Changes in the pattern of fibrosis in the rat lung with repetitive orotracheal instillations of gastric contents: evidence of persistent collagen accumulation.

Recurrent aspiration of gastric contents has been associated with several interstitial lung diseases. Despite this association, the pathogenic role of aspiration in these diseases has been poorly studied and little is known about extracellular matrix (ECM) changes in animal models of repetitive events of aspiration. Our aim was to study the repair phase of lung injury induced by each of several instillations of gastric fluid in Sprague-Dawley rats to evaluate changes in ECM and their reversibility. Anesthetized animals received weekly orotracheal instillations of gastric fluid for 1, 2, 3, and 4 wk and were euthanized at day 7 after last instillation. For reversibility studies, another group received 7 weekly instillations and was euthanized at day 7 or 60 after last instillation. Biochemical and histological measurements were used to evaluate ECM changes. Lung hydroxyproline content increased progressively and hematoxylin and eosin, Masson's trichrome, and alpha-SMA stains showed that after a single instillation, intra-alveolar fibrosis predominated, whereas with repetitive instillations this fibrosis pattern became less prominent and interstitial fibrosis progressively became evident. Both type I and III collagen increased in intra-alveolar and interstitial fibrosis. Imbalance between matrix metalloproteinase-2 (MMP-2) activity and tissue inhibitor of metalloproteinase-2 (TIMP-2) expression was observed, favoring either collagen degradation or accumulation depending on the number of instillations. Caspase-3 activation was also dose dependent. ECM changes were partially reversible at long-term evaluation, since Masson bodies, granulomas, and foreign body giant cells disappeared, whereas interstitial collagen accumulated. In conclusion, repetitive lung instillations of gastric fluid induce progressive fibrotic changes in rat lung ECM that persist at long-term evaluation.

Elastin degradation products in acute lung injury induced by gastric contents aspiration.

BACKGROUND: Gastric contents aspiration is a high-risk condition for acute lung injury (ALI). Consequences range from subclinical pneumonitis to respiratory failure, depending on the volume of aspirate. A large increment in inflammatory cells, an important source of elastase, potentially capable of damaging lung tissue, has been described in experimental models of aspiration. We hypothesized that in early stages of aspiration-induced ALI, there is proteolytic degradation of elastin, preceding collagen deposition. Our aim was to evaluate whether after a single orotracheal instillation of gastric fluid, there is evidence of elastin degradation. METHODS: Anesthesized Sprague-Dawley rats received a single orotracheal instillation of gastric fluid and were euthanized 4, 12 and 24 h and at day 4 after instillation (n = 6/group). We used immunodetection of soluble elastin in lung tissue and BALF and correlated BALF levels of elastin degradation products with markers of ALI. We investigated possible factors involved in elastin degradation and evaluated whether a similar pattern of elastin degradation can be found in BALF samples of patients with interstitial lung diseases known to have aspirated. Non-parametric ANOVA (Kruskall-Wallis) and linear regression analysis were used. RESULTS: We found evidence of early proteolytic degradation of lung elastin. Elastin degradation products are detected both in lung tissue and BALF in the first 24 h and are significantly reduced at day 4. They correlate significantly with ALI markers, particularly PMN cell count, are independent of acidity and have a similar molecular weight as those obtained using pancreatic elastase. Evaluation of BALF from patients revealed the presence of elastin degradation products not present in controls that are similar to those found in BALF of rats treated with gastric fluid. CONCLUSIONS: A single instillation of gastric fluid into the lungs induces early proteolytic degradation of elastin, in relation to the magnitude of alveolar-capillary barrier derangement. PMN-derived proteases released during ALI are mostly responsible for this damage. BALF from patients showed elastin degradation products similar to those found in rats treated with gastric fluid. Long-lasting effects on lung elastic properties could be expected under conditions of repeated instillations of gastric fluid in experimental animals or repeated aspiration events in humans.

Acute lung injury by gastric fluid instillation: activation of myofibroblast apoptosis during injury resolution.

BACKGROUND: Gastric contents aspiration in humans has variable consequences depending on the volume of aspirate, ranging from subclinical pneumonitis to respiratory failure with up to 70% mortality. Several experimental approaches have been used to study this condition. In a model of single orotracheal instillation of gastric fluid we have shown that severe acute lung injury evolves from a pattern of diffuse alveolar damage to one of organizing pneumonia (OP), that later resolves leaving normal lung architecture. Little is known about mechanisms of injury resolution after a single aspiration that could be dysregulated with repetitive aspirations. We hypothesized that, in a similar way to cutaneous wound healing, apoptosis may participate in lung injury resolution by reducing the number of myofibroblasts and by affecting the balance between proteases and antiproteases. Our aim was to study activation of apoptosis as well as MMP-2/TIMP-2 balance in the sub-acute phase (4-14 days) of gastric fluid-induced lung injury. METHODS: Anesthesized Sprague-Dawley rats received a single orotracheal instillation of gastric fluid and were euthanized 4, 7 and 14 days later (n = 6/group). In lung tissue we studied caspase-3 activation and its location by double immunofluorescence for cleaved caspase-3 or TUNEL and alpha-SMA. MMP-2/TIMP-2 balance was studied by zymography and Western blot. BALF levels of TGF-ß1 were measured by ELISA. RESULTS: An OP pattern with Masson bodies and granulomas was seen at days 4 and 7 that was no longer present at day 14. Cleaved caspase-3 increased at day 7 and was detected by immunofluorescence in Masson body-alpha-SMA-positive and -negative cells. TUNEL-positive cells at days 4 and 7 were located mainly in Masson bodies. Distribution of cleaved caspase-3 and TUNEL-positive cells at day 14 was similar to that in controls. At the peak of apoptosis (day 7), an imbalance between MMP-2 activity and TIMP-2 expression was produced by reduction in TIMP-2 expression. CONCLUSIONS: Apoptosis is activated in Masson body-alpha-SMA-positive and -negative cells during the sub-acute phase of gastric fluid-induced lung injury. This mechanism likely contributes to OP resolution, by reducing myofibroblast number and new collagen production. In addition, pre-formed collagen degradation is favored by an associated MMP-2/TIMP-2 imbalance.

Gestational Diabetes: Glycemic Control in the Last Two Weeks Before Delivery Contributes to Newborn Insulinemia.

INTRODUCTION: Fetal hyperinsulinemia in gestational diabetes mellitus (GDM) not only is important during intrauterine life, a time when it can result in macrosomia, but also at delivery, since it can result in neonatal hypoglycemia and hyperbilirubinemia. The question is, how long before delivery does maternal glycemic control contribute to newborn insulinemia in GDM? METHODS: In 72 women with GDM, we calculated Spearman's rank (rs) correlations between umbilical cord blood C-peptide at birth (a biomarker of insulin secretion), and both maternal glycosylated hemoglobin (HbA1c) and mean blood glucose (MBG) recorded in the last two visits prior to delivery. Iterative correlations were done between umbilical cord blood C-peptide at birth, and maternal glucose control, at 0, 1, 2, 3, 4, and 5 weeks before delivery. RESULTS: At an early visit (32.95 ± 1.8 weeks), rs = 0.353 (P = 0.07) between HbA1c and C-peptide, whereas rs = 0.244 (P = 0.186) between MBG and C-peptide. At the latest visit (35.04 ± 1.6 weeks), rs = 0.456 (P = 0.004) between HbA1c versus C-peptide, and rs = 0.359 (P = 0.023) between MBG versus C-peptide. Iterative correlations between MBG and C-peptide became significant at 2 weeks before delivery. CONCLUSION: To further reduce the risk of hypoglycemia and hyperbilirubinemia in infants born to women with GDM, besides applying a strict in-patient glucose control protocol at delivery, it is necessary to improve even more the quality of maternal glucose control during the last 2 weeks prior to delivery.

47-Fold rise of diabetes in childbearing age Chilean women: Markov model and cost-effectiveness of prevention of birth defects.

AIM: Prevalence of type 2 diabetes mellitus (T2DM) during childbearing age in Chile had a 47-fold rise in 7 years, reaching 120 844 women, half of which are unaware of their condition. We aimed to project pregnancies and births among Chilean women of childbearing age (WCBA) with T2DM and report the incidence of birth defects and the associated years of life lost and lifetime costs. METHODS: Markov model of cohort of WCBA with T2DM (WCBA-DM) with a 20-year time horizon (2018-2037), using data from previous studies. Two scenarios were assessed: scenario A: no universal detection of T2DM and scenario B: universal screening of T2DM using glycosylated hemoglobin levels. Both lifetime costs and disability-adjusted life years (DALY) were calculated with a 5% discount rate (US$ of 2017). RESULTS: In scenario A, 12 163 infants with birth defects could be born among the analyzed cohort, resulting in 243 260 years of life lost, 296 652 DALY and in lifetime costs of US$ 1 957 657 966. In scenario B, the first three figures could be reduced by 70.4% to 3599 infants with birth defects, 71 980 years of life lost and 87 794 DALY. Due to the addition of diabetes screening and new patient costs to scenario B, there would be a lesser reduction (67.3%) in total lifetime costs, to US$ 640 669 296. CONCLUSION: Screening of diabetes in WCBA would yield a 20-year reduction of 70.4% in the number of infants with birth defects, years of life lost and DALY. Total lifetime costs could be reduced by 67.3%.

Basal-bolus insulin therapy reduces maternal triglycerides in gestational diabetes without modifying cholesteryl ester transfer protein activity.

AIM: Macrosomia in the offspring of overweight/obese mothers with glucose-controlled gestational diabetes mellitus (GDM) is due to excessive rise of maternal triglycerides (TG). We aimed to ascertain whether basal-bolus insulin therapy (BBIT), or other components of the treatment, could reduce TG in GDM. METHODS: We studied the records of 131 singleton pregnancies with GDM, using stepwise multiple linear regression, Mann-Whitney, χ2 , and Jonckheere-Terpstra tests. As maternal TG increased steadily during normal pregnancy, these were transformed as z-scores. The atherogenic index of plasma (AIP) was calculated as a measure of cholesteryl ester transfer protein activity. RESULTS: Multiple regression showed that only BBIT (but neither limitation of weight gain nor metformin) reduced maternal TG z-scores (P = 0.011). When the 131 pregnancies were split into two groups - without BBIT (n = 58; HbA1c = 5.3 ± 0.3%) and with BBIT (n = 73; HbA1c = 5.4 ± 0.6; P = 0.2005) - we observed that BBIT (n = 73) reduced maternal TG z-scores in a dose-related fashion (Jonckheere-Terpstra P = 0.03817). The atherogenic index of plasma remained within normal range in both groups. CONCLUSION: BBIT (but not weight gain control nor metformin) reduced maternal TG in mothers with glucose-controlled GDM. This beneficial effect of BBIT was not related to changes in the cholesteryl ester transfer protein activity.

Acute lung injury induced by whole gastric fluid: hepatic acute phase response contributes to increase lung antiprotease protection.

BACKGROUND: Gastric contents aspiration in humans is a risk factor for severe respiratory failure with elevated mortality. Although aspiration-induced local lung inflammation has been studied in animal models, little is known about extrapulmonary effects of aspiration. We investigated whether a single orotracheal instillation of whole gastric fluid elicits a liver acute phase response and if this response contributes to enrich the alveolar spaces with proteins having antiprotease activity. METHODS: In anesthetized Sprague-Dawley rats receiving whole gastric fluid, we studied at different times after instillation (4 h -7 days): changes in blood cytokines and acute phase proteins (fibrinogen and the antiproteases alpha1-antitrypsin and alpha2-macroglobulin) as well as liver mRNA expression of the two antiproteases. The impact of the systemic changes on lung antiprotease defense was evaluated by measuring levels and bioactivity of antiproteases in broncho-alveolar lavage fluid (BALF). Markers of alveolar-capillary barrier derangement were also studied. Non-parametric ANOVA (Kruskall-Wallis) and linear regression analysis were used. RESULTS: Severe peribronchiolar injury involving edema, intra-alveolar proteinaceous debris, hemorrhage and PMNn cell infiltration was seen in the first 24 h and later resolved. Despite a large increase in several lung cytokines, only IL-6 was found elevated in blood, preceding increased liver expression and blood concentration of both antiproteases. These changes, with an acute phase response profile, were significantly larger for alpha2-macroglobulin (40-fold increment in expression with 12-fold elevation in blood protein concentration) than for alpha1-antitrypsin (2-3 fold increment in expression with 0.5-fold elevation in blood protein concentration). Both the increment in capillary-alveolar antiprotease concentration gradient due to increased antiprotease liver synthesis and a timely-associated derangement of the alveolar-capillary barrier induced by aspiration, contributed a 58-fold and a 190-fold increase in BALF alpha1-antitrypsin and alpha2-macroglobulin levels respectively (p < 0.001). CONCLUSIONS: Gastric contents-induced acute lung injury elicits a liver acute phase response characterized by increased mRNA expression of antiproteases and elevation of blood antiprotease concentrations. Hepatic changes act in concert with derangement of the alveolar capillary barrier to enrich alveolar spaces with antiproteases. These findings may have significant implications decreasing protease burden, limiting injury in this and other models of acute lung injury and likely, in recurrent aspiration.

Infinite Continuous Feature Model for Psychiatric Comorbidity Analysis.

We aim at finding the comorbidity patterns of substance abuse, mood and personality disorders using the diagnoses from the National Epidemiologic Survey on Alcohol and Related Conditions database. To this end, we propose a novel Bayesian nonparametric latent feature model for categorical observations, based on the Indian buffet process, in which the latent variables can take values between 0 and 1. The proposed model has several interesting features for modeling psychiatric disorders. First, the latent features might be off, which allows distinguishing between the subjects who suffer a condition and those who do not. Second, the active latent features take positive values, which allows modeling the extent to which the patient has that condition. We also develop a new Markov chain Monte Carlo inference algorithm for our model that makes use of a nested expectation propagation procedure.

Preventive letter: doubling the return rate after gestational diabetes mellitus.

To measure the impact of a "Preventive Letter" designed to encourage the return of gestational diabetes mellitus (GDM) mothers to follow up visit after delivery, in the context of a worldwide concern about low return rates after delivery of these patients. Mothers with GDM require medical evaluation and an oral glucose tolerance test (OGTT) 6 weeks after delivery, in order to: [a] confirm remission of GDM and [b] provide advice on the prevention of type 2 diabetes. In the year 2003 we developed a "Preventive Letter", containing three aspects: [a] current treatment, [b] suggested management during labor, and [c] a stapled laboratory order for OGTT to be performed 6 weeks after delivery. The return rate after delivery was assessed in two groups of GDM mothers: [a] "Without Preventive Letter" (n = 253), and "With Preventive Letter" (n = 215). Both groups, similar with respect to age (33.0 ± 5.4 and 32.3 ± 4.9 years respectively, p = 0.166) and education time (14.9 ± 1.8 and 15.0 ± 1.8 years respectively, p = 0.494), showed a significant difference in the 1-year return rate after delivery, as assessed by the Kaplan-Meier test: 32.0 % for the group "Without Preventive Letter", and 76.0 % for the group "With Preventive Letter" (p < 0.001). The 1-year return rate after delivery of GDM mothers was 2.4 times higher in the group "With Preventive Letter" than in the group without it. We believe that this low-cost approach could be useful in other institutions caring for pregnant women with diabetes.

[Treatment of diabetic ketoacidosis using 2009 American Diabetes Association guidelines]. / Cetoacidosis diabética: Casuística 2008-2012, epidemiología y fisiopatología.

BACKGROUND: During 2009, new guidelines for the treatment of diabetic ketoacidosis were published by the American Diabetes Association. AIM: To assess the impact of new treatment guidelines on the evolution of patients treated for diabetic ketoacidosis (KAD). PATIENTS AND METHODS: Anonymous data was obtained from computational medical records of patients treated for KAD at our institution two years before ("Traditional Protocol") and TWO years after ("ADA-2009 Protocol") the publication of the 2009 American Diabetes Association (ADA) KAD guidelines. RESULTS: Twenty three patients aged 36.5 ± 15.1 years were treated with the traditional method and 23 patients aged 44.4 ± 21.1 years were treated following 2009 ADA guidelines. Among patients treated with the traditional protocol and treated following ADA 2009 guidelines, the diabetes type 1/type 2 ratio was 18/5 and 19/16 respectively (p = NS), the glycosylated hemoglobin on admission was 12.6 ± 2.5 and 14.3 ± 2.7% respectively (p = 0.03), minimal blood pH was 7.15 ± 0.14 and 7.19 ± 0.09 respectively (p = NS), bicarbonate was required in seven and no patient respectively (p = 0.01), hypokalemia < 3.5 mEq/L occurred in 78.2 and 48.5% of patients (p = 0.03), the lapse until resolution was 28.7 ± 28.0 and 28.8 ± 20.6 hours (p = NS). Only one patient, treated following ADA 2009 guidelines, died. CONCLUSIONS: Introduction of the ADA-2009 protocol for the treatment of KAD resulted in decrease in the use of intravenous bicarbonate and a reduction in the incidence of hypokalemia. There was no impact neither in the lapse until resolution or lethality.

[Mitochondrial DNA heteroplasmy of the m.3243A>G mutation in maternally inherited diabetes and deafness]. / Heteroplasmia de la mutación del ADN mitocondrial m.3243A>G en la diabetes y sordera de herencia materna.

Maternally Inherited Diabetes and Deafness (MIDD) is caused by mutations in mitochondrial DNA (mtDNA), mainly m.3243A>G. Severity, onset and clinical phenotype of MIDD patients are partially determined by the proportion of mutant mitochondrial DNA copies in each cell and tissue (heteroplasmy). The identification of MIDD allows a corred treatment with insulin avoiding drugs that may interfere with mitochondrial electrón chain transpon. We estimated the degree of heteroplasmy of the mutation m.3243A>G from blood, saliva, hair root and a muscle biopsy using quantitative PCR (qPCR) in a femóle adult patient. For this purpose, PCR producís were inserted in a vector creating plasmids with 3243A or G. Mutant and wild-type vectors were mixed in different proportions to créate a calibration curve used to interpólate heteroplasmy percentages with qPCR threshold cycles. The proportions of m.3243A>G heteroplasmy were 62% (muscle), 14% (saliva), 6% (blood leukocytes) and 3% in hair root. Quantitative analysis of heteroplasmy showed marked variations in different tissues (highest in muscle and lowest in blood). Given the relatively high heteroplasmy found in saliva, this type of biológical sample may represent an adequate non-invasive way for assessing the presence of m.3243A>G mutations in epidemiologic studies.

[Nutrients other than glucose might explain fetal overgrowth in gestational diabetic pregnancies]. / La hipótesis de Pedersen no es suficiente: otros nutrientes además de la glucosa explicarían la macrosomía fetal en pacientes diabéticas gestacionales con sobrepeso y buen control glicémico.

Since 1964, the hypothesis of Pedersen has been used to explain fetal macrosomia observed in gestational diabetes mellitus (GDM), by a mechanism involving maternal hyperglycemia--fetal hyperglycemia--fetal hyperinsulinemia. However, since the 1980-89 decade, it is known that pregnant women with pre-gestational overweight not suffering from GDM still have a higher frequency of fetal macrosomia. Furthermore, pregnant women with GDM, despite being subjected to optimal glycemic control, still show unacceptably high frequencies of fetal macrosomia, a phenomenon that is concentrated in pregnancies with overweight or obesity prior to pregnancy. If glucose is not the single nutrient responsible for fetal macrosomia in pregnant women with gestational diabetes that undergo strict glycemic control, other nutrients may cause excessive fetal growth in pre-pregnancy overweight mothers. In this review, we propose that triglycerides (TG) could be responsible for this accelerated fetal growth. If this hypothesis is validated in animal models and clinical studies, then normal and pathological ranges of TG should be defined, and monitoring of triglyceride levels during pregnancy should be advised as a possible new alternative, besides a good glycemic control, for the management of fetal macrosomia in GDM women with overweight prior to pregnancy.

Handling Ill-Conditioned Omics Data With Deep Probabilistic Models.

The advent of high-throughput technologies has produced an increase in the dimensionality of omics datasets, which limits the application of machine learning methods due to the great unbalance between the number of observations and features. In this scenario, dimensionality reduction is essential to extract the relevant information within these datasets and project it in a low-dimensional space, and probabilistic latent space models are becoming popular given their capability to capture the underlying structure of the data as well as the uncertainty in the information. This article aims to provide a general classification and dimensionality reduction method based on deep latent space models that tackles two of the main problems that arise in omics datasets: the presence of missing data and the limited number of observations against the number of features. We propose a semi-supervised Bayesian latent space model that infers a low-dimensional embedding driven by the target label: the Deep Bayesian Logistic Regression (DBLR) model. During inference, the model also learns a global vector of weights that allows it to make predictions given the low-dimensional embedding of the observations. Since this kind of dataset is prone to overfitting, we introduce an additional probabilistic regularization method based on the semi-supervised nature of the model. We compared the performance of the DBLR against several state-of-the-art methods for dimensionality reduction, both in synthetic and real datasets with different data types. The proposed model provides more informative low-dimensional representations, outperforms the baseline methods in classification, and can naturally handle missing entries.

Automatic patient functionality assessment from multimodal data using deep learning techniques - Development and feasibility evaluation.

Wearable devices and mobile sensors enable the real-time collection of an abundant source of physiological and behavioural data unobtrusively. Unlike traditional in-person evaluation or ecological momentary assessment (EMA) questionnaire-based approaches, these data sources open many possibilities in remote patient monitoring. However, defining robust models is challenging due to the data's noisy and frequently missing observations. This work proposes an attention-based Long Short-Term Memory (LSTM) neural network-based pipeline for predicting mobility impairment based on WHODAS 2.0 evaluation from such digital biomarkers. Furthermore, we addressed the missing observation problem by utilising hidden Markov models and the possibility of including information from unlabelled samples via transfer learning. We validated our approach using two wearable/mobile sensor data sets collected in the wild and socio-demographic information about the patients. Our results showed that in the WHODAS 2.0 mobility impairment prediction task, the proposed pipeline outperformed a prior baseline while additionally providing interpretability with attention heatmaps. Moreover, using a much smaller cohort via task transfer learning, the same model could learn to predict generalised anxiety severity accurately based on GAD-7 scores.

Regularizing transformers with deep probabilistic layers.

Language models (LM) have grown non-stop in the last decade, from sequence-to-sequence architectures to attention-based Transformers. However, regularization is not deeply studied in those structures. In this work, we use a Gaussian Mixture Variational Autoencoder (GMVAE) as a regularizer layer. We study its advantages regarding the depth where it is placed and prove its effectiveness in several scenarios. Experimental result demonstrates that the inclusion of deep generative models within Transformer-based architectures such as BERT, RoBERTa, or XLM-R can bring more versatile models, able to generalize better and achieve improved imputation score in tasks such as SST-2 and TREC or even impute missing/noisy words with richer text.

A Quinoa Protein Hydrolysate Fractionated by Electrodialysis with Ultrafiltration Membranes Improves Maternal and Fetal Outcomes in a Mouse Model of Gestational Diabetes Mellitus.

SCOPE: Quinoa intake exerts hypoglycemic and hypolipidemic effects in animals and humans. Although peptides from quinoa inhibit key enzymes involved in glucose homeostasis in vitro, their in vivo antidiabetic properties have not been investigated. METHODS AND RESULTS: This study evaluated the effect of oral administration of a quinoa protein hydrolysate (QH) produced through enzymatic hydrolysis and fractionation by electrodialysis with ultrafiltration membrane (EDUF) (FQH) on the metabolic and pregnancy outcomes of Lepdb/+ pregnant mice, a preclinical model of gestational diabetes mellitus. The 4-week pregestational consumption of 2.5 mg mL-1 of QH in water prevented glucose intolerance and improves hepatic insulin signaling in dams, also reducing fetal weights. Sequencing and bioinformatic analyses of the defatted FQH (FQHD) identified 11 peptides 6-10 amino acids long that aligned with the quinoa proteome and exhibited putative anti-dipeptidyl peptidase-4 (DPP-IV) activity, confirmed in vitro in QH, FQH, and FDQH fractions. Peptides homologous to mouse and human proteins enriched for biological processes related to glucose metabolism are also identified. CONCLUSION: Processing of quinoa protein may be used to develop a safe and effective nutritional intervention to control glucose intolerance during pregnancy. Further studies are required to confirm if this nutritional intervention is applicable to pregnant women.

Continuous Assessment of Function and Disability via Mobile Sensing: Real-World Data-Driven Feasibility Study.

BACKGROUND: Functional limitations are associated with poor clinical outcomes, higher mortality, and disability rates, especially in older adults. Continuous assessment of patients' functionality is important for clinical practice; however, traditional questionnaire-based assessment methods are very time-consuming and infrequently used. Mobile sensing offers a great range of sources that can assess function and disability daily. OBJECTIVE: This work aims to prove the feasibility of an interpretable machine learning pipeline for predicting function and disability based on the World Health Organization Disability Assessment Schedule (WHODAS) 2.0 outcomes of clinical outpatients, using passively collected digital biomarkers. METHODS: One-month-long behavioral time-series data consisting of physical and digital activity descriptor variables were summarized using statistical measures (minimum, maximum, mean, median, SD, and IQR), creating 64 features that were used for prediction. We then applied a sequential feature selection to each WHODAS 2.0 domain (cognition, mobility, self-care, getting along, life activities, and participation) in order to find the most descriptive features for each domain. Finally, we predicted the WHODAS 2.0 functional domain scores using linear regression using the best feature subsets. We reported the mean absolute errors and the mean absolute percentage errors over 4 folds as goodness-of-fit statistics to evaluate the model and allow for between-domain performance comparison. RESULTS: Our machine learning-based models for predicting patients' WHODAS functionality scores per domain achieved an average (across the 6 domains) mean absolute percentage error of 19.5%, varying between 14.86% (self-care domain) and 27.21% (life activities domain). We found that 5-19 features were sufficient for each domain, and the most relevant being the distance traveled, time spent at home, time spent walking, exercise time, and vehicle time. CONCLUSIONS: Our findings show the feasibility of using machine learning-based methods to assess functional health solely from passively sensed mobile data. The feature selection step provides a set of interpretable features for each domain, ensuring better explainability to the models' decisions-an important aspect in clinical practice.

Gestational diabetes and pre-pregnancy overweight: possible factors involved in newborn macrosomia.

AIM: Good glycemic control in gestational diabetes mellitus (GDM) seems not to be enough to prevent macrosomia (large-for-gestational-age newborns). In GDM pregnancies we studied the effects of glycemic control (as glycosylated hemoglobin [HbA1c]), pre-pregnancy body mass index (PP-BMI) and gestational weight gain per week (GWG-W) on the frequency of macrosomia. METHODS: We studied 251 GDM pregnancies, divided into two groups: PP-BMI<25.0kg/m(2) (the non-overweight group; n=125), and PP-BMI≥25.0kg/m(2) (the overweight group; n=126). A newborn weight Z-score>1.28 was considered large-for-gestational-age. Statistical analysis was carried out using the Student's t-test and χ(2) -test, receiver-operator characteristic curves and linear and binary logistic regressions. RESULTS: Prevalence of macrosomia was 14.9% among GDM (n=202/251, 88.4%) with good glycemic control (mean HbA1c<6.0%), and 28.1% in those with mean HbA1c≥6.0% (n=49/251, P<0.025). Macrosomia rates were 10.4% in the non-overweight group and 24.6% in the overweight group (P=0.00308), notwithstanding both having similar mean HbA1c (5.48±0.065 and 5.65±0.079%, P=0.269), and similar GWG-W (0.292±0.017 and 0.240±0.021kg/week, P=0.077). Binary logistic regressions showed that PP-BMI (P=0.012) and mean HbA1c (P=0.048), but not GWG-W (P=0.477), explained macrosomia. CONCLUSIONS: Good glycemic control in GDM patients was not enough to reduce macrosomia to acceptable limits (<10% of newborns). PP-BMI and mean HbA1c (but not GWG-W) were significant predictors of macrosomia. Thus, without ceasing in our efforts to improve glycemic control during GDM pregnancies, patients with overweight/obesity need to be treated prior to becoming pregnant.

[A new physiopathological classification of diabetic neuropathy]. / Bases fisiopatológicas para una clasificación de la neuropatía diabética.

Nowadays, Diabetic Neuropathy (DN) is considered the most common cause of peripheral neuropathy in clinical practice. It can affect sensitive, motor or autonomic nerve fibers, with symmetric, asymmetric, acute or chronic presentations. Due to this variability, with multiple physiopathologic mechanisms involved, a complex clinical classification has been used until recently. The aim of this review is to present a new classification of diabetic neuropathy, based on its physiopathology. It is divided in metabolic microvascular and hypoxic, autoimmune and inflammatory, compressive, secondary to complications of diabetes and related to treatment. It must be understood that DN is not just a functional disease, but a complication of diabetes with molecular and pathological substrates caused by hyperglycemia. Therefore, normalization of blood glucose is a fundamental step towards the successful prevention and treatment of DN.

Medical Data Wrangling With Sequential Variational Autoencoders.

Medical data sets are usually corrupted by noise and missing data. These missing patterns are commonly assumed to be completely random, but in medical scenarios, the reality is that these patterns occur in bursts due to sensors that are off for some time or data collected in a misaligned uneven fashion, among other causes. This paper proposes to model medical data records with heterogeneous data types and bursty missing data using sequential variational autoencoders (VAEs). In particular, we propose a new methodology, the Shi-VAE, which extends the capabilities of VAEs to sequential streams of data with missing observations. We compare our model against state-of-the-art solutions in an intensive care unit database (ICU) and a dataset of passive human monitoring. Furthermore, we find that standard error metrics such as RMSE are not conclusive enough to assess temporal models and include in our analysis the cross-correlation between the ground truth and the imputed signal. We show that Shi-VAE achieves the best performance in terms of using both metrics, with lower computational complexity than the GP-VAE model, which is the state-of-the-art method for medical records.