An interpretable predictive deep learning platform for pediatric metabolic diseases.

OBJECTIVES: Metabolic disease in children is increasing worldwide and predisposes a wide array of chronic comorbid conditions with severe impacts on quality of life. Tools for early detection are needed to promptly intervene to prevent or slow the development of these long-term complications. MATERIALS AND METHODS: No clinically available tools are currently in widespread use that can predict the onset of metabolic diseases in pediatric patients. Here, we use interpretable deep learning, leveraging longitudinal clinical measurements, demographical data, and diagnosis codes from electronic health record data from a large integrated health system to predict the onset of prediabetes, type 2 diabetes (T2D), and metabolic syndrome in pediatric cohorts. RESULTS: The cohort included 49 517 children with overweight or obesity aged 2-18 (54.9% male, 73% Caucasian), with a median follow-up time of 7.5 years and mean body mass index (BMI) percentile of 88.6%. Our model demonstrated area under receiver operating characteristic curve (AUC) accuracies up to 0.87, 0.79, and 0.79 for predicting T2D, metabolic syndrome, and prediabetes, respectively. Whereas most risk calculators use only recently available data, incorporating longitudinal data improved AUCs by 13.04%, 11.48%, and 11.67% for T2D, syndrome, and prediabetes, respectively, versus models using the most recent BMI (P < 2.2 × 10-16). DISCUSSION: Despite most risk calculators using only the most recent data, incorporating longitudinal data improved the model accuracies because utilizing trajectories provides a more comprehensive characterization of the patient's health history. Our interpretable model indicated that BMI trajectories were consistently identified as one of the most influential features for prediction, highlighting the advantages of incorporating longitudinal data when available.

Gut microbial trimethylamine is elevated in alcohol-associated hepatitis and contributes to ethanol-induced liver injury in mice.

There is mounting evidence that microbes residing in the human intestine contribute to diverse alcohol-associated liver diseases (ALD) including the most deadly form known as alcohol-associated hepatitis (AH). However, mechanisms by which gut microbes synergize with excessive alcohol intake to promote liver injury are poorly understood. Furthermore, whether drugs that selectively target gut microbial metabolism can improve ALD has never been tested. We used liquid chromatography tandem mass spectrometry to quantify the levels of microbe and host choline co-metabolites in healthy controls and AH patients, finding elevated levels of the microbial metabolite trimethylamine (TMA) in AH. In subsequent studies, we treated mice with non-lethal bacterial choline TMA lyase (CutC/D) inhibitors to blunt gut microbe-dependent production of TMA in the context of chronic ethanol administration. Indices of liver injury were quantified by complementary RNA sequencing, biochemical, and histological approaches. In addition, we examined the impact of ethanol consumption and TMA lyase inhibition on gut microbiome structure via 16S rRNA sequencing. We show the gut microbial choline metabolite TMA is elevated in AH patients and correlates with reduced hepatic expression of the TMA oxygenase flavin-containing monooxygenase 3 (FMO3). Provocatively, we find that small molecule inhibition of gut microbial CutC/D activity protects mice from ethanol-induced liver injury. CutC/D inhibitor-driven improvement in ethanol-induced liver injury is associated with distinct reorganization of the gut microbiome and host liver transcriptome. The microbial metabolite TMA is elevated in patients with AH, and inhibition of TMA production from gut microbes can protect mice from ethanol-induced liver injury.

Associations of weight loss with obesity-related comorbidities in a large integrated health system.

AIMS: To determine the health outcomes associated with weight loss in individuals with obesity, and to better understand the relationship between disease burden (disease burden; ie, prior comorbidities, healthcare utilization) and weight loss in individuals with obesity by analysing electronic health records (EHRs). MATERIALS AND METHODS: We conducted a case-control study using deidentified EHR-derived information from 204 921 patients seen at the Cleveland Clinic between 2000 and 2018. Patients were aged ≥20 years with body mass index ≥30 kg/m2 and had ≥7 weight measurements, over ≥3 years. Thirty outcomes were investigated, including chronic and acute diseases, as well as psychological and metabolic disorders. Weight change was investigated 3, 5 and 10 years prior to an event. RESULTS: Weight loss was associated with reduced incidence of many outcomes (eg, type 2 diabetes, nonalcoholic steatohepatitis/nonalcoholic fatty liver disease, obstructive sleep apnoea, hypertension; P < 0.05). Weight loss >10% was associated with increased incidence of certain outcomes including stroke and substance abuse. However, many outcomes that increased with weight loss were attenuated by disease burden adjustments. CONCLUSIONS: This study provides the most comprehensive real-world evaluation of the health impacts of weight change to date. After comorbidity burden and healthcare utilization adjustments, weight loss was associated with an overall reduction in risk of many adverse outcomes.