudinal models. We also ascertained phytohemagglutinin-induced T-helper cell type 2 (Th2)-cytokine responses (PHA-Th2)., Measurements and Main Results: We identified six clusters of children based on their rhinovirus-16 responses, which were differentiated by the expression of four cytokine/chemokine groups: interferon-related (IFN), proinflammatory (Inflam), Th2-chemokine (Th2-chem), and regulatory (Reg). Clusters differed in their clinical characteristics. Children with an IFN mod Inflam highest Th2-chem highest Reg highest rhinovirus-16-induced pattern had a PHA-Th2 low response, and a very low asthma risk (odds ratio [OR], 0.08; 95% confidence interval [CI], 0.01-0.81; P = 0.03). Two clusters had a high risk of asthma and allergic sensitization, but with different trajectories from infancy to adolescence. The IFN lowest Inflam high Th2-chem low Reg mod cluster exhibited a PHA-Th2 lowest response and was associated with early-onset asthma and sensitization, and the highest risk of asthma exacerbations (OR, 1.37; 95% CI, 1.07-1.76; P = 0.014) and lower respiratory tract infection hospitalizations (OR, 2.40; 95% CI, 1.26-4.58; P = 0.008) throughout childhood. In contrast, the IFN highest Inflam mod Th2-chem mod Reg high cluster with a rhinovirus-16-cytokine pattern was characterized by a PHA-Th2 highest response, and a low prevalence of asthma/sensitization in infancy that increased sharply to become the highest among all clusters by adolescence (but with a low risk of asthma exacerbations)., Conclusions: Early-onset troublesome asthma with early-life sensitization, later-onset milder allergic asthma, and disease protection are each associated with different patterns of rhinovirus-induced immune responses.

Background: Sensitization in early childhood may precede respiratory allergy in adolescence., Methods: IgE reactivity against 132 allergen molecules was evaluated using the MeDALL microarray in sera obtained from a random sample of 786 children at the age of 4, 8 and 16years in a population based birth cohort (BAMSE). Symptoms were analyzed by questionnaire at ages 4, 8 and 16years. Clinically and independent relevant allergen molecules accounting for ≥90% of IgE reactivities in sensitized individuals and at all time-points were identified as risk molecules and used to predict respiratory allergy. The data was replicated in the Manchester Asthma and Allergy Study (MAAS) birth cohort by studying IgE reactivity with the use of a commercial IgE microarray. Sera were obtained from children at the ages of 3, 5, 8 and 11years (N=248) and the outcome was studied at 11years., Findings: In the BAMSE cohort 4 risk molecules could be identified, i.e.: Ara h 1 (peanut), Bet v 1 (birch), Fel d 1 (cat), Phl p 1 (grass). For MAAS the corresponding number of molecules was 5: Der p 1 (dust mite), Der f 2 (dust mite), Phl p 1 (grass), Phl p 5 (grass), Fel d 1 (cat). In BAMSE, early IgE reactivity to ≥3 of 4 allergen molecules at four years predicted incident and persistent asthma and/or rhinitis at 16years (87% and 95%, respectively). The corresponding proportions in the MAAS cohort at 16years were 100% and 100%, respectively, for IgE reactivity to ≥3 of 5 risk molecules., Interpretations: IgE reactivity to a few allergen molecules early in life identifies children with a high risk of asthma and/or rhinitis at 16years. These findings will be of importance for developing preventive strategies for asthma and rhinitis in children., (Crown Copyright © 2017. Published by Elsevier B.V. All rights reserved.)

Introduction: Asthma is no longer thought of as a single disease, but rather a collection of varying symptoms expressing different disease patterns. One of the ongoing challenges is understanding the underlying pathophysiological mechanisms that may be responsible for the varying responses to treatment. Areas Covered: This review provides an overview of our current understanding of the asthma phenotype concept in childhood and describes key findings from both conventional and data-driven methods. Expert Commentary: With the vast amounts of data generated from cohorts, there is hope that we can elucidate distinct pathophysiological mechanisms, or endotypes. In return, this would lead to better patient stratification and disease management, thereby providing true personalised medicine.

We are facing a major challenge in bridging the gap between identifying subtypes of asthma to understand causal mechanisms and translating this knowledge into personalized prevention and management strategies. In recent years, "big data" has been sold as a panacea for generating hypotheses and driving new frontiers of health care; the idea that the data must and will speak for themselves is fast becoming a new dogma. One of the dangers of ready accessibility of health care data and computational tools for data analysis is that the process of data mining can become uncoupled from the scientific process of clinical interpretation, understanding the provenance of the data, and external validation. Although advances in computational methods can be valuable for using unexpected structure in data to generate hypotheses, there remains a need for testing hypotheses and interpreting results with scientific rigor. We argue for combining data- and hypothesis-driven methods in a careful synergy, and the importance of carefully characterized birth and patient cohorts with genetic, phenotypic, biological, and molecular data in this process cannot be overemphasized. The main challenge on the road ahead is to harness bigger health care data in ways that produce meaningful clinical interpretation and to translate this into better diagnoses and properly personalized prevention and treatment plans. There is a pressing need for cross-disciplinary research with an integrative approach to data science, whereby basic scientists, clinicians, data analysts, and epidemiologists work together to understand the heterogeneity of asthma., (Copyright © 2016 The Authors. Published by Elsevier Inc. All rights reserved.)

Background: Skin prick tests (SPTs) and allergen-specific serum IgE (sIgE) measurements are the main diagnostic tools for confirming atopic sensitization. Results are usually reported as 'positive' or 'negative', using the same arbitrary cut-offs (SPT>3 mm, sIgE>0.35 kUA /l) across different ages and sexes. We investigated the influence of age and sex on the interpretation of allergy test in the context of childhood asthma., Methods: In a population-based birth cohort (n = 1051), we ascertained the information on asthma/wheeze (validated questionnaires) and performed SPTs and sIgE measurement to inhalant allergens (dust mite, cat, dog) at follow-ups between ages three and 11 years. We investigated the association between quantitative sensitization (sum of SPT mean wheal diameters [MWD] and sIgE titres to the three allergens) and current wheeze and asthma across ages and sexes., Results: We observed a significant association between the SPT MWD and sIgE titres and wheeze/asthma at most ages and for both sexes. However, the strength of this association was age- and sex-dependent. For SPTs, the strength of the association between MWD and asthma increased with increasing age; we observed the opposite pattern for sIgE titre. For any given SPT MWD/sIgE titre, boys were significantly more likely to express clinical symptoms, particularly in early life; this difference between males and females diminished with age and was no longer significant by age 11 years., Conclusions: Age and sex should be taken into account when interpreting the results of skin tests and sIgE measurement, and age- and sex-specific normative data are needed for these allergy tests., (© 2016 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd.)

Background: Although most children with asthma and rhinitis are sensitized to aeroallergens, only a minority of sensitized children are symptomatic, implying the underlying operation of efficient anti-inflammatory control mechanisms., Objective: We sought to identify endogenous control mechanisms that attenuate expression of IgE-associated responsiveness to aeroallergens in sensitized children., Methods: In 3 independent population samples we analyzed relationships between aeroallergen-specific IgE and corresponding allergen-specific IgG (sIgG) and associated immunophenotypes in atopic children and susceptibility to asthma and rhinitis, focusing on responses to house dust mite and grass., Results: Among mite-sensitized children across all populations and at different ages, house dust mite-specific IgG/IgE ratios (but not IgG4/IgE ratios) were significantly lower in children with asthma compared with ratios in those without asthma and lowest among the most severely symptomatic. This finding was mirrored by relationships between rhinitis and antibody responses to grass. Depending on age/allergen specificity, 20% to 40% of children with allergen-specific IgE (sIgE) of 0.35 kU/L or greater had negative skin test responses, and these children also expressed the high sIgG/sIgE immunophenotype. sIgG1 from these children inhibited allergen-induced IgE-dependent basophil activation in a dose-dependent fashion. Profiling of aeroallergen-specific CD4(+) TH memory responses revealed positive associations between sIgG/sIgE ratios and IL-10-dependent gene signatures and significantly higher IL-10/TH2 cytokine (protein) ratios among nonsymptomatic children., Conclusion: In addition to its role in blocking TH2 effector activation in the late-phase allergic response, IL-10 is a known IgG1 switch factor. We posit that its production during allergen-induced memory responses contributes significantly to attenuation of inflammation through promoting IgG1-mediated damping of the FcεRI-dependent acute-phase reaction. sIgG1/sIgE balance might represent a readily accessible therapeutic target for asthma/rhinitis control., (Copyright © 2015 American Academy of Allergy, Asthma & Immunology. Published by Elsevier Inc. All rights reserved.)