Evaluation of measurement errors in the Patient-Oriented Eczema Measure (POEM) outcome.

BACKGROUND: The Patient-Oriented Eczema Measure (POEM) is the recommended core outcome instrument for atopic dermatitis (AD) symptoms. POEM is reported by recalling the presence/absence of seven symptoms in the last 7 days. OBJECTIVE: To evaluate measurement errors in POEM recordings due to imperfect recall. METHODS: Using data from a clinical trial of 247 AD patients aged 12-65 years, we analysed the reported POEM score (r-POEM) and the POEM derived from the corresponding daily scores for the same seven symptoms without weekly recall (d-POEM). We quantified recall error by comparing the r-POEM and d-POEM for 777 patient-weeks collected from 207 patients, and estimated two components of recall error: (1) recall bias due to systematic errors in measurements and (2) recall noise due to random errors in measurements, using a bespoke statistical model. RESULTS: POEM scores have a relatively low recall bias, but a high recall noise. Recall bias was estimated at 1.2 points lower for the r-POEM on average than the d-POEM, with a recall noise of 5.7 points. For example, a patient with a recall-free POEM of 11 (moderate) could report their POEM score anywhere from 5 to 14 (with 95% probability) because of recall error. Model estimates suggested that patients tend to recall itch and dryness more often than experienced (positive bias of less than 1 day), but less often for the other symptoms (bleeding, cracking, flaking, oozing/weeping and sleep disturbance; negative bias ranging 1-4 days). CONCLUSIONS: In this clinical trial data set, we found that patients tended to slightly underestimate their symptoms when reporting POEM, with significant variation in how well they were able to recall the frequency of their symptoms every time they reported POEM. A large recall noise should be taken into consideration when interpreting POEM scores.

Evolution of Eczema, Wheeze, and Rhinitis from Infancy to Early Adulthood: Four Birth Cohort Studies.

Rationale: The relationship between eczema, wheeze or asthma, and rhinitis is complex, and epidemiology and mechanisms of their comorbidities is unclear. Objectives: To investigate within-individual patterns of morbidity of eczema, wheeze, and rhinitis from birth to adolescence/early adulthood. Methods: We investigated onset, progression, and resolution of eczema, wheeze, and rhinitis using descriptive statistics, sequence mining, and latent Markov modeling in four population-based birth cohorts. We used logistic regression to ascertain if early-life eczema or wheeze, or genetic factors (filaggrin [FLG] mutations and 17q21 variants), increase the risk of multimorbidity. Measurements and Main Results: Single conditions, although the most prevalent, were observed significantly less frequently than by chance. There was considerable variation in the timing of onset/remission/persistence/intermittence. Multimorbidity of eczema+wheeze+rhinitis was rare but significantly overrepresented (three to six times more often than by chance). Although infantile eczema was associated with subsequent multimorbidity, most children with eczema (75.4%) did not progress to any multimorbidity pattern. FLG mutations and rs7216389 were not associated with persistence of eczema/wheeze as single conditions, but both increased the risk of multimorbidity (FLG by 2- to 3-fold, rs7216389 risk variant by 1.4- to 1.7-fold). Latent Markov modeling revealed five latent states (no disease/low risk, mainly eczema, mainly wheeze, mainly rhinitis, multimorbidity). The most likely transition to multimorbidity was from eczema state (0.21). However, although this was one of the highest transition probabilities, only one-fifth of those with eczema transitioned to multimorbidity. Conclusions: Atopic diseases fit a multimorbidity framework, with no evidence for sequential atopic march progression. The highest transition to multimorbidity was from eczema, but most children with eczema (more than three-quarters) had no comorbidities.

Immunoinflammatory Biomarkers in Serum Are Associated with Disease Severity in Atopic Dermatitis.

BACKGROUND: A growing body of evidence links various biomarkers to atopic dermatitis (AD). Still, little is known about the association of specific biomarkers to disease characteristics and severity in AD. OBJECTIVE: To explore the relationship between various immunological markers in the serum and disease severity in a hospital cohort of AD patients. METHODS: Outpatients with AD referred to the Department of Dermatology, Bispebjerg Hospital, Copenhagen, Denmark, were divided into groups based on disease severity (SCORAD). Serum levels of a preselected panel of immunoinflammatory biomarkers were tested for association with disease characteristics. Two machine learning models were developed to predict SCORAD from the measured biomarkers. RESULTS: A total of 160 patients with AD were included; 53 (33.1%) with mild, 73 (45.6%) with moderate, and 34 (21.3%) with severe disease. Mean age was 29.2 years (range 6-70 years) and 84 (52.5%) were females. Numerous biomarkers showed a statistically significant correlation with SCORAD, with the strongest correlations seen for CCL17/thymus and activation-regulated chemokine (chemokine ligand-17/TARC) and CCL27/cutaneous T cell-attracting-chemokine (CTACK; Spearman R of 0.50 and 0.43, respectively, p < 0.001). Extrinsic AD patients were more likely to have higher mean SCORAD (p < 0.001), CCL17 (p < 0.001), CCL26/eotaxin-3 (p < 0.001), and eosinophil count (p < 0.001) than intrinsic AD patients. Predictive models for SCORAD identified CCL17, CCL27, serum total IgE, IL-33, and IL-5 as the most important predictors for SCORAD, but with weaker associations than single cytokines. CONCLUSIONS: Specific immunoinflammatory biomarkers in the serum, mainly of the Th2 pathway, are correlated with disease severity in patients with AD. Predictive models identified biomarkers associated with disease severity but this finding warrants further investigation.

Personalized prediction of daily eczema severity scores using a mechanistic machine learning model.

BACKGROUND: Atopic dermatitis (AD) is a chronic inflammatory skin disease with periods of flares and remission. Designing personalized treatment strategies for AD is challenging, given the apparent unpredictability and large variation in AD symptoms and treatment responses within and across individuals. Better prediction of AD severity over time for individual patients could help to select optimum timing and type of treatment for improving disease control. OBJECTIVE: We aimed to develop a proof of principle mechanistic machine learning model that predicts the patient-specific evolution of AD severity scores on a daily basis. METHODS: We designed a probabilistic predictive model and trained it using Bayesian inference with the longitudinal data from two published clinical studies. The data consisted of daily recordings of AD severity scores and treatments used by 59 and 334 AD children over 6 months and 16 weeks, respectively. Validation of the predictive model was conducted in a forward-chaining setting. RESULTS: Our model was able to predict future severity scores at the individual level and improved chance-level forecast by 60%. Heterogeneous patterns in severity trajectories were captured with patient-specific parameters such as the short-term persistence of AD severity and responsiveness to topical steroids, calcineurin inhibitors and step-up treatment. CONCLUSIONS: Our proof of principle model successfully predicted the daily evolution of AD severity scores at an individual level and could inform the design of personalized treatment strategies that can be tested in future studies. Our model-based approach can be applied to other diseases with apparent unpredictability and large variation in symptoms and treatment responses such as asthma.

Reliable Detection of Eczema Areas for Fully Automated Assessment of Eczema Severity from Digital Camera Images.

Assessing the severity of eczema in clinical research requires face-to-face skin examination by trained staff. Such approaches are resource-intensive for participants and staff, challenging during pandemics, and prone to inter- and intra-observer variation. Computer vision algorithms have been proposed to automate the assessment of eczema severity using digital camera images. However, they often require human intervention to detect eczema lesions and cannot automatically assess eczema severity from real-world images in an end-to-end pipeline. We developed a model to detect eczema lesions from images using data augmentation and pixel-level segmentation of eczema lesions on 1,345 images provided by dermatologists. We evaluated the quality of the obtained segmentation compared with that of the clinicians, the robustness to varying imaging conditions encountered in real-life images, such as lighting, focus, and blur, and the performance of downstream severity prediction when using the detected eczema lesions. The quality and robustness of eczema lesion detection increased by approximately 25% and 40%, respectively, compared with that of our previous eczema detection model. The performance of the downstream severity prediction remained unchanged. Use of skin segmentation as an alternative to eczema segmentation that requires specialist labeling showed the performance on par with when eczema segmentation is used.

Detecting Eczema Areas in Digital Images: An Impossible Task?

Assessing the severity of atopic dermatitis (AD, or eczema) traditionally relies on a face-to-face assessment by healthcare professionals and may suffer from inter- and intra-rater variability. With the expanding role of telemedicine, several machine learning algorithms have been proposed to automatically assess AD severity from digital images. Those algorithms usually detect and then delineate (segment) AD lesions before assessing lesional severity and are trained using the data of AD areas detected by healthcare professionals. To evaluate the reliability of such data, we estimated the inter-rater reliability of AD segmentation in digital images. Four dermatologists independently segmented AD lesions in 80 digital images collected in a published clinical trial. We estimated the inter-rater reliability of the AD segmentation using the intraclass correlation coefficient at the pixel and the area levels for different resolutions of the images. The average intraclass correlation coefficient was 0.45 ( standard error = 0.04 ) corresponding to a poor agreement between raters, whereas the degree of agreement for AD segmentation varied from image to image. The AD segmentation in digital images is highly rater dependent even among dermatologists. Such limitations need to be taken into consideration when AD segmentation data are used to train machine learning algorithms that assess eczema severity.

EczemaPred: A computational framework for personalised prediction of eczema severity dynamics.

BACKGROUND: Atopic dermatitis (AD) is a chronic inflammatory skin disease leading to substantial quality of life impairment with heterogeneous treatment responses. People with AD would benefit from personalised treatment strategies, whose design requires predicting how AD severity evolves for each individual. OBJECTIVE: This study aims to develop a computational framework for personalised prediction of AD severity dynamics. METHODS: We introduced EczemaPred, a computational framework to predict patient-dependent dynamic evolution of AD severity using Bayesian state-space models that describe latent dynamics of AD severity items and how they are measured. We used EczemaPred to predict the dynamic evolution of validated patient-oriented scoring atopic dermatitis (PO-SCORAD) by combining predictions from the models for the nine severity items of PO-SCORAD (six intensity signs, extent of eczema, and two subjective symptoms). We validated this approach using longitudinal data from two independent studies: a published clinical study in which PO-SCORAD was measured twice weekly for 347 AD patients over 17 weeks, and another one in which PO-SCORAD was recorded daily by 16 AD patients for 12 weeks. RESULTS: EczemaPred achieved good performance for personalised predictions of PO-SCORAD and its severity items daily to weekly. EczemaPred outperformed standard time-series forecasting models such as a mixed effect autoregressive model. The uncertainty in predicting PO-SCORAD was mainly attributed to that in predicting intensity signs (75% of the overall uncertainty). CONCLUSIONS: EczemaPred serves as a computational framework to make a personalised prediction of AD severity dynamics relevant to clinical practice. EczemaPred is available as an R package.

Impact of environmental factors in predicting daily severity scores of atopic dermatitis.

BACKGROUND: Atopic dermatitis (AD) is a chronic inflammatory skin disease that affects 20% of children worldwide. Environmental factors including weather and air pollutants have been shown to be associated with AD symptoms. However, the time-dependent nature of such a relationship has not been adequately investigated. This paper aims to assess whether real-time data on weather and air pollutants can make short-term prediction of AD severity scores. METHODS: Using longitudinal data from a published panel study of 177 paediatric patients followed up daily for 17 months, we developed a statistical machine learning model to predict daily AD severity scores for individual study participants. Exposures consisted of daily meteorological variables and concentrations of air pollutants, and outcomes were daily recordings of scores for six AD signs. We developed a mixed-effect autoregressive ordinal logistic regression model, validated it in a forward-chaining setting and evaluated the effects of the environmental factors on the predictive performance. RESULTS: Our model successfully made daily prediction of the AD severity scores, and the predictive performance was not improved by the addition of measured environmental factors. Potential short-term influence of environmental exposures on daily AD severity scores was outweighed by the underlying persistence of preceding scores. CONCLUSIONS: Our data does not offer enough evidence to support a claim that weather or air pollutants can make short-term prediction of AD signs. Inferences about the magnitude of the effect of environmental factors on AD severity scores require consideration of their time-dependent dynamic nature.