Generation and application of a convolutional neural networks algorithm in evaluating stool consistency in diapers.

AIM: The aim of the study was to develop a deep convolutional neural networks (CNNs) algorithm for automated assessment of stool consistency from diaper photographs and test its performance under real-world conditions. METHODS: Diaper photographs were enrolled via a mobile phone application. The stool consistency was assessed independently according to the Brussels Infant and Toddler Stool Scale (BITSS) by paediatricians. These images were randomised into a training data set and a test data set. After training and testing, the new algorithm was used under real-world conditions by parents. RESULTS: There was an overall agreement of 92.9% between paediatricians and the CNN-generated algorithm. Post hoc classification into the validated 4 categories of the BITSS yielded an agreement of 95.4%. Spearman correlation analysis across the ranking of 7 BITSS photographs and validated 4 categories showed a significant correlation of rho = 0.93 (95% CI, 0.92, 0.94; p < 0.001) and rho = 0.92 (95% CI, 0.90, 0.93; p < 0.001), respectively. The real-world application yielded further insights into changes in stool consistency between age categories and mode of feeding. CONCLUSION: The new CNN-based algorithm is able to reliably identify stool consistency from diaper photographs and may support the communication between parents and paediatricians.

The impact of incorporating toilet-training status in the pediatric Rome IV criteria for functional constipation in infant and toddlers.

BACKGROUND: The diagnosis of functional constipation (FC) is based on the Rome criteria. The last edition of the criteria (Rome IV) for infants and toddlers modified the criteria to differentiate toilet-trained (TT) and non-toilet-trained (NTT) children. These changes have not been validated. We aimed to understand the impact of adding toilet training to the diagnostic criteria and to assess the prevalence of FC. METHODS: Parents of infants and toddlers from six outpatient clinics (four public, two private) located in three geographically dispersed cities in Colombia completed validated questionnaires to diagnose functional gastrointestinal disorders according to Spanish version of Rome IV criteria (QPGS-IV). RESULTS: A total of 1334 children (24.4 months ±15.0) participated: 482 (36%) TT and 852 (64%) NTT. The prevalence of FC was 21.1%. The prevalence increased with age, 0-1 years 7.7%; 2 years 18.2%; 3 years 23.7%; and 4 years 37.2%. TT vs NTT for FC 41.9% vs 9.3%, respectively (OR 7.06, 95% CI 5.26-9.47, P < .0001). TT more likely to report ≥ 3 criteria (OR = 2.43, 95% CI 1.41-4.21, P = .0015). 18.3% of TT had episodes of fecal incontinence that met the frequency required by Rome for FC (≤1 episode/week). However, 87.1% had fecal incontinence less often. 7.4% of them characterized as large quantity. CONCLUSION: We found no changes in the prevalence of FC using the Rome IV criteria vs Rome III. TT children are more likely to have FC. Study suggests that changes in Rome IV criteria were potentially clinically relevant and to have adequate face validity. Future studies should confirm our findings.

Urine collection methods used for non-toilet-trained children in pediatric emergency departments in France: A medical practice analysis.

INTRODUCTION: Since April 2015, the French Society of Pediatrics has encouraged suprapubic aspiration (SA) and urethral catheterization (UC) for urine collection in non-toilet-trained children suspected of having urinary tract infections (UTIs) and has tried to reduce the use of urine bag (UB). OBJECTIVE: To analyze the medical practices concerning urine collection methods used for non-toilet-trained children in pediatric emergency departments in France. METHODS: We conducted a descriptive medical practice study in October 2017. All members of the French Society of Pediatric Nephrology received two questionnaires about urine collection methods used for non-toilet-trained children, distinguishing between male and female patients, and about the corresponding analgesic protocols used in their pediatric emergency departments. RESULTS: In total, 26 centers completed questionnaires concerning female patients. UC was performed in cases of fever associated with urinary tract malformations in 14 of 26 centers (54%). UB was used in cases of fever of unknown origin lasting for more than 48h in 17 of 26 centers (65%), in cases of fever associated with UTI symptoms in 14 of 26 centers (54%), and in cases of fever in infants under 3 months of age in 16 of 26 centers (61%). The questionnaires concerning male patients were completed by 30 centers. UB was the initially used urine collection method in all situations with, respectively, 22 of 30 (73%), 27 of 30 (90%), 23 of 30 (77%), and 22 of 30 (73%) centers. The analgesic protocol for urine collection is not well established in France. CONCLUSION: UC for urine collection in pediatric emergency departments in France is underused despite the national recommendations and the greater diagnostic power of this method compared with UB.

Urine collection methods for non-toilet-trained children in biological monitoring studies: Validation of a disposable diaper for characterization of tebuconazole exposure.

Young children differ from adults in their exposure and susceptibility to environmental chemicals (e.g. pesticides) because of various factors such as behavior, diet and physiology. Their heightened vulnerability to environmental stressors makes it important to obtain appropriate urine samples for exposure characterization. However, collecting urine from non-toilet-trained children has been shown to be methodologically and practically challenging. Four urine collection approaches were tested: a disposable diaper, a urine bag, a collection pad and the clean catch. The success rate and the user rating of each method was evaluated. The success rates were 67%, 21%, 17% and 4% for the disposable diaper, urine bag, collection pad and clean catch, respectively. The average user ratings on a 0-10 (0 = inconvenient, 10 = convenient) scale were 9.0, 4.7, 7.3 and 2.5, respectively. Subsequently, the best rated method, the disposable polyacrylate diaper was tested with hydroxy-tebuconazole as an exposure biomarker for the fungicide tebuconazole and creatinine for urine density adjustment. After LC-MS/MS analysis, the recoveries of hydroxy-tebuconazole in the range of 0.05-25 ng/mL were on average 106%, and for creatinine 87%. Precisions (relative standard deviation) were for both 3%. The overall procedure including collection and extraction was assessed, resulting in three out of seven positive samples. Based on this study, the disposable diaper is a suitable method for urine collection of non-toilet-trained children for biomonitoring of tebuconazole. This method can serve as a basis for extension to other substances of interest.